Insight Manager Standby Server

[Simon Doy]

Dear All,
We have our live HP SIM server and we are looking to implement a standby server and was wondering if anyone else had already done something similar.
I have been looking for whitepapers on the HP website but to no avail.
Here is the proposed solution, both live and dr insight manager boxes will have HPSIM SP2 installed with database replication between live and dr occuring (using NSI's Doubletake). Is this going to cause problems? (e.g Does HPSIM database holds the name of the server that its installed on).

SNMP traps will be looked after by creating a DNS CNAME record called insightmanager and that will be pointed to the live insight manager box.
The management agents of the servers will trust both insight manager boxes and will accept SNMP connections from both as well.

My main worry is the database replication as it may pull over data which is specific to the insight manager box and this will cause problems. Does anyone have any insight (excuse the pun) into what data may have to be changed? Any thoughts? Does this make sense?

I will keep this forum up to date with my progress.

Regards
Simon Doy
Infrastructure Consultant
Link Interchange

[David Claypool]

The only supported solution is using Microsoft Cluster Services.

http://www.hp.com/go/hpsim --> Information Library --> "Deploying HP Systems Insight Manager on MSCS Clusters"

[Brent Seizer]

How about using 2 servers running Win2K with cluster services installed.

Each server can be connected to an MSA.
HP SIM can be installed on the D drive (MSA).
The cluster will have a virtual CNAME which can be called IManage, and each server can be called IManage1 or IManage2.
SNMP will be configured on all 3 CNAMES.
SNMP will be received through the virtual name, so it doesnt matter which server is running the cluster.

If a server failure occurs, the other server will take over the MSA and run HP SIM from the second server.
Failover takes less than 5 minutes.

[OlivierV]

Hi.

This is our solution :

We have set up a second SIM server used for testing. All agents send trap to the production and the test servers but the test server only knows himself, none of the other servers are in the local DB.
Each week, we create a host file from the production server and a task on the test one download this file. In case the production server is crashed, we import the host file into the test server. It wil receive traps and untill the production server is back on we will use the second one.
May be not very sophisticated but we had problem with SQL replication and this solution seems easier.