Features How the latest IT systems are helping data deluge could … · 2014. 1. 13. · Axell Wireless advocates the use of a Distributed Antenna System. It says this takes the signal

by Rahiel Nasir

register online @ www.networkingplus.co.uk to receive your free monthly copy managed

High-tech in the High StreetHow the latest ITsystems are helping retailers cash inReal World Networks,p9!

FeaturesWhat next afterSpanning Tree?PLUS deployingvideo conferencingFeatures, pp10-17!

Ofcom has announced a major trial of ‘whitespace’ technology which could help supportthe next wave of wireless innovation. Overthe next six months, around 20 public andprivate organisations will be participating inthe pilot to test a variety of applications.

For example, BT and technology special-ist Neul will work with the Department forTransport to test the potential enhancementof traffic information as part of a widerproject along the A14 between Felixstoweand Cambridge. Using white spaces to trans-mit data on traffic congestion and varyingtraffic conditions to vehicles, the technologyis designed to improve information todrivers and could even improve road safety.

Microsoft has already been conductingwhite space trials in parts of Africa. It willtest how the technology can provide access

to free Wi-Fi in Glasgow, which has thelowest level of broadband take-up of all UK cities. Working with the University ofStrathclyde’s Centre for White Space Com-munications, the company will also examineusing white spaces to link a network of sen-sors around Glasgow to create a ‘smart city’.

Additionally, Microsoft will work withMediaTek and Canada-based wireless spe-cialist 6Harmonics to conduct the world’sfirst network trial of the IEEE 802.11afstandard for white space devices. This willalso take place in Glasgow where the tech-nology will be used to provide mobile andfixed broadband services using 6Harmonics’Adaptive Radio Network system.

White space technology utilises thegaps between the frequency bands used tobroadcast digital terrestrial TV. Unlike

www.networkingplus.co.uk O C T O B E R 2 0 1 3

High-speed mobile broadband will be rolledout across the busiest parts of Britain’s railnetwork under plans announced by thegovernment at the end of last month.

Transport secretary Patrick McLoughlinsaid the programme will tackle areasalong rail corridors with intermittent orpoor coverage of mobile phone signals,and deliver a consistent and reliableservice for passengers on key routes.

He said 70 per cent of passengers willbenefit from the new technology by 2019,with noticeable improvements during2015. Network Rail and the industry willnow work on a business case to outlinehow they will fund the improvements.

Some experts say the plans do not go farenough. UK-based Axell Wireless, whichspecialises in wireless coverage solutionsfor public safety and rail applications, saysthe government needs to promote an inte-grated comms infrastructure that supports

all cellular coverage onboard trains. Butthis presents significant technologicalhurdles according to Ingo Flomer, thefirm’s product management director.

“The coverage would have to extendthroughout the entire 14,480km passengerand freight network, along with thenotorious black spots found in 6,300km ofcuttings and 335km of tunnels,” he says.

Flomer adds that there are also majorissues posed by modern train rolling stock:“RF signals generally glance off theoutside of multi-layered, metallic carriages,which results in a reduced level of signalpropagation inside carriages and thereforepoor quality mobile coverage.”

Axell Wireless advocates the use of aDistributed Antenna System. It says thistakes the signal from a mobile operator’sbase station or an off-air repeater and ampli-fies it to enhance reception in enclosedspaces such as train carriages or tunnels. "

Government needs to do more toimprove broadband on rail network

Ofcom launches major trial totest ‘white space’ technology

F I X E D & W I R E L E S S N E T W O R K S F O R E N T E R P R I S E U S E R S

other forms of wireless technologies suchas Bluetooth and Wi-Fi, Ofcom says theradio waves used by white space devicescan travel longer distances and are able topass through solid objects more easily.

It adds that some of these gaps may beused by other applications, such as wirelessmicrophones, but only at certain times.White space devices would access the

spectrum at times when it is vacant bycommunicating their locations to a databasedesigned to minimise the risk of interferencewith any existing users. A number of firms,including Google, Nominet, LS telcom,iconectiv, Key Bridge, Fairspectrum andSpectrum Bridge, have expressed interest intesting such intelligent databases.

(continued on page 2)

High-speed in the HighlandsRapier deploys wireless broadband network in the Cairngorms News, p3! Off-the-shelf, p18!

White space technology usesthe gaps – or ‘white space’ –that sit in the low frequencyband used to broadcast digitalterrestrial TV. As a result, theradio waves can travel longerdistances and more easilythrough solid objects.

Switching hourThe horrors of the data deluge could mean it’s time to get a new switch

White space technology usesthe gaps – or ‘white space’ –that sit in the low frequencyband used to broadcast digitalterrestrial TV. As a result, theradio waves can travel longerdistances and more easilythrough solid objects.

Net+ 1310 p1 (cover) AM RN.qxd 11/10/13 16:34 Page 1

news

2

register online @ www.networkingplus.co.uk to receive your free monthly copy

Councils need to integrate onlineservices with back-office systemsIT experts are warning that the lack ofintegration between the front end of acouncil’s website and its back-officesystems is a barrier to delivering servicesvia online channels. This presents seriousconcerns for cash-strapped local authoritieswho are hoping to claw back costs byincreasing online services in line with thegovernment’s Digital by Default initiative.

For its ninth annual Integration andEfficiency Report, specialist software houseNDL questioned 270 senior IT staff, whichbetween them represent two thirds of UKlocal authorities. It revealed that threequarters of councils now view increasingthe amount of transactions done online as the ‘main focus’ for maintaining orimproving service delivery.

Seventy per cent of those questionedbelieve the link between the front end of acouncil’s website and its back-officesystems will be provided by CRM systems.But the report says that a third of councilsstill integrate less than 10 per cent of CRMservices with back-office systems, and morethan 70 per cent admit that re-keying datagathered by CRM is “common practice”.

NDL says services do not have to berouted via CRM systems to reach the back-office, and believes e-forms could be a“nimbler, cost-efficient” alternative. But it

found that these are even less integratedthan CRM systems, and 55 per cent ofcouncils re-key more than three quarters ofthe data captured via e-forms.

According to NDL MD Declan Grogan,it is vital that councils’ websites areproperly integrated with wider IT systemsfor service delivery to be improved. “Ourexperience of working with around a thirdof local authorities is that systemsintegration rarely reaches the top of theagenda, partly because the cost of ‘workingaround’ poorly integrated systems is rarelydocumented or understood.”

Grogan adds that the problems arecompounded by the fact that when the ITdepartment attempts to implement asolution, it is also expected to shoulder thecost even though it’s the end userdepartment which ultimately benefits.“This impasse can scupper many otherwiseviable integration projects,” he says. !

LSO deploys Wi-Fi in 18th centurychurch – but it wasn’t easy

White space technology trial

nn ee tt ww oo rr kk ii nngg october 2013

The London Symphony Orchestra (LSO)now has a Wi-Fi network at St Luke’s, its18th century venue in London’s Old Street.The building is a Grade 1 listed HawksmoorChurch that has been restored to become thehome of the orchestra’s community andmusic education programme. It plays host toconcerts and workshops, and is available forrehearsals and recordings as well as forcorporate and private events.

The LSO has installed a Wi-Fi systemfrom Xirrus throughout St Luke’s. Eightaccess points (APs) cover the entire venue,including its main entrance areas, concerthall, café, offices, and conference facilities.

Quentin Bradley, the LSO’s ICTmanager, says Xirrus was chosen as it hasexperience of deploying Wi-Fi in difficultenvironments. “The church has thick stonewalls and a reinforced concrete slabbetween the basement and ground floor, aswell as a free-standing internal metalstructural skeleton – this layout andbuilding structure had implications on the

strength of wireless signal,” he says.Xirrus says its APs were strong enough topenetrate the stone walls where othershadn’t been able to. It adds that its “moremodern” network was able to overcome allinfrastructural obstacles to ensure robustwireless throughout St Luke’s.

While the LSO won’t reveal details abouthow its W-Fi system has been secured,Xirrus says that Spiceworks is mostly usedfor network monitoring and management.

The LSO plans to make maximum useof its new Wi-Fi network: “With visitorsexpecting seamless access to digitalapplications, Wi-Fi usage is dictated on anevent-by-event basis and we want to beusing it as fully as we can,” says Bradley.“Wi-Fi is critical to demonstrateapplications in rehearsal or teachingsessions, and DJs and musicians requirewireless access so they can stream playlistsusing services such as Spotify.”

The LSO may extend the Wi-Fi roll out toits Barbican site in the near future. !

St Luke’s Church has thick stone walls, a reinforced concrete slab between flooring, and aninternal metal skeleton – a difficult environment for wireless signals. © MATTHEW WEINREB

Enterprises are ditching traditionalapproaches to static WAN optimisationbecause of their inability to cope with thedemands of increasingly complexapplication environments, according to thelatest study from TRAC Research.

Around half the respondents in TRAC’s2013 WAN Management Spectrum reportsaid they would replace first generationWAN optimisation solutions. They saidthese solutions had failed to scale topromised levels, did not provide adequatevisibility into the performance of key

applications, and were not able to copewith rollouts of new applications.

TRAC says increased adoption of WANoptimisation is being driven by the need tosupport new initiatives such as SaaS, BigData, virtualisation, real-time applications,and increased volumes of mobilecomputing. This adoption is being supportedby a transition to lower cost, service-baseddeployment models. The research showsenterprises are moving to either a virtualappliance or cloud services and away fromphysical hardware deployments.

TRAC’s president Bojan Simic reckonsthat the top challenge for managingenterprise network traffic is controlling newusages such as unified communications,internet traffic, VDI and BYOD.

“Moving beyond WAN optimisation toachieve dynamic application performanceguarantee is a critical step for ITdepartments seeking to transition from atechnical cost centre to the role ofadvisers,” he says. “[They] can then workwith business stakeholders to ensure theirusers are as productive as possible.” !

First generation WAN optimisationsolutions have “failed”

Auriga Consulting has launched a range of Onboarding Services to enablegovernment departments, public sectororganisations and third party suppliers totake advantage of the PSN.

The firm says it will guide public andprivate sector organisations through theprocesses of migrating to the PSN, sellingservices to it, or gaining accreditation. Itwill offer assistance with frameworkdocumentation, compliance, prescribedtechnical architecture, service managementand security policies.

Auriga says that its services aredelivered by CESG listed advisors andtechnical architects who are familiar withgovernment requirements. “We haveassisted clients through PSN processes andunderstand what needs to be done to satisfythe various bodies who approveaccreditation of services and connectivity,”

says MD, Louise T. Dunne. “Our PSNOnboarding Services enables us to sharethat knowledge and assist public and privateorganisations as they seek to gain entranceto and exploit the benefits of the PSN.”

Auriga, which has offices in Londonand Reading, adds that its range of PSNservices is further complemented by avariety of specialist offerings. It says thatthese cover secure business processanalysis, design and development,transition management, and more. !

(continued from page 1)The amount of white space available in

the UK varies by location, the power levelof devices, and the point in the day at whichthey access the frequencies. Unlike someother parts of the radio spectrum, whitespaces will be available to use on a licence-exempt basis, potentially allowing for fasttake-up and innovation by manufacturers.

Ofcom CTO Steve Unger says: “In thefuture it won’t be just mobiles and tabletsthat are connected to the internet; billions

of other things including cars, crops,coffee machines and cardiac monitors willalso be connected, using tiny slivers ofspectrum to get online.

“This is likely to deliver large benefitsto society; however, there isn’t anunlimited supply of spectrum to meet thisextraordinary demand. This is why weneed to explore new ways of unlocking thepotential of spectrum – like white spacetechnology – to get the most from thisvaluable national resource.” !

Auriga can help organisationssatisfy complex PSN criteria

Net managers may have to move beyond staticWAN optimisation if they don’t want users toget frustrated with application performance.

MD Louise T. Dunnesays Auriga has a lotof PSN knowledge thatit can share.

NDL MD Declan Grogansays council websiteshave to be properlyintegrated with widerIT systems for theirservice delivery to beimproved.

Net+ 1310 p2 (news) AM RN.qxd 11/10/13 16:40 Page 2

newsregister online @ www.networkingplus.co.uk to receive your free monthly copy

3 october 2013 nn ee tt ww oo rr kk ii nngg

With data threatening to deluge enterprisenetworks, IT architects and managers arestruggling to keep their heads above water.To accelerate performance and bolstertheir networks, they have two options:deploy systems beefed-up with moregeneral-purpose processors; or switch tosystems with intelligent silicon poweredby purpose-built hardware acceleratorsintegrated with multi-core processors.

Performance can be improved by addingfaster general purpose processors tonetworking kit. But they bring increasedsystem costs and power demands whiledoing little to address latency, a majorcause of network performance problems.In contrast, smart silicon minimises oreliminates performance choke points by reducing latency for specific tasks.

In the past, hardware and softwarelargely progressed in step: as processorperformance increased, software sophisti-cation increased. These parallel advancesmade it possible to create more abstractedsoftware, enabling much higher function-ality to be built quicker and with lesseffort. Today, these layers of abstractionare making it difficult to perform morecomplex tasks with adequate performance.

Regardless of core count and clock rate,general purpose processors are too slowfor functions such as classification, crypto-graphic security and traffic management

that must operate deep inside each andevery packet. What’s more, these functionsmust often be performed sequentially,restricting the opportunity to processthem in parallel in multiple cores.

By contrast, these and other specialisedtypes of processing are ideal applicationsfor smart silicon, making it increasinglycommon to have multiple intelligentacceleration engines integrated withmultiple cores in specialised System-on-Chip (SoC) communications processors.

The number of function-specific acceler-ation engines available continues to growand shrinking geometries make it possibleto integrate more engines onto a singleSoC. It is even possible to integrate a sys-tem vendor’s unique intellectual propertyas a custom acceleration engine within anSoC. Taken together, these advances makeit possible to replace multiple SoCs with asingle one to enable faster, smaller, morepower-efficient networking architectures.

Adopting emerging smart siliconsolutions is the best way to efficiently andcost-effectively address the rapid, complexchange taking place in enterprisenetworks and harness the opportunities of the data deluge. Moving forward,design engineers will increasingly deploysmart silicon to achieve the benefits ofsignificantly higher performance andgreater efficiencies in cost and power.

THE WORLD ACCORDING TO...Greg Huff, chief technology officer, LSI

Dealing with the data deluge using smart silicon solutions

Fife-based Rapier Systems has installed asecure, high-speed wireless link fromAviemore to the CairnGorm MountainVisitor Centre, a distance of around 11km.

At 130Mbps, Rapier says the link to themountain far exceeds the 10Mbpsbroadband cable feed into Aviemore, and is“immune” to the extremes of weather thatare common on the 1,245m peak. Afteridentifying locations for antenna mountingthat would provide good line of sight whilegiving as much protection as possible fromthe elements, Rapier connected the centre(at the base of the funicular railway) to theMacDonald Hotel in Aviemore.

It built the link using a CambiumNetworks’ PTP 250 microwave system.This was installed primarily to provide aninternet connection and replaced theprevious solution which featured fourbonded DSL lines that provided 400kbpsdown and 600Kbps up – in good conditions.

At the hotel, Rapier deployed a wirelesslink providing 130Mbps between the twosites. It says this is capable of 250Mbpsusing dual polarity transmission/reception.Equipment was also installed that allowedthe existing bonded circuit to remain forredundancy. The entire installation ismanaged and monitored remotely usingVPN connectivity at the hotel end, givingcomplete visibility of the system and theability to change configuration.

The next phase is already under wayand will see local businesses in andaround the Cairngorms connected to thehigh-speed network. Rapier says thatthese businesses currently have very poorinternet connections.

In the final phase, the firm will addpublic and corporate Wi-Fi for all buildingsat the ski resort (including those at the topand bottom of the railway) and hot spots inand around the ski-lift areas. !

High-speed wirelessfor Highland ski resort

Princess Yachts has migrated from atraditional TDM system to virtualisedunified communications as part of anextensive network upgrade. The platform,provided by UC specialist Mitel and itspartner NG Bailey, is enabling more than400 employees at the luxury brand toconnect quickly and seamlessly acrossmultiple locations, driving up productivitylevels while streamlining costs.

Mitel says its solution provides PrincessYachts with a “highly flexible” platformfor communication and collaboration. Ituses a multi exchange platform to provideconnectivity to six of its sites in Plymouth,and the use of VoIP has enabled staff to

move around sites whilst remaining“seamlessly” connected.

According to the vendor, PrincessYachts has now put into place an “easilyscalable” IT strategy that will support theexpansion of telephony in the future.

The firm plans to integrate with Mitel’sMiCollab collaboration application whichprovides real-time access to a variety ofapplications and features for everyonewithin the organisation, either on or offpremise. Mitel adds that the applicationwill also extend the capabilities of mobiledevices and allow employees to stayconnected on the move without expensivecall or roaming charges. !

Virtualisation to give yacht firmincreased flexibility and mobility

EDITORIAL:

Editorial director: Rahiel [email protected]

Deputy editor: Ian Grant [email protected]

Designer: Alan McClenaghan [email protected]

Contributors: Geoff Huff,Polycom

ADVERTISING & PRODUCTION:

Sales executive: Andy [email protected]

Production: Suzanne [email protected]

Publishing director: Kathy [email protected]

is published monthly by:Kadium Limited, Brassey House, NewZealand Avenue, Walton-on-Thames,Surrey, KTI2 IQD, United Kingdom.Tel: +44 (0) 1932 886 537Fax: +44 (0) 1932 886 539Annual subscription: £80 (UK); £95(Europe), £100 (USA and the rest of theworld) airmail. Cost per single copy is£8.25.Printed in England by Williams Press.© 2013. All rights reserved.

The contents of the magazine may not bereproduced in part or whole, or stored inelectronic form, without the prior writtenconsent of the publishers. The viewsexpressed in this magazine are notnecessarily those shared by the editor orthe publishers.

ABC audited circulation:19,6791 Jul 2011 – 30 Jun 2012

Main photo: the CairnGorm Mountain visitorcentre. Inset: Rapier recommended the use oflarger antennas to provide better throughput andavailability in bad weather – these antennas aredesigned to survive higher wind speeds.


Following the rapid uptake of space in itsfirst hall, Everest Data Centre (EDC) isnow accelerating the development of itssecond data hall and has commissionedBorri to supply further UPS units.

EDC owner Ed Butler says the first hallat the Reading-based co-location facility isnow more than 40 per cent occupied, andhas “far exceeded” expectations.

“We have already accumulated aforward order book for our second data halland we are actively looking to add anothernetwork provider to the carrier neutral site,to bring our list of carriers on site up to six.The final data halls, 3 and 4, have not beenfitted out yet and are available as bespokedesign and build facilities.”

Named earlier this year as EDC’s soleUPS systems provider, Borri has already

completed the supply, installation andcommissioning of four B9600FXS 400kVAunits and two B9000FXS 125kVA systems.It says these are now fully operational andproviding reliable backup power.

The vendor will now supply two moreB9600FXS units for EDC’s second hall.

These will be placed in a dual corded A+Bconfiguration which, according to Borri, willdeliver complete redundancy and removeany single points of failure from the site.

Butler adds that the physical design ofBorri’s UPS systems makes them “veryaccessible” and “so easy” to maintain. !

news register online @ www.networkingplus.co.uk to receive your free monthly copy

4nn ee tt ww oo rr kk ii nngg october 2013

ICT investment inschools set to peakFunding for ICT investment by schools willreach an all-time high in 2014, according tothe British Educational Suppliers Association(BESA). In a recent survey of 1,238 schools,it found that ICT expenditure in the 2014/15academic year is forecast to be higher in cashterms than at any other time on record. BESAsays investment in hardware replacement,peripherals, software and technical supportwill reach £14,220 per primary school and£65,570 in each secondary school. Theprevious highest estimate of ICT expenditurewas in 2008/9 when allocations averagedaround £14,000 in primaries and £65,400 insecondaries – a total of around £320macross all UK maintained schools. !

UK charities offeredfree recycled ITPrism Disposal, a subsidiary of Prism Total ITSolutions, says the vast majority of ITequipment its collects for waste disposal havenot reached their end of life. The firm says it isconstantly receiving computers that can be re-used, and invites UK charities to get intouch with an outline of their needs which itwill then match with machines as they becomeavailable. Prism adds that all data on thehardware it collects is deleted to Ministry ofDefence standards. According to governmentestimates, about one million tonnes ofelectronic waste is dumped in landfill sites inthe UK every year and much more is sittingunused in storage. Prism adds that none ofthe waste it collects ends up in landfill. !

Modular data centres“more affordable”Research from DCD Intelligence reveals that theactual cost of deploying a modular data centresolution is around 13-14 per cent less than thebuild of a traditional data centre of similarcapacity. The study compared all associatedcosts of both approaches, and also found thatin many cases the greater density and moreefficient use of space means that modularfacilities are able to achieve significantly betterPUE. DCD’s research was undertaken with thesupport of three key players in the modularspace: IO, Colt and Gardner, all of whichprovided valuable information that made thecost comparisons possible. Other prominentplayers in both the modular and traditional buildspace also contributed data. !

Allied Telesis claims its new frameworkcould help enterprises reduce the complexityand cost of network management.

According to the firm, most softwaredefined networking (SDN) solutions aretargeted at large data centres which putsthem out of reach of the typical enterprisecustomer. It reckons AMF (Allied TelesisManagement Framework) offers apractical way for business users to takeadvantage of the benefits of SDN withoutthe “hefty price tag”.

AMF has been developed to ease theburden of managing enterprise networks by

automating many tasks that today require alevel of involvement by netadmins. The firstfeatures to be implemented within theframework include: centralised manage-ment; auto-backup; auto-provisioning; auto-recovery; auto-upgrade; plug-and-playnetworking; and zero-touch management.

The company adds that AMF offerscentralised management of an entirenetwork from any single device through a“simple and intuitive” CLI. It says thatconfiguration and firmware files areautomatically backed up regularly and areavailable for regenerating failed devices,

while configuration changes can be made onmultiple devices at the same time.

Jun Kohara, the company’s vicepresident of product marketing, says:“Unlike SDN, which requires a software-based controller external to the networkitself, Allied Telesis has embedded thecontroller functionality within the network,simplifying the delivery of this newtechnology. This enables many of thebenefits of SDN to be delivered without thecost, complexity and overhead it brings.” !SDN and alternatives to the Spanning TreeProtocol – feature pp10-13.

Enterprise SDN without the “hefty price tag”

Borri UPS provides the backup for Everest

e-skills UK has set up the Cyber Academyto help the UK stay at the forefront of thecyber security field and develop the skillsneeded to manage future threats.

The employer-backed academy will workclosely with education and government inan effort to inspire young people to considercareers in cyber security. It aims to providenew entry routes into the sector, and improveaccess to relevant and high-quality training.

Its work will include: embedding cybersecurity-related content in the school anduniversity curricula; organising effectiveinteraction between employers and school

pupils; and creating the first nationallyavailable degree-level apprenticeships incyber security.

The academy’s enterprise backersinclude the fast growing small companies

in the Malvern Cyber Security Cluster, aswell as large firms such as Atos, GeneralDynamics, IBM, John Lewis Partnershipand National Grid. It’s also supported byorganisations such as CREST and CyberSecurity Challenge.

Karen Price, CEO of e-skills UK, says:“Our research shows the cyber securityworkforce is an ageing one, with only sevenper cent of professionals under the age of29. For the UK to retain its innovative edgein this fast-moving field, we need to domore to bring new talent into the industryand continue to upskill existing staff.” !

e-Skills UK CEO KarenPrice says only sevenper cent of cybersecurity professionalsare aged under 29.

Employer-backed Cyber Academylaunches to boost infosec skills

Borri has now supplied eight UPS systems for Everest Data Centre’s colo facility in Reading.


Net+ 1310 p5 (IBM).qxd 10/10/13 17:12 Page 5

The Cloud Security Alliance (CSA) and theBritish Standards Institution (BSI) haveunveiled what’s claimed to be a “rigorous”third-party independent assessment of thesecurity of a cloud service provider.

By achieving ‘STAR Certification’, theCSA says that cloud providers of every sizewill be able to give prospective customers agreater understanding of their levels ofsecurity controls. It says that by providing auser-centric assessment, the certificationwill provide the additional layer of trans-parency the industry has been calling for.

“In light of recent government revela-tions, both consumers and providers ofcloud-based services have been asking forindependent and technology-neutralcertification to help them make moreinformed decisions about the servicesthey purchase and use,” says DanieleCatteddu, the CSA’s EMEA MD.

The certification programme is basedupon achieving the ISO/IEC 27001:2005management system standard and thespecified set of criteria outlined in theCSA’s Cloud Controls Matrix. This covers

11 areas: compliance; data governance;facility security; human resources;information security; legal; operationsmanagement; risk management; releasemanagement; resiliency; and security

architecture. The independent assessmentcarried out by an accredited CSAcertification body – such as the BSI – willassign a ‘Management Capability’ score toeach of these control areas.

Each will be scored on a specific maturityand measured against five managementprinciples. An internal report will showorganisations how they could improve toreach an optimum level of maturity(designated as either Bronze, Silver orGold). Approved organisations will be listedon the CSAregistry as STAR Certified. !

news register online @ www.networkingplus.co.uk to receive your free monthly copy

nn ee tt ww oo rr kk ii nngg october 2013 6

No more cuckoos in cloud landThe end of 2014 will be a “pivotal moment”for the enterprise cloud, according to arecent survey by Virtustream. The IaaSspecialist found that 69 per cent of largeorganisations in the UK are planning tomigrate their business-critical applications,such as ERP systems, into the cloudduring the coming months.

“ERP and other mission-critical applica-tions have mainly been deployed conven-tionally – the cuckoos in cloud land,” saysSimon Aspinall, the firm’s CMO. “Thenext 18 months will see these criticalapplications pushed out of their in-housedata centre nests and migrated to the cloud.”

Virtustream conducted its surveyamongst senior IT decision makers in 100medium and large organisations in both thepublic and private sectors. 84 per cent saidthey run an ERP system (mainly eitherOracle or SAP) and that they also manageparticularly complex IT estates. SAP users

in particular run an average of 530applications of which 38 per cent arelegacy and not built for cloud.

Virtustream reckons such SAP usersare at the forefront of cloud adoption: “Of all the ERP vendors, SAP users are inthe right position to reap the benefits ofbusiness-critical applications in thecloud,” says the company.

It adds that a number of importanttechnological advances have helped seniorIT teams pragmatically assess the cloudopportunity and dispel the traditionalfears over security, business risk and lossof control. These developments includegeo-tagging data, chip-level authenticationthrough IntelTXT, application-level servicelevel agreements, and legislation.

59 per cent of respondents expect a cloudmigration to result in IT budget savings.On average, these would be around£1.5m, or nearer £2m for SAP users. !

CSA and BSI launch STAR security certificationThe CSA’s DanieleCatteddu says usersand providers of cloudservices have beencalling for independentcertification.

SPAR UK will use real-time collaborationtechnologies delivered via the cloud in aneffort to increase efficiency across sevenbusiness units.

The convenience store chain’s businessunits previously used basic tele-conferencing systems, but these provedinefficient and ineffective forcollaboration, and executives ended uptravelling to have face-to-face meetings.As a result, SPAR needed a reliable,scalable, collaborative video solution toensure employees were able tocommunicate effectively without loss oftime in travelling.

After evaluating several systems, thecompany chose Avaya’s Scopia platform.This allows SPAR to have multipleconnections in high-definition to each ofits seven business units.

“The ease-of-use of the Scopia systemensures our adoption curve will be moreeasily managed than with other solutions,and we expect our use to grow as the

system beds in,” says Roy Ford, SPARUK’s IT controller.

The system uses a cloud-based, hostedplatform from Videonations. Businessdevelopment manager Richard Savillesays his firm provides full end-to-endsolutions for companies that wish to moveeffectively from a capex to an opexmodel, and therefore gain the best of thetechnology without the headaches ofmaintenance. “In the case of SPAR,Videonations demonstrated and suppliedthe endpoints and incorporated everythinginto a complete solution based on theAvaya Scopia platform,” he says.

Avaya claims that its technologies are“optimal” for use both in hosted as well asowned environments. It adds that byusing Scopia, customers can interact face-to-face whenever and wherever theywish, “without the hassle” of traditionaltelepresence systems. !Deployment considerations for videoconference systems – feature pp14-17.

A new pay-as-you-go cloud service aimsto enable customers from any industry toeasily make cloud computing an integralpart of their IT strategy.

Capita Private Cloud from Capita ITServices is built upon VCE’s Vblockconverged infrastructure system. It hasbeen designed for companies wanting ablend of public and private cloud serviceswithout the complexity of managingmultiple suppliers and contracts.

“Clients we work with, from NHS trustsand local authorities through to retailers andmanufacturers, have told us they want thebenefits of cloud but are concerned aboutsecurity and complexity,” says Andy Parker,deputy chief executive at Capita. “CapitaPrivateCloud takes that uncertainty away by

offering a simple, cost-effective solutioncustomers can access within minutes.”

Parker reckons the combination of publicand private cloud services, with the supportof a dedicated account manager and techni-cal experts, means businesses can meet alltheir IT needs in one place. “In addition,managing all cloud platforms together inCapita’s UK data centres guarantees datasecurity and sovereignty – a key for manypublic or highly regulated companies, suchas banks and pension providers,” he says.

As well as being able to choose exactlywhat they need and only pay for what theyuse, Capita says that customers have accessto 2,000 pre-tested cloud applications via aself-service portal, allowing them to easilytailor, monitor and manage services. !

Avaya claims its Scopia video

collaboration system canbe used “without thehassle” of traditionaltelepresence systems.

Pay-as-you-go service usespublic and private platform

SPAR UK chooses hostedvideo conferencing system

Net+ 1310 p6 (news) NEW.qxd 11/10/13 16:47 Page 6

Net+ 1309 p11 (BT).qxd 19/9/13 17:55 Page 11

Net+ 1310 p8 (APC).qxd 10/10/13 17:17 Page 8

real world networksregister online @ www.networkingplus.co.uk to receive your free monthly copy


Downtime is not cheap atPoundland Poundland has almost 500 stores acrossthe UK and is said to serve more than 4.5million customers every week. Thecompany’s core network is essential to itsday-to-day operations. It handles every-thing from financial and staff manage-ment systems to the integration of salesand stock information, and a 24-hourwarehousing operation that despatchesmore than 800 million items every year.

IT director Mike Gray says: “Ourbusiness is highly reliant on IT – fororders, stock control, transport andlogistics and warehousing. IT resilience iscritical as unplanned downtime can quicklydamage our business.” As a result, he saysPoundland has now put in place proactiveand preventative service measures tominimise business impact and ensure astable platform for its ongoing growth.

Poundland has worked with NextiraOnesince 2004, and under a multi-year deal,the systems integrator is now providing avariety of managed services for theretailer’s Cisco-based network infra-structure. The contract covers the datacentre, voice communications and LANinfrastructure across its HQ, together withPoundland’s two regional distributioncentres and nationwide estate of stores.

NextiraOne says it designed its managedservices to be “flexible but comprehen-sive” in order to support and enhancePoundland’s operations efficiently andcost-effectively. Under the agreement, it isproviding 24/7 break-fix support as wellas performance monitoring, optimisation,and flexible additional on-demand services.The firm says the latter provides additionalresource to supplement the retailer’s in-house IT team during trading peaks and unforeseen circumstances, thus“streamlining” the outsourcing process.

NextiraOne adds the managed servicesprovide Poundland’s team with a singleoverview of its entire IT estate. This allowsit to make informed business decisionsand take pre-emptive action before issuesthat could impact the business develop,thus avoiding costly downtime andassociated loss of productivity.

Mark Petty, NextiraOne’s managedservices director, says: “By combiningtechnical and operational support withperformance monitoring and networkoptimisation, plus additional on-demandresources, Poundland has the best of allworlds – cost-effective and highly efficientservices. We designed the service to beflexible and scalable, so that as thePoundland business grows and new storesor sites are added, the company hassufficient data to understand theinfrastructure investments required.”

Optical Express realises itsnetwork visionFounded in 1991, the Optical ExpressGroup’s (OEG) portfolio has developed toinclude laser eye surgery, private dentistry,healthcare services and cosmetic surgicaland non-surgical treatments, as well asmaintaining the core optics division ofspectacles and contact lenses.

It currently has more than 170 retailsites in Croatia, Germany, France,Ireland, the Netherlands and the UK, aswell as operations in the US.

All of OEG’s national and internationalsites are linked via an MPLS network.This is used for fixed line voice and data,as well as for a VoIP system that wasoriginally supplied by Avaya. But in2011, the group began to have issues withthe network and the service supplied byits incumbent provider.

IT manager Craig Duffy (picturedabove) says OEG then spoke to a numberof telecoms firms to find a solution.But none were able to provide a fullpackage of calls, lines and mobileservices together with an MPLSnetwork. The group had already beenworking with Vodafone for mobileservices, and so the two companiesbegan discussions about a moresophisticated network which wouldinclude unified communications.

Vodafone was commissioned to rollout a fully managed service acrossOEG’s business, including an MPLSnetwork. This now securely connects its head office, call centre and stores.

Duffy explains that all of OEG’ssmaller stores don’t have an externalphone system, so they route theirexternal calls up to head office over thenetwork and then break onto the voicenetwork from there. “The bigger storeshave their own external lines and a localswitch which also acts as a standalonesystem in the event of a system failure,allowing the store still to receive andmake calls. All of our calls within thecompany are done for free over VoIP.”

As well as QoS technology to prioritisevoice traffic, Vodafone has providedhardware acceleration to maximise theeffective speed of the network. This helpsOptical Express share data, such as highresolution eye images, between its sites.It’s claimed the new network is up to eighttimes faster than its predecessor whichcouldn’t cope with such images.

The faster network has also enabledOEG to stream video from its headoffice to its stores, for example, toshow videos to customers on theconsultation process. Previously, it hadto manually send each video to each PCin a store – a process that consumed alot of time for the IT help desk.

High-tech on the high streetFrom checking stock information to sending out orders, good customer service oftenstarts with the retailer’s IT network and back-office systems.

Cosmetics firm launchesthe ‘Lush Cloud’Since being established 18 years ago,Lush Cosmetics has been driven byinnovation and its ethics. It’s currentlypresent in 51 countries with 856 shopsand also has global manufacturing sites.

The firm is now using a cloud-basedfile sharing system from Storage MadeEasy. SME claims its Enterprise FileShare and Sync solution is unique in thatit unifies files from disparate sources thatcould be on-cloud or onsite. It says thesystem provides various control pointsfor data including full file life-cycleaudit and remote file encryption forprotection against potential snooping.

Richy Baxter-Freeman from LushInternational Support says SME waschosen because of its flexibility: “It hasbeen very easy to manage large numbersof users and file structures from all areasof the business in over 50 countries. Ourstaff can access data on the go throughtheir mobile devices, through a standardweb browser, or directly using thedesktop integration functions.

“The solution is also flexible in thesense that we are able to manage multipledata sources and, most importantly, keepour data sources and management of

these separate, meaning we can easilymove data sources if necessary.”

Lush ran an initial two-month trial ofthe system with 50 people prior to rollingit out department by department. It expectsto reach more than 1,000 users by the timethe solution is fully deployed.

The SME solution resides on servershosted by Memset (pictured below). TheUK-based cloud service specialist alsoprovides the back-end data store, MemsetObject Storage, which is a customisedversion of the OpenStack Swift platform.

“The SME appliance is hosted usingXen, although SME also supportsVMware, KVM or Hyper-V – in fact,any OVF compliant hypervisor,” saysSME CEO Jim Liddle. He adds that hisfirm’s Mac and Windows desktop toolshave all been custom branded for theclient, as has the storage solution whichis known as the Lush Cloud.

Net+ 1310 p9 (rwn) AM RN.qxd 11/10/13 16:50 Page 9

E thernet has become the true linguafranca of the digital machine age.What started as a LAN system hasproved supremely flexible in coping withfaster speeds and with linking devicesinside the data centre and the world beyondthe firewall. However, the rapid rise of virtualisation and the resulting increase inthe complexity of the network landscape isnow coming to a head. Networking todayrequires a different approach.

As virtual machine migration (VMwarevMotion or Microsoft Live Migration)requires Layer 2 extension, existingEthernet routing protocols like spanningtree protocol (STP) and its extensions,RSTP or MSTP, are facing majorchallenges and limitations. They areincreasingly becoming less fit for purpose.

A quick reviewSTP was invented in 1985 by RadiaPerlman at Digital Equipment Corporation.It is a Layer 2 (data link) protocol thatruns between Ethernet bridges to helpcreate a loop-free network topology, anecessary condition for virtual LANs ifthe user is not to lose control of dataflows and have the network crash.

The idea behind spanning tree is thatbridges (switches) can discover a subsetof the network topology that is loop-free,i.e. a ‘tree’. It also makes certain there isenough connectivity to reach everyportion of the network – in other words, itspans the entire LAN. Bridges willperform the spanning tree algorithm whenthey are first connected, as well as anytime there is a topology change.

STP uses a version of the Bellman-Forditerative algorithm to send Bridge ProtocolData Units (BPDUs – packets sent betweenEthernet switches which are essentiallymulti-port bridges) to ‘elect’ a root bridge

and to look for the optimal route, selectingthe ‘closest’ candidate every time. Everyswitch accepts and retains only the bestcurrent root bridge information. The switchthen blocks alternate paths to the rootbridge, leaving only the single optimal (interms of path cost) uplink, and continuesrelaying the optimal information.

If a switch learns about a better rootbridge – such as a more effective bridgeID or shorter path to the root – the oldinformation is erased and the new oneimmediately accepted and relayed. Theswitch stores the most recent BPDUs withevery port that receives them. Therefore,for a given switch, there is a BPDU storedwith every root or alternate (blockedport). The resulting tree, with the root atthe top, spans all bridges in the LAN.

This is fine while Ethernet is confined tobarracks, as it were. But it has escaped intothe wild beyond the firewall, and virtualisa-tion has increased the rate of change in thenetwork topology inside the firewall. Allthis makes STP a nightmare for netadmins.

Scott Hogg, director of advanced tech-nology services with Global TechnologyResources, says the default configurationsettings with STP are “efficient” atpreventing loops. As a result, many firmssimply accept the defaults and ignore STPin their network designs. Years later theymay discover network issues related tospanning tree. “There have been manyoptimisations to STP but, if they have notbeen configured, the network is not bene-fiting from these new features,” he says.

Chris Hay, solutions architect at AlliedTelesis, agrees: “STP is wasteful in termsof cost because you need to put in a link,which then has to sit within the solution asa ‘redundant’ item, waiting for failure. STPsare slow and cumbersome, and many of ourcustomers have big issues with their STPbreaking the network rather than fixing it.

With STP you enable it and expect it towork, but it needs a lot more maintenancethan people think. As a result, networkoutage failure times tend to be long.”

Ethernet devices running STP have beenimplemented in networks since the early1990s. But because of the cost and com-plexity issues, a number of improvementshave emerged. These include Rapid STP(RSTP) and Multiple STP (MSTP).

Nigel Bragg, network architectureprincipal at network equipment supplierCiena, says: “Until now, spanning treehas been the primary protocol forEthernet networking. While in the pastvarious short-term fixes for the issueswith STP, like RSTP and MSTP, havebeen implemented, these solutions havenot resolved the issues completely.”

Network suppliers are naturally keen to develop answers they hope will becomede facto standards. For example, Hay saysthe alternatives to STP are a mixture oftechnologies that are put together duringthe design phase with due consideration for the desired and required end result,including link aggregation and loopprotection. “Allied Telesis offers twosolutions, the Ethernet ProtectionSwitching Ring (EPSRing) and VCStack.There is no one best alternative but thedesign needs to ensure that loops areprotected and that resiliency is included.The design then becomes a ‘resilientsolution’ rather than a ‘redundant’ one.”

EPSRing is claimed to enable rings torecover from link or node failures in aslittle as 50ms, depending on port type andconfiguration. STP could take up to 30seconds, and even Rapid STP one to threeseconds. Much like STP, EPSRing providesa polling mechanism to detect ring-basedfaults and failover accordingly. But unlikeSTP, it uses a fault detection scheme toalert the ring that a break has occurred.

The ring then takes immediate actioninstead of going through an STP-likereconvergence (discovery of the newoptimal path and perhaps root bridge).

VCStack allows multiple switches toappear as a single virtual chassis whichacts as a single switch, simplifying man-agement. “When put together in the rightcombination, the mixture of technologiesensures a solution that is simple to managewhen compared with STP and ensures noimpact on the user’s experiences shouldany part of the system fail,” says Hay. “The intention is that only the networkadministrator would know that there hasbeen a network failure – the users wouldnot experience any lack of service.”

Shortest path bridgingWhile Allied Telesis offers a proprietarysolution, the industry is lining up behindthe IEEE 802.1aq standard that came intoforce on 29 March 2012. This is the so-called shortest path bridging (SPB)protocol. Ciena’s Bragg says SPB canachieve big improvements in Ethernetbehaviours by applying IS-IS1 link-statetechnology to replace spanning tree. Theimprovements include:! Optimal routing: all traffic flows travel

on the shortest paths, whereas withspanning tree each flow travels on a tree rooted on a single bridge

! SPB uses all available links so it doesnot strand capacity, whereas spanningtree must block redundant paths toprevent loops, sometimes leading tonetwork resources being operational but unavailable

! Faster convergence: with link-statecontrol a network can reconverge in100ms after a topology change,whereas spanning tree can requireseconds or even minutes.

STP alternatives register online @ www.networkingplus.co.uk to receive your free monthly copy

Spanning tree protocol, which has kept Ethernet networks up and running for 30 years,is no longer fit for purpose. IAN GRANT plunges into acronym hell to find alternatives.


Getting thereGetting there

It looks neat and tidy now, but the logicalview of the network may tell a different

story. Virtualisation, BYOD and M2M areincreasing network topological

complexity and the rate of change isforcing netadmins to find better ways

than STP to keep up.

It looks neat and tidy now, but the logicalview of the network may tell a different

story. Virtualisation, BYOD and M2M areincreasing network topological

complexity and the rate of change isforcing netadmins to find better ways

than STP to keep up.

Net+ 1310 p10-13 (Spanning Tree) NEW.qxd 11/10/13 17:01 Page 10

Net+ 1310 p11 (ads).qxd 11/10/13 17:06 Page 11

With link-state routing there is no interrup-tion to traffic that is unaffected by a fault.Spanning tree can shut down forwardingon the entire network completely untilreconvergence has been achieved.

SPB eliminates the need for hop-by-hop signalling protocols when setting upmulticast services. This is handled by IS-IS directly, and thereby achieves much quicker multicast setup.

“The IEEE places great importance on backwards compatibility, and the twovariants of the SPB protocols (SPBV andSPBM) run with minimal modification on Q-in-Q2 and MAC-in-MAC3 [a.k.aProvider Backbone Bridge or PBB]hardware respectively,” says Bragg.

Ever since the formerly cosy relationshipbetween HP and Cisco started breaking upabout five years ago, the two have foughtto win the standards battle. David Sturgess,

HP Networking’s EMEA product linemanager, says his firm has been playing a“very active” role in both the IEEE and theIETF efforts to standardise two new Layer2 STP replacement technologies, namely802.1aq (SPB) and TRILL (TRansparentInterconnection of Lots of Links).

“HP’s David Law, Mark Pearson, PaulBottorff, Anil Rijhsinghani are all activecontributors to the IEEE and IETF,” hesays. “Anil is one of the contributors toTRILL, making HP uniquely positioned toinfluence the standard as well as drivingthe engineering team towards implementa-tion. Paul Bottorff is recognised as one ofthe key contributors to the PBB technology[a.k.a 802.1ah], and is actively working onthe SPB standard as well.”

According to Sturgess, TRILL is animportant part of HP’s data centre strategy:“The TRILL stack has been successfully

tested for compliance to the standard bythe University of New Hampshire. TRILLhas been available since December 2012in our Comware v7 OS and our top-of-rack (59xx) product line. Since December,TRILL has been ported to our aggregationswitch (the 11900) and has been availablefrom the end of September on our newcore data centre switch, the 12900. TheSPB feature will be made available to alldata centre platforms by end of this year.”

FabricsAccording to Sturgess, HP has been simpli-fying both the architecture and managementof its networking systems since 2009, theaim being to allow customers to supportvery large Layer 2 domains without STPthrough its unique Intelligent ResilientFramework technology. HP acquired IRF

when it bought Ethernet inventor BobMetcalfe’s company 3Com in 2010. IRF isa software virtualisation technology, and itsmain idea is to connect multiple networkdevices through physical IRF ports, performthe necessary configurations, and thenvirtualise these devices into a distributedservice. This scheme includes thecooperation, unified management, and non-stop maintenance of multiple devices.Arch-rival Cisco implements equivalentproducts and services using its VSS (VirtualSwitching System) and vPC (virtual Portchannel) technologies.

Sturgess reckons IRF allows the creationof large network fabrics that consist ofmultiple switches at a single layer (access,aggregation or core) that operate andappear logically as a single switch.

“By virtualising all L2 and L3 networkfunctions, such as routing, gateway, link aggregation, and QoS and securityfunctions and extending the control planeacross the 10GbE and now 40/100GbEnetwork fabrics, IRF distributes devicemanagement such that you only need asingle configuration file and one softwareimage. Devices inserted into an IRF domainautomatically update their configurationfile and software, preventing you frommodifying one device in the domain inisolation from the others.”

He points out that a cluster of IRF-enabled switches can also interwork withnon-HP switches (at different networklayers) through standardised interfaces/protocols (such as LACP or link aggrega-tion control protocol) or routing protocolslike OSPF (open shortest path first), BGP(border gateway protocol) and MPLS.

“If you now use IRF domains to thedevices in the access layer and again to theupstream core and/or aggregation layer, thiscomplex environment is vastly simplifiedto ‘look’ logically like two switches, theaccess connecting to the core.”

Sturgess claims this means there is norequirement for STP or the virtual routerredundancy protocol (which eliminatesthe single point of failure inherent in thestatic default routed environment); bothserver uplinks and switch uplinks cansimply use normal LACP link aggre-gation for physically meshed distributedtrunking. “All links are active and load-balanced, IP addresses and interfacecomplexity is virtually eliminated, andthe environment provides a guaranteedsub-50ms recovery time with no singlepoint of failure,” he says.

And it can scale. By extending a controlplane across Ethernet fabrics, IRF providesa logical ‘single switch’ of severalthousands of 10G/40G/100G ports.

Sturgess adds that IRF and the newstandards are not mutually exclusive. “AnIRF fabric can be built using either TRILLor PBB/SPB. This makes the fabric morescalable and faster to reconverge. As switchfabrics get faster and port densities increase,so will IRF domain sizes. The same tech-nology applies to campus switches, as wellsome recently released core routers, makingthe value proposition ubiquitous acrossHP’s networking portfolio, and scaling to the largest cloud hosting providers.”

What about SDN?Software defined networking (SDN) ismore correctly defined as the separation ofthe control plane and the data plane in datacentre communications and the identifica-tion and allocation of resources undercentralised programmatic control (also seefeature, June 2013 issue). SDN is the much-hyped response to the complexity alsoaddressed by IRF, VSS, etc, so much sothat people are now starting to talk about

STP alternatives




the ‘software defined data centre’. Thefirst SDN standard is the OpenFlowprotocol that enables the networkcontroller to interact with the forwardingplane and make adjustments to thenetwork so it can adapt better to changingbusiness needs. With OpenFlow, entriescan be added and removed to the internalflow-table of switches (and potentiallyrouters) to make the network moreresponsive to real-time traffic demands.

Any device that wants to participate inthis environment must support OpenFlowwith a standardised interface. This interfaceexposes the internal workings of the device,enabling the controller to ‘push down’changes to the flow-table. Once the devicesare OpenFlow-enabled, network administra-tors can use them to partition traffic, controlflows for optimal performance, and starttesting new configurations and applications.

The excitement that has greeted SDNand OpenFlow has led to a swathe of newnetworking suppliers such as Nuage andVyatta alongside more traditional oneslike Brocade, Cisco, HP and Juniper aswell as the older telecoms equipmentfirms like Alcatel-Lucent, Ericsson,Huawei and Nokia Solutions andNetworks (formerly Nokia SiemensNetworks). Presently, each has a slightlydifferent take on the subject.

The available technology is still largely inpilot projects, but initial results are promis-ing. The Open Networking Foundation isoverseeing SDN standards formation efforts.It has published an early case study of aNEC implementation of OpenFlow at theKanazawa University in Kanazawa, Japan.It quotes Keisuke Nagase, professor ofmedicine, health care administration, andmedical informatics, saying: “We areenjoying rapid recovery time and flexibilityin a network with reduced maintenance andoperational costs. The time for recoverywas reduced to seconds rather thanminutes with STP [and] RSTP. We nowhave flexibility in introducing multipleindependent LANs as and when neededby equipment or the medical teams.”

Nagase went on to say the acquisitioncost of the hardware was almost the sameas the legacy network, but the operationalexpenses and maintenance cost weresignificantly reduced. “I estimate savingsof 80 per cent on my operational expenses,including reduction in staff hours requiredto manage the network. We also expect theprice of OpenFlow switches and controllerswill be reduced further as a result ofcompetition in the market. Furthermore,with the flexible configurability ofOpenFlow, a full mesh configuration isnot required, and our next phase will berealised in less cost per switch.”

It is still early days. Most SDN productstarget specific uses and are definitely notone-size-fits-all according to Ethan Bankswho hosts and edits the PacketPushersblog. “The OpenFlow specification, whichis a core component of the centralisedcontroller model of SDN, is running aheadof the capabilities of currently availablesilicon to perform all potential matchingoperations in hardware.”

Banks believes that netadmins canachieve most of what SDN promises withexisting fabrics like VSS and IRF, butthey require a homogeneous data centrenetwork. SDN will allow firms to usebest of breed kit and also allow networkfunctions to run on commodity servers.

“ECMP [equal cost multi-pathing] andMLAG [multi-chassis link aggregation]creation is a relatively trivial task for acentral controller with a holistic view ofthe network, as there is no greatdistinction to be made between physicaldevices,” he says.

Banks adds that another advantage is thathaving a centralised network view allowsnetadmins to set up policies such as latencyand hop count to define end-to-end pathsfor data to follow, not merely source anddestination addresses. “Because a centralcontroller sees the network as a whole,there’s no need for distributed protocols todetermine a loop-free, best-path topology.An individual switch no longer has tofigure out for itself how to get to a remotedestination; the switch is told how toforward by the controller. ‘Best path’ canmean whatever a network designer wants itto mean, and not what a group of protocoldesigners decided it meant.”

That’s not to say firms can ditch TRILLand SPB for SDN/OpenFlow just yet.Banks points to SDN’s untested scalability,its undetermined scope, unverifiedperformance, and unknown compatibility

with legacy systems. Portugal Telecom istesting various suppliers’ SDN productsthroughout its network, including its hugenew data centre in Covilhã (see last month’snews). Technical director Salas Pires saysthere’s little uniformity in the way suppliershave implemented OpenFlow: “I expect itwill be 12 to 18 months before we seestandards, but I am planning to makedecisions within a year.” !

Notes1 Intermediate System to Intermediate System

is a routing protocol to move informationefficiently within a computer network, agroup of physically connected computers, orsimilar devices. It works by reliably floodinglink state information throughout a network ofrouters. Each IS-IS router independentlybuilds a database of the network’s topology,aggregating the flooded network

information. It uses Dijkstra’s algorithm forcomputing the best path through thenetwork, and works within a domain orfirewall rather than, as with border gatewayprotocol, between autonomous networks.

2 Q-in-Q allows multiple VLAN headers to beinserted into a single frame, an essentialcapability for implementing Metro Ethernetnetwork topologies.

3 MAC-in-MAC or provider backbone bridges(PBB) is a set of architecture and protocolsfor routing over a provider’s network thatallows interconnection of multiple providerbridge networks without losing eachcustomer’s individually defined VLANs. It was created by Nortel which offered it tothe IEEE 802.1 committee as a standard. Thestandard was approved by the IEEE in June2008 as IEEE 802.1ah-2008.

STP alternativesregister online @ www.networkingplus.co.uk to receive your free monthly copy



The next few years are expected to seeexplosive growth in the use of videoconferencing as a fundamental toolfor businesses to enhance communicationand collaboration between employees,partners and customers. The technologyhas developed considerably from earlyadopters to its current form of mass marketrollout. It’s anticipated that nearly half ofinformation workers will have some typeof personal video solution in 2016, up fromaround 15 per cent today (Forrester –Preparing for Uneven Corporate Adoptionof Video Communications, May 2011).

At a simple level, a video conference isan online meeting between two parties inwhich the participants can see and heareach other. As well as the hardware, thecomponents necessary to make this happeninclude the codec software plus a programthat bridges both parties together across adigital connection, managing the exchangeof voice and video between participants. A management tool for the scheduling ofconferencing sessions can also be used.

At a slightly more advanced level, it is also possible to provide the ability toshare content from a device during a

video conferencing register online @ www.networkingplus.co.uk to receive your free monthly copy


Joining the bridge

In multi-point video calls, three or more locations are connected together and a central independent software platform processes the data streams ofvoice, video and content. More sophisticated versions of these software programs can offer much more in terms of dynamic network functionality.

With video conferencing becoming a core component of IT infrastructure, telepresencespecialist POLYCOM looks at the basic components of the technology and what youneed to consider when deploying a solution.

Net+ 1310 p14-17 (Telepresence) NEW.qxp 11/10/13 17:24 Page 14

video conferencingregister online @ www.networkingplus.co.uk to receive your free monthly copy


video call. The quality and type ofcontent that can be shared depends on therate of data exchange during the call.

The points of videoconferencingVideo-enabled meetings happen in twodistinct ways: either point-to-point or withmulti-point. In point-to-point, the simplestscenario is where one person or group isconnected to another. The components (i.e.microphone and camera) that enable themeeting to take place are often integratedinto computers, or can be combined intodedicated, room-based hardware solutions.

In multi-point video calls, three or morelocations are connected together. All theparticipants can see and hear each other, aswell as see any content being shared duringthe meeting. In this scenario, data streams ofvoice, video and content are processed bycentral, independent software. Combiningthe individual participant’s video and voicetraffic, the program re-sends a collectivestream back to the meeting in the form ofreal-time audio and video imagery.

The software which creates this virtualmeeting room and the digital processinghardware on which it resides is oftencalled a video bridge or multi-pointcontrol unit (MCU). The process ofdialling into and participating in a virtualmeeting is known as ‘joining a bridge’,while different virtual meeting rooms areassigned unique ‘bridge numbers’ andusers join a video call by dialling them.

Depending upon the technical capabilityof the video conferencing system beingused, images seen by participants are eitherclassified as ‘active speaker’ or ‘continuouspresence’. In active mode, the screen onlyprovides an image of the person that isspeaking. In continuous presence, thebridge divides the image on the screen intoa number of different areas. The personspeaking at any point in time is presentedin a large central area which is surroundedby images of the other participants.

Whereas point-to-point conferencing isrelatively simple, the creation andmanagement of multi-point conferencescan be complex. An MCU must be able tocreate, control and facilitate multiplesimultaneous live video conferencingmeetings. A further complexity is addedwhen different locations may connect tothe meeting over digital or analoguestreams at different speeds, with differentdata transport and signalling protocolsemployed to facilitate the communication.

To link these users into a common,virtual meeting, the MCU must thereforebe able to understand and translate betweenseveral different protocols (i.e. H.264 forcommunication over IP, and H.263 forISDN). The MCU will also allow thosejoining the video bridge to do so at thehighest speed and the best possible qualitythat their individual system can support.Although there are two separate processestaking place here, this is often collectivelyreferred to as ‘transcoding’.

It’s important to note that not all bridgesprovide such transcoding capability, andfailure to do this can seriously impact thequality and experience of video calls. Whentranscoding isn’t provided and users dialinto a bridge over different connectionspeeds, it is possible that the bridge mayonly be able to support the video meetingby establishing the connections at thelowest common denominator.

To illustrate the negative effect of this,consider a meeting that takes place withmost users joining the bridge from the high-speed corporate network, but where one ortwo individuals dial into the meeting from

home on low-bandwidth DSL or ISDN. Inthis case, the experience of the corporateusers is downgraded to the lowest commondenominator of the home-users, potentiallymaking the video call ineffective.

Where effective transcoding is supportedby the MCU, those on the corporatenetwork will continue to enjoy HD videoquality, while remote users receive qualitycommensurate with their connection speeds.

When an MCU is designed well –integrating multiple vendors and allowingusers to call in at the data rate available –the result is an easy, seamless experiencefor all users, allowing people to focus onthe meeting, not the technology.

Building bridgesAs video conferencing technology hasevolved, two main protocols have emerged

to provide the signalling control for theestablishment, control and termination ofvideo conferencing calls: SIP and H.323.

For the encoding and decoding of visualinformation, the industry is moving towardsthe H.264 standard which was developedto provide high-quality video at lowerbandwidth over a wide range of networksand systems. Scalable Video Coding is anextension to the H.264 protocol and wasestablished to facilitate the enablement ofvideo conferencing on a wider range ofdevices, such as tablets and mobile phones.

With multiple calls taking place at once,software analyses all the different datastreams coming into the bridge processors,and assigns data streams accordingly.

At the simplest level, the processingworkload for bridges is dependent uponfour factors: the number of locations thatdial into each bridge; the number of

conferencing calls that each bridge musthandle simultaneously; the amount of datathat are being received on each digitalstream (higher resolutions of images andsound generate more data that need to beprocessed); and the degree of transcodingthat the bridge must perform while handlingcalls being received at different connectionspeeds and utilising different protocols

As the workload increases, each bridgemust process more data. Performance canbe improved by increasing the number ofdigital signalling processors (DSPs) used todecode and encode the digital streamsentering and leaving MCUs. If the bridgingfunction becomes overloaded, video andvoice data may be lost, causing latency tobe introduced into calls. All of this willdegrade the meeting experience.

Extra processing resource can be provided for the bridging function by either


using a more powerful bridge (with agreater number of DSPs) or via a virtualsoftware approach, where the software thatcontrols the signalling function can operateindependently of the physical hardware.

A conference call with an assignedconference number does not have to takeplace or be processed by a dedicated pieceof hardware. The call can be virtualised andassigned to whatever physical bridge hasthe correct resource or capacity to handleit. In extreme (but rare) circumstances, thevirtualisation manager may assign resourcesfor a call across several different physicalbridges that work in tandem together.Known as ‘auto-cascading’, the resourceswithin the physical bridge can be instructedby the software to operate in a ‘parent-child’ arrangement, with one bridgeowning the conference call, and theothers sharing the workload.

In the continuous presence mode ofpresentation, the bridge will automaticallyprovide the screen templates in which theviewers will see the other participants. Thebridge can also provide some administrativefunctionality for the call, such as assigningpasswords to enter each meeting andproviding IVR functionality where callparticipants can be greeted and instructedby customised voice greetings.

Although most participants will activelydial into a meeting, the bridge can beprogrammed to automatically dial out toparticipating locations and connect them toa conference. For example, the bridge couldautomatically wake up the cameras in apre-scheduled call, and participants wouldsimply have to walk into the video room atthe correct time and join the meeting.

Intelligent softwareIn order to build an architecture that scales,the conferencing software platform must beable to provide call signalling functionalityand dynamically manage the setup andmaintenance of a large number of calls.The software architecture has to be capableof re-configuring itself in real-time so theseresources are used to their best ability.

In addition, the software architecture hasto understand the bandwidth requirementsof each call being placed, the policy that isassociated with each one (its prioritisationand importance), and where the participantsare geographically located. By under-standing this, the software platform can uselocal resources instead of redirecting datastreams and call signalling to resources thatare far away – an approach that would eatup large amounts of bandwidth on WANlinks that are very costly.

The software should also be able toinstantly detect any failure of hardwareresources or loss of communication acrossinfrastructure links. This enables it to re-direct traffic and re-establish calls usingalternative resources, without overlyimpacting video calls or their quality.

When systems on different customerpremises try to join the same video callusing devices which run different protocols(i.e. H.323, RTV or SIP), the conferencingplatform must first perform protocolconversion to a common language so theinfrastructure can understand and processinformation correctly. In other words, thesoftware platform should provide intrinsicgateway functionality between devices thattalk different languages.

For instance, Polycom’s RealPresenceDMA sits in front of the bridges as aninterface between the outside world and thebridging resources. This optimises howincoming video calls are handled by thevirtual resources at its disposal. The DMAcan apply business rules that help it placeincoming meetings on bridges that makethe most sense, either for capacity,geography, or other priority rules. Thefollowing three scenarios show how this approach simplifies the process:

In our first example, Customer A inCalifornia wants to meet with Customer Bin New York, Customer C in London andCustomer D in Paris. They use video bridgesin Denver and Paris, and a virtualisationmanager on a server in London. In thissituation, the virtualisation managementsoftware would identify that two userswanted to join the call from the US andmay, for example, direct them to theresources on the Denver bridge. Likewise,the European participants may be directedto the Paris bridge with overall control ofthe call being given to the Master Denverbridge. Under this scheme, large amountsof data are not shipped over a transatlanticWAN, potentially providing cost savings.

video conferencing



The development needed for cablingarchitecture to support greater speeds in the datacentre is not a slight change or a gradual evolution; it’s a completelynew way of thinking about cabling. Thebasic components used to build a40Gbit/sec or 100Gbit/sec network arefundamentally different from those used to build a 10Gbit/sec network. It’snot an overstatement to describe the differences as ‘drastic’.

Enabling speeds of 40Gbit/sec andbeyond means moving from a 2-fibreduplex LC connector based system to a 12-fibre MTP/MPO connector basedsystem, with the inherent problem offailing to utilise the available fibrecapacity.

A traditional network utilises a simple 2-fibre duplex LC connector-based systemwith 10Gbit/sec of data transmittingdown one fibre and returning up theother. Transmission at 40Gbit/sec isbased on using eight fibres in the link –four transmitting at 10Gbit/sec in eachdirection. In a simple pass-through connection scheme, this means, for40Gbit/sec only eight of the fibres in astandard 12 fibre MTP/MPO Connectorare used.

It must be recognised that you cannotretrofit an MTP/MPO based system to aLC duplex system. An analogy would be;trying to run a modern high-speed trainon an old narrow-gauge track. It’s notthat it won’t work well; it won’t work at all.

What is needed is a suite of componentsthat enable datacentres to easily and cost effectively migrate their cablinginfrastructure to next-level applications,

including parallel optics and integratednetwork monitoring.

The Pretium EDGE AO solution offers thesimplest and most cost effective upgradepath for 40G and 100G using existingPretium EDGE infrastructure. Just swapout the modules and harnesses andleave your hardware and trunk cables in place, while maintaining 100% fibreutilisation and no increase in rack space.

Pretium EDGE AO solutions for paralleloptics consist of conversion modules andharnesses that allow networks to fullyutilise existing 12 fibre trunks whenmigrating to 40G, which uses eight fibres(four fibres transmitting at 10Gbit/sec ineach direction). Without this conversion,datacentres running 40Gbit/sec paralleloptics on their existing fibre backboneonly use 66 percent of the installed fibre.

Enabled by bend-insensitive Corning®ClearCurve® multimode fibre, thePretium EDGE AO solution enables thenext level of performance in your datacentre and Storage Area Network, whileproviding unequalled rack density,improved air flows, ease of access, quickdeployment and simpler MACs. The system also fully manages link polaritywith the ability for easy onsite changes.

The Pretium EDGE AO solution is a further enhancement to the PretiumEDGE product family and provides innovative flexible design options to suit your particular needs:

! Cross connect with a module-based design to give capacity for72 x 40G ports (576 fibres) in a 1Ufootprint, or

! Deploy a harness-based design toincrease capacity to 108 x 40G ports(864 fibres) in a 1U footprint.

Pretium EDGE AO Solutions provideadditional return on investment at100G, as the conversion modules andharnesses can remain in place for 100Gtransmission once the proposed 4x25GIEEE 802.3bm Ethernet standard isapproved (four fibres transmitting at 25G in each direction).

As part of Pretium EDGE AO Solutions,Corning is also introducing the firstintegrated port tap module for networkmonitoring of Ethernet 40GBASE-SR4multimode fibre parallel optical circuits.This passive tap device integratesdirectly into the Pretium EDGE Solutions infrastructure, and its all-MTP® footprint enables seamlessmigration with 40G electronics.

PRETIUM EDGE Conversion HarnessPRETIUM EDGE Conversion Module with reverse polarity MTP adapters

EDGE AOThe Painless Upgrade Path for 40G and 100G

Transcoding capability is crucial for high-quality video conferencing. For example, some meetingsmay have users who dial in from home on low-bandwidth DSL or ISDN. Without transcoding, theexperience of the many corporate users, who are on the higher speed corporate network, isdowngraded to the lowest common denominator of participants joining in from home.


The second example assumes the UScustomers are using an H.264 based systemwhile in Europe they’re using MicrosoftLync-enabled conferencing based uponRTV. In this scenario, the virtualisationmanagement software on the London serveracts as a gateway between the Lync and USvideo resources, converts the Microsoftsignalling, and establishes the whole callusing the bridges in the US and Paris.

This brings us to a third scenario. Let’ssay the above call is proceeding but thebridge in Denver suddenly stops function-ing due to a fire in the data centre. Thevirtualisation manager in London detectsthis and redirects the video traffic acrossthe WAN link to the Paris bridge. Usersconnecting via H.323 simply redial to re-join the call, with the administration andmanagement being performed seamlesslyin the background. However, for SIP-basedcalls there is an added advantage: theplatform will detect the problem andreconnect the participants back into the callautomatically, hopefully before the user haseven noticed that there was a problem.

To enable large-scale deployment andmanagement of video conferencing solutions, the software platform providesfor the management and maintenance ofhardware infrastructure componentsthrough a separate functional area: theDevice Manager. This will monitor andhelp troubleshoot any problems withdevices, and when software updates arerequired it will also help deploy them.

The Device Manager will also providereporting and comprehensive details ofvideo calls, processing the information to evaluate current system usage andexpansion plans for the video network.

Many organisations who have investedin video conferencing will inevitably needto be able to assist mobile or home workerswanting to dial into the company networkfor video calls with colleagues. Thesoftware platform must therefore providethe capability to enable and manage this.

Likewise, video conferencing-enabledorganisations will also want to use the tech-nology to communicate with their partnersand customers. This will only be possible ifvideo traffic is able to securely traverse thefirewalls from one customer to another.Firewall traversal is a particular challengeto video, as the data firewalls try to re-organise data packets. The implementationof a video firewall can eliminate this issue.

Moving beyond meetingsHistorically, the primary motivating factorfor most companies has been to use videoconferencing as a way of saving businesstravel costs. Recently, organisations havebegun to understand that the benefits ofvideo conferencing can also impact manydifferent areas including marketing,training, education, compliance, internalcommunications, advertising, PR, etc.

As the usage of video conferencing inthese fields has begun to grow, customershave discovered the potential to not onlyuse it to communicate in real-time, butalso to uncover the possibilities that existfor re-using digital recordings of pastevents and communications.

Moving beyond meetings, the sametechnology is being used to create digitallyencapsulated rich media which can then beedited, enhanced, archived, and broadcastacross multiple media. These assets canbe made available to target audiences on-demand. For example, the softwareplatform can be used for:! Live event multicasting – the streaming

of recorded webcasts, supporting boththe push and pull of video to thestreaming servers

! Video-on-demand – the automaticcreation of archived versions of any liveevent webcast so that customers canreplay them on demand, as desired

! Media management – enabling thecontrol of how video content will beaggregated, approved, categorised,edited and published

! Storage and archiving – establishingrules for the lifecycle of storage forbandwidth-intensive video content.Customers can determine how thecontent will be retained, transcoded andstored in the cloud, or across corporateresources without daily, hands-onmaintenance. "

The above article is based on the originalwhite paper, An Introduction to the Basicsof Video Conferencing, first published byPolycom in 2013.

video conferencingregister online @ www.networkingplus.co.uk to receive your free monthly copy


Many organisations will need to be able to assist mobile or home workers wanting to participate invideo calls. The software platform must therefore provide the capability to enable and manage this.



off-the-shelf: switches

Changing technologies and the data deluge can provide horrifyingmoments for networks. Perhaps the time has come to switch.

Buffalo Technology reckons its latest intel-ligent PoE switches allow SMEs to flexiblygrow their networks as their needs expand.

Two ranges are available. The BSL-PS-G21xxM is described as a series of “efficientsmart managed” Layer 2 switches toprovide SMEs with a “quick and easy”solution for otherwise complex networkimprovements and expansions. They comewith either eight or 16 ports, all with amaximum of 30W power supply per port.

The BS-POE series of managed Layer2 switches are available with eight, 16 or24 ports. They are said to come with anexpanded feature set, including integratedsecurity settings, designed for mediumand large businesses.

They also have a maximumport power supply of 30W which,according to Buffalo, provides

more than enough power to run IP cameras,wireless access points, and even forconcurrent use of 2.4GHz and 5GHz with 1000Mbps bandwidth.

All the new switches include PoEprofiler functions which automatically turn them on and off at a scheduled time or date. They also come with Buffalo’sLoop Guard warning function whichimmediately signals when and where anetwork problem occurs. Other featuresinclude VLAN tagging and supported USB flash recovery which the vendor says significantly eliminates down time.

As more businesses adopt cloud-basedservices, Allied Telesis reckons it canhelp them to beat bandwidth congestionand service delivery problems with itsx210 and x510 series of switches. Theysupport IPv6, and are said to include afull range of security and resiliencyfeatures, coupled with easy management.

The x210 Series gigabit Layer 2+switches are the entry-level products inAllied Telesis’ x-Series range. The firmsays they offer “powerful” control overnetwork traffic, secure management options,and multi-layered security features. Forinstance, they include tri-authentication toensure only known users and devicesaccess the network, with sepa-rate controlled access for guests.

The x510 Series of stackablegigabit switches includes twoPoE+ models, the AT-x510-

28GPX and AT-x510-52GPX. It’s claimedeach one can provide up to 370W of powerto support high-powered devices such assurveillance cameras and IP phones, thuseliminating the need for electrical rewiringand reducing office noise pollution.

Allied Telesis says a “great advantage”of the x510 Series is its upgradeability fromLayer 2+/basic Layer 3 functionality to fullLayer 3. Customers can either purchase theproduct with the full Layer 3 premiumlicence, or upgrade their devices in thefuture as their business and network needsgrow. The firms reckons this ability toincrease the capability of existing productssignificantly increases return on investment.

Dell says its S4820T top-of-rack (ToR)switch is purpose-built for deployment inhigh-performance data centre and cloudcomputing environments. According to thefirm, it leverages the popularity of twisted-pair copper cabling and provides the “rightmix” of scalability, performance, opera-tional simplicity and cost-effectiveness.

The switch is powered by Dell’sForce10 operating system and is designedto support both Layer 2 and Layer 3 func-tionality required for Web 2.0, enterpriseand cloud server provider data centres.

It offers 10GBASE-T connectivity forseamless migration from 1Gbto 10Gb servers: Dell saysthis is especially forcustomers looking toconsolidate server I/Oat 10G, as it enables

low-co

Documents

Features How the latest IT systems are helping data deluge could … · 2014. 1. 13. · Axell Wireless advocates the use of a Distributed Antenna System. It says this takes the signal