Hi folks,
I've been trying to get back in touch with what's happening in the storage space as the relevant hardware isn't something I'm knowledgeable on and in light of the fact that a year or so ago when I was last reading a great deal of content from Jose Barreto's blog (here), I was under the impression that when costs permitted wider adoption, RDMA would become the best thing since sliced bread.
Having a brief read of some sites including the TechNet community today (I found this post to be an excellent overview for a luddite like me to use as catch-up material), it "feels" like RDMA hasn't really amounted to much, and furthermore, that vendors once committed to some flavour of it are now moving away (see Intel's converged adapter summary here purely as an example of iWarp appearing to be "abandoned").
I could probably keep reading older and current material until the cows come home, but rather than do that, I'm going to take the easy way out and ask a few things like: What has happened to RDMA? What has succeeded it - if anything? Are people just working with vanilla 10GB FCoE and the like?
So much was/is touted about it in Server 2012 that I simply assumed this was something everyone would be pursuing as part of adopting SMB 3.0.
To give this some context, the place I'm at currently is small to medium sized at best and now that we've gotten rid of VMware View, we don't really generate a lot of IOPS - even the spikes aren't that large (< 1,200 IOPS). We make use of a pair of NetApp FAS3240's which in having only one SAS shelf and one SATA don't offer stellar performance but there's nothing I can do about that for the time being.
What I was hoping is that in a year or so when we look to replace our current blades which don't support RDMA that we look at a solution that does, but frankly, today's reading has left me rather uninspired and wondering if I'm thinking along the wrong track?
What are your thoughts/opinions? Do you know anyone that has committed to RDMA and what was/is their opinion?
Cheers,
Lain