Hi,

I have this question can some one clarify it.

What is Clustering in DB2 ?

Is it the same as HADR available.

In my environment, i want to install Db2 in 2 servers and have a cluster , how to configure those cluster on it ?

I need can cluster be done without creating database ?

pls help me ....

Thanks!

Read about DB2 pureScale:

IBM - DB2 pureScale Feature

There are multiple ways to cluster in DB2:

• Shared disk (similar to simple clustering on other db's)
• HADR (separate disk for both primary and standby database which are replicated via DB2 transaction logs)
• Data Partitioning Feature (DPF). A single table is spread across multiple partitions which can be on different servers. This is normally used for data warehouses but can be used for OLTP (see the TPC-C Benchmark results where DB2 now holds the world record).
• PureScale - similar to Oracle RAC.

In most cases, HADR is the best and easiest to implement solution.

Thanks Marcus,

i had a view on this.
Now my doubt is im planning to do in such a way..

2 server both should have the DB2 9.5 and then the data and log files are available in the storage SAN.

How to configure Clustering in this Databases ?

Can we configure during installation / it has to be done after installation ?

There is manual called Data Recovery and High Availability Guide and Reference that you can download from the IBM website. You might want to also look at the Partitioning and Clustering Guide.

hey i had a look into that book,

very basic question ..don mistake me.

I need to know that do we need database to be created only then we can do clustering ??


why im asking is because in MS-SQL you will have a option to install the cluster- during the installation..


pls clear it ..

Many Thanks!

Hi,

if you know the HA-Features of SQL-Server.

MS Cluster ist the Shared Disk Approach Marcus Mentioned.
HADR is like SQL-Server Database Mirroring.

You could use SAN for both HADR and Shared Disk. You need to install DB2 and setup the Database before configuring HADR.

Thanks all..

I got a view as how HADR and Clustering is ..

Now i have a question as the procedure of Clustering using the IBM Tivoli SAMP.

can some tell me the procedure... i am not gettting the clear idea of which has to be done first and which to be done next.

help me...

Hi,

- install DB2 on both Servers (Primary and Standby). (Same Version of DB2 on both Servers)
- Setup Database on Primary (Same Filesystem-Structure for Database on both Servers)
- configure HADR (Initializing high availability disaster recovery (HADR) - IBM DB2 9.7 for Linux, UNIX, and Windows)
- run db2haicu (Configuring a cluster for high availability - IBM DB2 9.7 for Linux, UNIX, and Windows and ftp://public.dhe.ibm.com/software/da...R_db2haicu.pdf)

General Link:
Configuring for high availability - IBM DB2 9.7 for Linux, UNIX, and Windows

Hey Nvk,

Thanks for the info,

I have a silly doubt, for configuring cluster in the procedure mentioned by you.. it requires the database,

if suppose i want it to be instance level clustering or if im adding another database will it be automatically added to cluster ?

y im asking is because i want to have the cluster from DB2 level..not with respect to Database level.

Is it possible in DB2 ?

Hi,

HADR + TSAMP is working on the Database Level. If you want to work on Instance Level, then you have to use a Shared Disk approace or a little bit more sophisticated PureScale.

Cheers,
nvk

nvk, you made it clear after lot many days...

see i will let you my setup...

I have 2 linux box which is with the DB2 v9.7
both is sharing the same (data & log files ) which is in Storage SAN.

which does not contains Databases as of now.

now i want to configure failover on this.....

how can i do it, pls put in your suggestion.

Well,
i've never setup an environment like this.

Since you want to work on a Instance-Level, you should create your Instance on a Filesystem or a Filesystemstructure which could be mounted on both Servers. This includes the Home-Directories for the Inst-Owner, the Fenced User and all the Directories for the Instance (diag-path, Working Directories) and the Databases (Data, Logs, ...).

If a Server fails, you could mount the Filesystems on the Standby-Server and restart the Instance. The tricky Part is to automate the Failover of the Instance and to avoid Split-Brain Situations. Off the top of my head i don't know a Tool to easily configure such an Environment.

Thanks, don you mistake me.. as im very new to this environment..

i jus need some suggestion as

1. can i configure the clustering on that setup using SAMP / dbhaicu

2. Is db2purescale possible in that setup ?

3 .else what kind of setup can i undergo..

gimme your views ..which will be very helpfull.

Hi,

db2haicu is a script to configure a database for automatic failovers with HADR using SAMP. So i won't help. SAMP should work for your needs, but don't have that much knowledge about SA MP.

No, PureScale requires a special hardware-setup, e.g. InfiniBand


If you are new to this topic, you should read the Documentation in the Information Center regarding High Availability (High availability with IBM Data Server - IBM DB2 9.7 for Linux, UNIX, and Windows). This may help you to identify which approach fits your needs.