Minutes of the GSE UK IMS Working Group Meeting
IBM Warwick, 8^th September 2000

Chairman: Frank Fleming (frank.fleming @ barclays.co.uk) - 01565-613581

Secretary: Peter Armstrong (peter_armstrong @ bmc.com) – 0831 400154

Administration

• Introduction
  • Frank welcomed us all, but the turnout was pretty low – only 20 people. Not sure if this is location or because we are DB only or timing or what? Little pack of agenda etc distributed by Frank including problems form, feedback form etc.
  • Frank ran through and overview of agenda, and welcomed the new IBM rep Niel Kenyon.
  • Andrew has unfortunately had to resign due to health reasons and Alan has finished his spell as IBM rep – any ideas for presents to Frank please.
  • Don’t forget to pay Ian. Please update the membership list, which was passed round.

• Correspondence
  • Websphere, MQ and CICS conference in Amsterdam in October – flyer
  • Unix technical university – 9 to 12 October in Disneyland Paris
  • GSE 5^th technical symposium in Montpelier on 15/16 October

  These events are jointly organised by IBM and GSE – you get a voucher back after the event, which can be used as a discount for future events / membership costs.

  • Minutes from most recent large systems user group – there is a bogus MQ user group (there is a real MQ user group, this is someone pretending). They are calling up members and claiming to be from the user group – user groups do not do cold calling – we don’t give out membership lists or email ids. Let us know if you are being hassled chairman@mqug.org.uk
  • GSE ISV pricing initiative going on – started in Germany. GSE UK is running a meeting on 28^th September at a hotel near Slough. You should have got a letter from Phil about it.
  • Every two years the region Manager has to be (re)elected. Phil Howard-Knight has been doing the job for almost four years and there is a nomination form – any volunteers?
  • Date conflict with DB2 working group for next meeting? We moved this meeting to avoid conflict – haven’t heard from DB2 chairman yet, but they are trying to move their meeting. We will pursue further.

IBM What’s New – Niel Kenyon, IBM

• Education details. As usual contact Pete or Alan of you want non-scheduled courses. New course being looked at – IMS DB and TM Systems Programming course for people installing IMS for the first time. Thought is to have a DBCTL version as well – let Frank know if you are interested.
• IMS Tech Conference in Anaheim – 23 to 26 October, 2000. GA date for IMS V7 is 27 October.
• IBM Fault Analyzer for OS/390 – supports IMS.
• Some interesting APARs on SHARK, on FICON for OSAM, buffering for batch backout, shared queues etc.
• Recent benchmarks have reached 11246 transactions/sec with IMS data sharing and shared queues on a 2-CPC sysplex – gosh, some of us remember when we were happy to reach 3 trans/minute.
• Several new red books including HALDB guide – SG24-5751-00. Pete says there is a new book on increasing availability – look at www.redbooks.ibm.com and search on IMS for this and other interesting bits. Have to register now, but worthwhile as you can download books etc.
• Ian whinged about www.dbazine.com not being updated - Peter to check.

OSAM – the best thing since ….. – Pete Sadler, IBM

• What was the best thing before sliced bread?
• Pete reminded us of the wonderful OSAM, which some people forget about in their enthusiasm for the Very Strategic Access Method. Indexes have to be VSAM as ISAM is dead – hooray! But you can have HIDAM with KSDS index and OSAM data for instance, and people do. OSAM is part of IMS (used for logs, message queues etc.) and hence is strategic too and has some neat features that VSAM doesn’t have like chained writes and parallel writes. It is also more flexible in terms of blocksize, buffer pools etc.
• Pete also compared OSAM Sequential Buffering with HSSR, Background Write, the 8G limit. Watch out for multi-volume datasets in OSAM, but this is well documented. There are some other issues to convert (e.g. put blocksize in DBD, update buffer pools etc.), but they are not major – the conversion process is simple.
• HALDB supports OSAM and VSAM, but there is a limit of 4G for each OSAM – however, it doesn’t matter as you just have more partitions. Pete recommends more small partitions rather than one colossal DB (and the same applies for OLDS).

• Comment from Peter A – would be nice if VSAM Hiperspace did writes to bufferspace (rather than write through) as well – yes, but that means you have to recover from the log if it goes wrong. This is concept used by VSO DEDB and that is why you are recommended to have two copies of them in case something goes wrong.
• Pete recommends DEDB, followed by HDAM OSAM followed by the rest.

Handouts were provided.

DB Control Suite – Niel Kenyon, IBM

• Niel took us through this product from IBM, which is designed to make life easier for IMS DBAs. Product is designed from a task point of view rather than a utility point of view, and it is all ISPF driven. It is also designed for those sad people (Peter A’s words) who haven’t gone to DBRC yet – it will build all the DBRC registration stuff for you.
• He was then very brave and gave us a live demo of the product. You can use it with standard IMS utilities, BMC, IDI, Neon or IBM DB Tools. This is not the old Swiss Bank product – that became DBICF and is very much a utility oriented DBRC front-end, whereas this is designed as a task-oriented management tool.
• It does not force you to run all the steps, but you can also run the jobs / steps in batch.
• Product runs from skeletons, which you can tailor to your standards, and you can group your DBs in all sorts of different ways – e.g. by application or whatever.

Hints & Tips, Problems and Contacts

• Beware moving UCBs above the line – people have done this and killed V5.1 as the logger uses OSAM which can't access a UCB > 16Mb.

• Name:Chris Huffington, Friends Provident

  Problem:Is anyone successfully using IMS V6 with ACF2 successfully?

  Contacts/Comments:Contact vendor? AIB?

LUNCH

RECON Performance – Peter Armstrong, BMC Software

• There had been a spate of entries on the IMS Listserver recently about RECON performance, RECON backup and repair etc. Ian asked Peter to run through them and summarise the correct answers. Peter ran through the procedures for backing up and repairing RECON, monitoring RECON, how to reorganise it whilst IMS is running, and how to set up the buffers etc.
• Being a tedious old fool, Peter had also read the manual (not his, the DBRC Reference manual), which actually tells you that you should reorganise your RECONs regularly and also tells you how to do it (actually the same way as he recommends in his book, which is probably where he stole it from)
• There was also a discussion about the buffering for RECON and whether it had changed in V7 –
• Alison: "At the IMS GSE Meeting on Friday the recommendation to increase the number of buffers for the RECON was mentioned via DSPBUFFS. You said that the default number has been increased. If the defaults are as documented in the Customization Guide then for V6 for LSR they were 6,12 ; and in V7 they are now 60,120. When we wrote the Redbook on Data Sharing we added this section: "Give DBRC plenty of buffers; 16 index and 64 data buffers should be sufficient." So hopefully we can just leave the system to default now ?"
• Here is the answer from Ron Blake of IBM:

  "Yes, for V7 the default LSR values of (60,120) coded in DSPBUFFS should be fine. For V6, the default values coded in DSPBUFFS were increased to (12,24) and may be sufficient for many customers. If not, customers can use the Redbook or the V7 DSPBUFFS values for guidance."

  Handouts were provided.

Buddy, can you spare a €? - Discussion Topic

• Ian introduced the problem of Euro – on Jan 1^st 2002 most of Europe moves over to a single European currency. Does this matter to us or not?
• As a starting point, it is just another currency, and the issue is displaying it externally – your underlying DB will probably not change, but can you display the symbol when required? If we join, then what will the rules be? Now you have to handle dual currencies for a while. If we finally get to one currency, then it gets easier again, but there is definitely significant work required between here and there. Conversion has to be done to 6 decimal places.
• Are there options in IMS to help me? How about the Y2K conversion exit, or the compression exit? Will it convert to the same both ways? Do you have to store the data twice?
• Conclusion – you have to change all your application code to handle it properly. Can’t do it all from systems programmer point of view.
• General discussion on why etc.

AOB

Thanks for coming – see you all next time on December 7^th at IBM South Bank.

Feedback

It looks as though most people enjoyed some of the topics yesterday.

The scores on the board are :

Topic	Comments	Score
IBM - What’s New?	Nice to be told about the web sites (red books etc). Would be nice to have a handout at the time. I would have expected V7 to feature more prominently. Missed it. N/A - came late.	4.07
OSAM - The best thing since …	More ammo to use to persuade people to consider the benefits of OSAM for volatile shared databases. Excellent. Could have some useful application. Preaching to the converted N/A - came late.	4.06
DB Control Suite	No operational DBAs. Handouts. Looks useful. I’ve seen it already (Barcelona) but still learnt a lot.	3.69
RECON Performance	Excellent thought provoker. I’ve read Peter’s book - over and over.	4.35
Buddy, can you spare a Euro?	When we get the rules. We’ve already done it.	3.29
Rating of IBM, Warwick	Food good, room fine. Easy to get to by car. It’s local! Suits me, I live in Northampton. I prefer to travel by train. Tea ran out. Yes - come again (I’m biased (local))	4.31

Offers of presentations

• ODBA
• Recovery without WADS

Thank you - we will be in touch.

Requests for topics

None received