So IBM offers SSD's to their servers?
With all this EMC Lightning kerfuffle and Dell, HDS and NetApp joining in – with HP pointedly saying "No comment" like we really believe it's not playing too – there's a Big Blue elephant we're not hearing or seeing. Don't worry, IBM definitely has skin in this game – and it's called eXFlash. This is solid state storage local …
Why does IBM need hot swapping for SSD, does IBM's solution lead to more SSD failures?
The throughput is easily outgunned by the 410,000 IOPS and the 2,800MB/s read and write of a 1.6TB SSD PCIe.
Looks to me if you want to have a decent server you, er, take your earthing kit and screwdriver and build it yourself.
Actually IBM has been selling the full suite of PCIe based solid state since before the announced the EXFlash.
While the PCIe are faster, they do have the downside of not being hot - swap and only able to use software RAID to protect data unlike the EXFlash which sits behind a hardware RAID card allowing for a lot more flexibility and aggregation if required.
But hey, you seem to know everything already :)
The IBM RAID cards have been supporting SSD devices for a long time now, this is nothing new.
Regarding TRIM, that is a function where the SSD is relying on the O/S to govern recycling of deleted data cells. In the SSD's in use in the servers, the SSD's have an internal garbage collection mechanism that takes care of this function.
As I mentioned in my earlier post, putting the SSD's behind a RAID card ( an HBA is supported as well) gives you flexibility on hardware assisted data protection as well as drive aggregation for capacity increases.
Biting the hand that feeds IT © 1998–2020