Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft Eppler 12 May 2003GSI Darmstadt Slow Control System for Neutrino Experiment KATRIN Wolfgang Eppler,

Forschungszentrum Karlsruhein der Helmholtz-GemeinschaftEppler 12 May 2003 GSI Darmstadt

Slow Control System for Neutrino Experiment KATRIN

Wolfgang Eppler, Forschungszentrum Karlsruhe

Armen Beglarian, Forschungszentrum Karlsruhe

Volker Hartmann, Forschungszentrum Karlsruhe

Surik Chilingarian, Yerevan Physics Institute

Simon Kelly, Bristol University of the West of England


Neutrino Experiment KATRIN

- tritium beta-decay

- measures mass of electron neutrino to a precision of 0.35

eV

- scales up to previous experiments by an order of magnitude


Slow Control System Architecture

Convert

Convert

DB

WebServer

Supervisor

Control

XML OPC-HDR

XMLQuery

relationalOracle

STxx

OperatorControl

LabViewDSC

XMLSchema

Extreme use of standards: protocols, interfaces, concepts

anydatabase

WebClient

HTTP

NetscapeInternet Explorer

DataAccess

MagnetHeating/Cooling

Ultra-HighVacuum

High Voltage

DetectorCRYO-system

FP-1 FP-2 FP-3 FP-4 FP-5 FP-6

DistributedControl


Experimental Impressions

Testcylinder

LabView GUI

Web GUI (TOSKA, not

KATRIN)


Steps Towards a Highly Standardized System (1)

Oracle DB

RT Control

Admin Measure

Clobs Tables

data2xml

OCI Driver

Web Server

relat2xml

PL/SQLOCI Driver

XQuery

XSU

Oracle DB

RT Control

Web Server

PL/SQL

XQueryon Clobs XSU

Oracle specific

In more detail

administration data

measurement data


Steps Towards a Highly Standardized System (2)

XML DB

RT Control

Admin Measure

OPC-XML Server

JDBC Driver

Web Server

relat2xml

OPC-XML

XQuery

XQuery

OPC-XML Client

XQuery

XML DB

RT Control

Web Server

XQuery

XQuery

OPC-XML

Oracle DB

RT Control

Web Server

PL/SQL

XQueryon Clobs XSU

Oracle specific

administration data

measurement data

In more detailNext step


OPC vs OPC-XML

- COM/DCOM not available on non-Microsoft platforms

- Consequently no integration into enterprise applications

without COM/DCOM interface

- Firewall problem as DCOM uses dynamically allocated TCP/IP ports only

- Data representation of OPC-XML as text:

-> more network traffic (<35%), more CPU ressources

- OPC History and OPC Alarms and Events not yet available as XML

- Restricted data subscription with XML

OPC Drawbacks:

OPC-XML Drawbacks:


OPC-XML/HDR (High Data Rate)

Purpose:

• Higher data rate by better binary data encoding und use of UDP

• Compatible with OPC XML DA

• Better security

• Stringent use of XML paradigm for user‘s data presentation

• Support of multicasting

• Support of data subscription

2 alternatives:1. SOAP message with attachment

• Still working draft in W3C• No support by major XML libraries• No multicasting

2. HTTP message with MIME multipart type and XLink• No multicasting, but separate connection as done in FTP protocol• New user defined data type described by XML Schema


HTTP/MIME/XLink-ExampleContent-Type: multipart/related;

boundary=xxxxxx--xxxxxxContent-Type: text/xmlContent-ID:OpcXmlHdrMessage<?xml version=”1.0”?>......<Value xsi:type=”BinaryStream” ><BinaryStream type=”SomeType”

records=”5” recordsize=”64” xlink:type=”simple” xlink:href=”cid:BinaryData1”/></Value><Value xsi:type=”MulticastStream”><MulticastStream type=

”SomeType” recordsize=”64” xlink:type=”simple” xlink:href=”mcast:224.0.0.1”/></Value><Value xsi:type=”XMLStream”><XMLStream xlink:type=”simple” xlink:href=cid:TypeDescription1/></Value>......

--xxxxxxContent-Type: application/binaryContent-Transfer-Encoding: Little-EndianContent-Length:Content-ID:BinaryData1..... Binary Data ......

--xxxxxxContent-Type: text/xmlContent-ID:TypeDescription1... XSD Schema ...


Implementation of OPC XML HDR

Internal DataRepresentation

Secure XML DataRepresentation

XLink / MIME / HTTP Message

Transport Level

XML DataRepresentation

LibOpcXml

LibOpcHttp

Libraries

LibOpcSecure

LibOpcTransport

Web Server

OPC Server

Servers

XML ParserXML Validation by XSD Schema

Access Rights CheckXML Document Signature CheckXML Document Decryption

MIME/XLink/HTTP Encapsulation

Message Communication by IPv4 / IPv6


Which XML Parser?XML libraries used for Benchmarking:

• LibXML2 + GDome + LibXSLT + XML Security • Apache Xerces for C + Apache Xalan for C • IBM XML4C + IBM Lotus XSL • Expat + CenterPoint XML + Sablotron + Arabica • RXP Parser • Oracle XDK for C/C++ • Oracle XDK for Java • QT XML Module • Sun XML Pack Summer + Apache XML Security

Performance measured in total and in following sub-tasks:

•Non-Validating Parsing with Native, SAX, DOM Engines Benchmark

•Creating + Serializing DOM tree Benchmark •Schema Validation Benchmark•XSL Transformation Benchmark •XML Security (Signature, Encryption) Benchmark

Expat Xerces LibXML Oracle Sun0

1

2

3

4

5

6

7

8

9

10 121,46

Parsing (SAX)

Parsing (DOM)

Building DOM

Validating

XSL Transform

Security: Signature

Security: Encryption

Overall

-> LibXML2 was chosen

(In spite of no XML Schema validation)


XQuery vs SQL

0

200

400

600

800

1000

1200

1400

010

0020

0030

0040

0050

0060

0070

0080

0090

00

1000

0

records

tim

e (

se

c) 1 Signal - SQL

1 Signal - Xquery

1 Signal - no document build

4 Signals - SQL

4 Signals - Xquery

XQuery

XQuery (1 Signal)

XQuery (4 Signals)

SQL (1-4 Signals)


XQuery vs XSU

0

200

400

600

800

1000

1200

1400

1 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

records

tim

e (

se

c)

XQuery (4 Signals) XQuery (1 Signal)

XSU (8

Signals)

XSU (1-4 Signals)

(XSU: XML SQL Utility (Oracle))

Each signal with own XML structure


XSU Improvements (1)

0

50

100

150

200

250

300

350

1 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

records

tim

e (

se

c)

21 Signals

16 Signals 12 Signals

1-8 Signals

1. (big) improvement: 1 timestamp for all signals


XSU Improvement (2)

0

50

100

150

200

250

300

1 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

records

tim

e (

se

c)

1 Signal

4 Signals

8 Signals

12 Signals

16 Signals

21 Signnals

2. (small) improvement: Timestamps in separate XML structure


XSU Improvement (3)

0

5

10

15

20

25

1 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

records

tim

e (s

ec)

1 Signal

4 Signals

8 Signals

12 Signals

16 Signals

21 Signals

3. (big) improvement: use of internal XSLT


Conclusion

Still many problems with XMLEspecially problematic:

• XML Query (-> Oracle XSU)• OPC XML (-> OPC XML HDR, OCI, LOGOS)

Significance of XML in future undoubtful More and more same techniques in

• Industrial automation• Office applications• Enterprise solutions• Physical experiments

because of same components:• database, internet, browser, documentation, GUI, graphics, ...

and of same problems:• fast data access, good searching capabilities, flexible software, ...

-> Connection to Grid Computing

Documents

Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft Eppler 12 May 2003GSI Darmstadt Slow Control System for Neutrino Experiment KATRIN Wolfgang Eppler,