Hello,
I red a thread about db2 clustering, and somebody suggested to link together smp machines .
Please help me to find an article , redbook and something else to read the base of this technology.

Thanks,
Szilva

Are you referring to DPF - Data Partitioning Feature.

I think, no. I want to build such application as similar to oracle cluster. I think , the HADR is similar to oracle standby. I red an letter on dbforum, and it was suggested to build frame with smp machines.

Are you referring to PureScale, may be?

DB2 pureScale Feature solution overview

All computers, even laptops, are SMP these days as they have multiple processors (or cores).

There are three main kinds of clustering in DB2:

• Traditional shared disk cluster with virtual IP that switches between servers.
• DB2 HADR
• IBM PureScale

The first two options can be controlled with IBM Tivoli System Automation, which is included in some DB2 editions of DB2 and automatically installed when you install DB2 (ESE, WSE, but not sure about others).

PureScale is like Oracle RAC.

DPF is a parallel shared nothing database architecture designed primarily for data warehouses, and now is part of InfoSphere Warehouse.

The first clustering method is intresting to me. I would like to ask what method use db2 to syncronize update and insert origaneted from different servers.
Oracle uses cache coherency.

Thanks.

In the first method, shared disk means that there is only one set of disks that are shared by both servers. When you failover, you have to unmount the disk from one server, and remount on the other server (they could be simultaneously mounted on both servers if you make absolutely sure the DB2 instance is only started on one of the servers at a time).

The method where mutiple servers are using the same disk is DB2 PureScale, which is functionaly like Oracle RAC. You can google PureScale to find out the details of how it works.