Transcript ppt

ITPA Expert Group Meetings, October 2002, Cadarache, France
STATUS and PLANS for the ITPA PROFILE DATABASE
C M Roach, M Walters
EURATOM/UKAEA Fusion Association, Culham Science Centre, Abingdon, Oxon,
OX14 3DB, UK.
CONTENTS
• Brief Infrastructure Overview
• Recent Data additions
• Tool Developments
• Integrating the ITB Profile Data
• Conclusions
This work was jointly funded by the UK Department of Trade and Industry and EURATOM
ITPA Expert Group Meetings, October 2002, Cadarache, France
Block Diagram of Profile Database Infrastructure
Main JET Firewall
Profile DB sits on
PC, OUTSIDE MAIN
JET FIREWALL.
backed-up inside
firewall
OUTSIDE MAIN JET
FIREWALL http://tokamakprofiledb.ukaea.org.uk
INSIDE MAIN JET
FIREWALL
Data mirrored,
backups.
RedHat Linux7.2
Web Site: with tools and
documentation
http://tokamakprofiledb.ukaea.org.uk
Public and working versions
of profile database
DB Manager
Users:
Access through: ftp, http and
MDSplus:
RDB search tool
http / ftp /
MDSplus
Less restrictive
Firewall
remote access to
DB for user
tools, MDSplus,
r/w, display,
analysis
ITPA Expert Group Meetings, October 2002, Cadarache, France
INFRASTRUCTURE OVERVIEW:
What is There?
• Read access to data available through
 http and ftp to Ufiles (also a few eqdsk files)
 MDSplus to trees containing Ufile data
• Various graphical tools are available, and there are a number of example routines
to read the data.
• Relational database now developed and this provides an indexing system to help
users to find discharges of interest
 relies on structured comments and global 0D data
• Data submissions presently through emailing Ufiles to the database manager.
ITPA Expert Group Meetings, October 2002, Cadarache, France
INFRASTRUCTURE OVERVIEW (ctd):
What is Still to Come?
• Plan improved automation in data submission procedure
 present system has not been problematic so far.
• Direct write access through MDSplus and ftp.
 agreed in principle, as security holes will not threaten JET/UKAEA.
 machine itself may of course become vulnerable.
 automatic checking procedure, prior to uploading database.
• Need tools to write MDSplus trees, conforming to a set of required standards.
• Users need to learn to use these tools.
• Improvements to the relational database.
• More physics tools (from users)
ITPA Expert Group Meetings, October 2002, Cadarache, France
Addition of data
User adds new
Ufile or tree
using ftp or
MDSplus server
Incoming
datastore
cron job
loops over
files
Run
checking
code
or to
r
r
e or
n
O err er
nd vid
e
s p ro
If successful
Conversion
Convert to
Ufile/tree
move Ufiles
and tree
Load
IPRED
Main
datastore
ITPA Expert Group Meetings, October 2002, Cadarache, France
NEW DATA SUBMISSIONS to the PROFILE DATABASE
There have been a number of data submissions since the Princeton Meeting
• D3D: #99411, a high performance ELMy H-mode (M Murakami)
• JET: #46664, 51599, 53521, 53532, 53537, ITB discharges (P Stubberfield and V
Parail)
• TS: #18302, 18305, 18319, 18368, 18370, 18372, 18488, 18490, 18500, 18496,
18504, 18507, 18520, 18765, 18768, 18769, 18771, 18774, 18775, 18778, 18780,
18790, 18792, 18799, 18801, 18805, 18807, 28 electron heated discharges (T
Aniel, and T Hoang)
Promises of discharges from TCV and T10
ITPA Expert Group Meetings, October 2002, Cadarache, France
TOOL Developments
NTCC Data Server:
Java package for plotting profile database data at any web browser (J Carlsson, J
Cary and A Kritz).
– Presently operational for PR98 data
– Working on authentication to access the working data.
– Will be advertised when fully functioning for working data
ITPA Expert Group Meetings, October 2002, Cadarache, France
TOOL Developments
Relational Database Search Tool:
ITPA Profile RElational Database, IPRED
– Relational database, using MIMER RDBMS
– IPRED contains all the signal metadata and the comments and zerod data.
Currently the only metadata stored is the signal size information.
– Also contains all the discharge, file and MDSplus node name information.
– makes possible an extensive range of searches or data retrievals eg:
 select all discharges providing twod variable DELTAR
 find discharges where STATE=‘STEADY’
 seek discharges with ‘ITB’ appearing somewhere in the comments data
 obtain dates of all discharges submitted to the database
– web interface at: http://tokamak-profiledb.ukaea.org.uk/IPRED.html
– IPRED is an excellent tool to find discharges of interest
– user feedback would be much appreciated, so please use it !
ITPA Expert Group Meetings, October 2002, Cadarache, France
IPRED table structure
tblCommentLabels
tblComments
PK
PK labelID
tbl1d
commentID
PK
FK1 dischargeID
FK2 labelID
comment
label
tblFileType
PK filetypeID
fileExtension
mdsNode
fileCSVExtension
fileDescription
tableName
tblTypes
PK type
typeLabel
missingValue
tblFile
PK
oneDID
FK1 dischargeID
name
size
tblPulse
fileID
PK dischargeID
tbl2d
FK1 filetypeID
FK2 dischargeID
signalID
name
FK1 type
units
PK
tbl0dData
tbl0d
tbl0dNames
PK
tokamak
pulse
PK
zeroDID
FK1 dischargeID
FK2 signalID
PK
dataID
FK1 zeroDID
timesliceID
data
twoDID
FK1 dischargeID
name
sizeX
sizeY
ITPA Expert Group Meetings, October 2002, Cadarache, France
TOOL Developments
IPRED (ctd):
Future Plans:
– client library will be made available so searches can be made in software
– SQL interface will be provided
Proviso:
High quality and complete ZEROD and COMMENTS data is crucially important
in helping to maximise the usefulness of any relational database tool such as
IPRED.
ITPA Expert Group Meetings, October 2002, Cadarache, France
INTEGRATING ITB PROFILE DATA
ITB database has 31 discharges with profiles: D3D (10), JT60U (9), RTP(6), TFTR(6)
12 out of the 31 ITB discharges were already in the Profile Database.
ITB database uses UFILE formats that are very similar to Profile Database, though
zerod information is very different and is all stored for all discharges in a single file
 integration possible, and desirable
 converging the databases allows sharing of tools and infrastructure
All discharges from d3d, tftr and rtp (22 out of 31) have now been integrated into the
required formats for submission to the profile database
 only 3 of the d3d discharges have zerod information (from csv file)
 see http://tokamak-profiledb.ukaea.org.uk/ITB/index.htm (‘itbdb’ , ‘wk-grp’)
 also converted to MDSplus trees treename itb_tok, MDSplus server
‘tokamak-profiledb.ukaea.org.uk:8002’ (NB access requires registration)
 jt60u discharges just need comments files and appropriately formatted 2D
ufiles (jt60u 2d ufiles are presently written as sequence of 1d ufiles)
Various difficulties encountered in integrating the data, including missing comments
files and UFILE formatting issues
ITPA Expert Group Meetings, October 2002, Cadarache, France
INTEGRATING ITB PROFILE DATA (ctd)
Incompatible 0D data is biggest issue
– 78 0D variables in PDB, but 126 0D variables in latest ITB database
– nevertheless extra 0D variables can be accommodated in PDB
0D variables have unfortunately (but inevitably) diverged slightly:
• Some ITB 0D variables are stored at >50 timeslices => isn’t this really 1D data?
• Identical physical data has been stored similarly but differently :-(
– eg WALMAT (PDB) => WALLMAT (ITB)
– eg PGASA and PGASZ (PDB, integers)=> PGAS1 (ITB, string)
– data access will be more convenient if we can resolve these (trivial)
incompatibilities agree and maintain uniform conventions
• tok = JET or EFDA-JET , and DIIID cf D3D (PDB)
… better to remove these trivial differences?
ITPA Expert Group Meetings, October 2002, Cadarache, France
INTEGRATING ITB PROFILE DATA (ctd)
How Integrated should the ITB profile data be with the profile database?
ITB profile database now stored separately from the profile database
 which avoids incompatibility issues for now
 but … Is this really necessary in the longer term?
Possibly more convenient for users if ITB discharges are simply part of the profile
database (integration issues).
 Would this be desirable?
If we desire STRONGER Integration:
• Profile Database manual should be updated in collaboration with ITB group to
define a consistent set of variables incorporating additional ITB variables
ITPA Expert Group Meetings, October 2002, Cadarache, France
CONCLUSIONS
• Infrastructure for the ITPA Profile Database is improving continually
• Further development requires and will respond to user feedback.
– please contribute to this process!
• Integration of the ITB profile data should be mutually beneficial, and will hopefully
encourage increased use, but there are a number of important but straightforward
issues that require to be resolved.