It's all about the answers!

Ask a question

What is the recommended topology for CLM 6.0 (with fewest number of servers)?


Eduardo Bello (4401621) | asked Jun 29 '15, 10:53 a.m.
 Hi guys,

  I can use a little help to define a new topology for CLM 6.0. 
 
  With version 5.0.2 the topology for approximately 400 users (begin 100-200 simultaneously) would be something like that:

   CLM Server 1 (active):
      Disk: 60gb 
      RAM: 32Gb
      Processor: 8x Cores, 3.2ghz
      Applications: jts, ccm, qm, rm, converter, rs and converter

   CLM server 2 (standby): Exactly the same of CLM Server 1

   RRDI Server: 
      Disk: 60gb 
      RAM: 16Gb
      Processor: 8x Cores, 3.2ghz
      Applications: rrdi

   Server Monitoring: 
      Disk: 20gb 
      RAM: 4Gb
      Processor: 4x Cores, 3.2ghz
      Applications: csm


  But now in CLM 6.0 we have new applications: Lifecycle Query Engine (/lqe), Link Index Provider (/ldx) and Global Configuration (/gc). And now Data Collection Component (/dcc) is the only way to ETL, so we have to install it as well.

  My question is, which of those new applications can I install on the same CLM server and which one do I have to install on a dedicated server?

  I did read that Data Collection Component (/dcc) would be recommended to have it own server. But what about lqe, ldx and gc?

  The goal is to have the minimal number of servers but without compromise the performance. What would you guys recommend?

One answer



permanent link
Guido Schneider (3.3k137699) | answered Jun 29 '15, 3:56 p.m.
We run the whole CLM environment on a single physical server. But the server has sufficient memory, currently 128GB (planned to extend to 256GB), and 32 logical cores (2 cpu with 8 physical cores, 2.1GHz).

Because we use Websphere as application server and the fact, Websphere is an excellent virtualization layer, we can run all applications in paralell, but each in its own WAS profile with its own Jaca process and virtual memory.

So we have about half of the memory used for Was profiles (64GB) and rest is for DB2 process and DB caching.

This performs great for our environment with about 1000 active users.

Only issue we see is the datawarehouse 32k tablespace. The disk queue can go up to 5 or more when somebody loads dashboard widgets with large number of workitems the first time. For this we plan to put the DW tablespace and index onto a SSD.

Cheers
Guido


Your answer


Register or to post your answer.