http://searchstorage.techtarget.com.au/articles/35033-How-to-buy-clustered-NAS-for-a-virtualized-environment

Cheers
gs

http://searchstorage.techtarget.com.au/articles/35033-How-to-buy-clustered-NAS-for-a-virtualized-environment

Cheers
gs

Glad you liked the post.

Good note about a follow-up post, if you look at some of the links including presentations, there are examples of various block, file and other clustered storage solutions, perhaps not in the detail you would like, however that could be addressed in the future.

Your comment about clustering and performance is interesting to me as I hear it regularly from vendors of server and storage clusters. The common line is that performance scales linearly with no degradation when adding nodes while maintaining data integrity (cough cough ).

Now and then I hear from vendors who will say the performance scales on a near linear basis, with little to no perceived overhead while scaling. Want to take a guess which one Im more inclined to believe of the cuff?

There is the other common myth that adding more controllers, more ports, faster ports, faster disks, more cache results in better performance which may or may not be true as it’s how all of those pieces work together that makes the differences.

In the cluster game, some want you to believe that simply adding more controllers will result in more performance which may be true, however what about the overhead of data integrity in the cluster? Some do better than others, some just don’t want to talk about it, and others want to talk around it.

With the current SSD/flash craze, it’s interesting to watch vendors all of a sudden talk about IOPS however few are talking about latency, even with clusters, hmmm…

Again thanks for the comments.

Cheers
gs

Glad you liked the post.

Good note about a follow-up post, if you look at some of the links including presentations, there are examples of various block, file and other clustered storage solutions, perhaps not in the detail you would like, however that could be addressed in the future.

Your comment about clustering and performance is interesting to me as I hear it regularly from vendors of server and storage clusters. The common line is that performance scales linearly with no degradation when adding nodes while maintaining data integrity (cough cough ).

Now and then I hear from vendors who will say the performance scales on a near linear basis, with little to no perceived overhead while scaling. Want to take a guess which one Im more inclined to believe of the cuff?

There is the other common myth that adding more controllers, more ports, faster ports, faster disks, more cache results in better performance which may or may not be true as it’s how all of those pieces work together that makes the differences.

In the cluster game, some want you to believe that simply adding more controllers will result in more performance which may be true, however what about the overhead of data integrity in the cluster? Some do better than others, some just don’t want to talk about it, and others want to talk around it.

With the current SSD/flash craze, it’s interesting to watch vendors all of a sudden talk about IOPS however few are talking about latency, even with clusters, hmmm…

Again thanks for the comments.

Cheers
gs

This is great stuff though.

This is great stuff though.