Operating System - HP-UX
1748155 Members
3804 Online
108758 Solutions
New Discussion юеВ

Moving to High Availability

 
Colin Workman
Occasional Contributor

Moving to High Availability

Scenario

2 RX6600 servers running a bespoke server app and ORacle 11g database. THe databases are kept in synch with Oracle Replication. failoer is manual process.

If host1 has an issue we can start services on host2 and make it the active server.

Quest
We want to make this a high availability system, possible clustered using Data Guard and Service Guard but we would have to do this with as littel downtime as possible.

Any suggestions as to how a cluster can be made but still allowing the system to operate?
2 REPLIES 2
Rita C Workman
Honored Contributor

Re: Moving to High Availability

Well...DataGuard is having two databases, a primary and a secondary. Service Guard is where you have one database that can be run on either server. So...I'm not exactly sure how you are implying to handle this or which way you are leaning.
You mention that you server is 'bespoke' and I take that as a custom built server.

Now...you can build a Service Guard cluster with the server/nodes up and running. But at a certain point you do need to give the 'disk', or more accurately, the volume groups, over to the cluster to control. So there would be a point of downtime required to deactivate/activate the vg's using vgchange. And, of course, there would be time required to test out that the package properly 'failed' over to the secondary node and all came up appropriately.

So, maybe someone out there as dealt with setting up such an environment. Myself, I've never seen one doing both.

Regards,
Rita
Alzhy
Honored Contributor

Re: Moving to High Availability

I think the question is how to convert his existing environment to a CLustered configuration without incurring DOWNTIME.

My few cents (specially if you will find ServiceGuard prohibitbly expensive):

Since you already it seems have a MANUAL process in place -- why don't you just improve on those manual processes and script the entire process? ServiceGuard really is just a couple os smart scripts that automates "everything". I call this recipe -- "Poor Man's Failover" -- all done via smart scripts -- from plumbing up interfaces on the failover host, checking and importing storage needed and startup of services (apps and DBs) on the failover node.



Hakuna Matata.