Bladewatch.com

I was talking with Chris today to see how he’s getting on, there have been some more job cuts announced around the city and naturally we’re all getting a bit nervous.

Anyway, he was telling me about an issue that had arisen which made me put together on my post-it note some bullet points about clusters. The meeting had involved an architect recommending everything be clustered for high availability, a rather nervous application guy going so this cluster thing…. and Chris thinking of the issues involved in supporting it all.

Clustering can be a very effective platform for business, deployed in the right way it can form a high availability solution and provide flexibility for planned maintenance or upgrades. That you can simply fail resources between the cluster nodes as the business or infrastructure teams demand it, can be an invaluable aspect of the solution.

However consider the following:

• Is the application cluster aware?
• Does the data need to follow the application or is it realtime/static data – to fail it over do I need to just restart some services or do we need the whole drive to be presented?
• What’s the failover time now and what will it be as you scale it up in terms of data replication?
• What level of access do the application teams need to the cluster and why
• What comfort level operationally and technically do you have with clusters
• How will the monitoring differ and cope with the concept of a cluster?
• What level of backup do we need for the cluster? Do we backup the cluster nodes or just the data

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.