1

We're looking to improve our Disater revoery and fail over capabilies or our Web server and Webpshere applicaiton server.

We have 2 sites in the UK [HQ and Callcentre] and want to host a DR Webserver and application server at the call center site. So if HQ is out of action Callcenter server can take over.

We were thinking running two VMware ESx boxes and mirroring the config changes between them (how i'm not sure)

An alterntive we thought about is using Websphere Culstering.

These seem like good options?

2 Answers 2

1

You have two challenges :

  • Choose a data replication strategy. It depends on how you store your data (database, raw files, etc.)
  • Choose a requests distribution strategy. It depends on your network infrastructure. You can rely on load balancers, DNS servers, IP routing, etc.

Your choices need to fit your network and application architectures.

0

One option that I have seen implemented (it was on physical servers but could easily be adapted for virtual servers) was to have the production system running at the primary site on some given hardware, with all the storage SAN backed (both system and data volumes). The SAN then manages the replication to another SAN at the DR site at the cluster level in real time.

If the primary server fails, then the copy of the primary server at the DR site can be started on the standby hardware, and it will come back up with the config and data from almost immediately before the failure. The OS will think that something went wrong with the hardware (like a power outage) and recover as it would normally.

The advantage to this setup is the standby is updated in near real time and you don't have to make any changes to the applications to support this sort or fail over. The downside to this setup is the infrastructure overheads required (dual SANs, dual servers) that may or may not be available.

You must log in to answer this question.