It's a ratchet: Old storage guard face incoming tech squeeze Squeezed between the giant pressures of hyper-convergence, the public cloud, object storage and software-defined/cheap commodity hardware, the old guard storage suppliers face shrinkage into shadows of their former dominant selves. Ten years ago and before that Dell, EMC, HDS, HPE, IBM and NetApp represented the bulk of on- …
Do storage and processor upgrade cycles align? AFAIK they don't. Moore's law is slowing down but is there a storage Moore's law to compensate?
Secondly how fast do those flash arrays wear out? Yes the spinning rust can crash, which is why you stripe the data.
I get that saving power, volume and area are important but how does that compare to overall system reliability of flash Vs conventional
There was a reason people started splitting systems into big processor arrays and big storage arrays. It still seems pretty valid to me.
Neither VMware's vSAN or EMC's ScaleIO are mentioned in this article. Does El Reg really have that big of a beef with Software Defined Storage or is it just bias against Virtzilla and EMC? vSAN is the fastest growing hyperconverged/SDS product out of all competitors yet doesn't even rate a mention in the article. I find that rather strange.
Transparency: I'm a Senior Systems Engineer in SDS for VMware.
LOL. I love the analysis that says the newbie storage hybrid guys "have morphed into all-flash array vendors" meaning that were not AFA but magically are now?? Yet, HPE, HDS (growing?), EMC (excluding the now shunned XtremIO) are seen as not able to "morph" into AFAs. So, the engineers are better at storage-morphing outside of those companies. Puhleeze.
Let's all realize that by and large (save some hail-mary NVMe products that don't exist yet), these things are all storage systems. With data reduction getting more pervasive the distinction between "General Purpose" and AFA is only useful for selling analyst papers. Do you store, protect and manage my data well, or not.
Hyperconverged is a bit more interesting... and will still be a wild ride as vendors grow, contract, merge and die. I don't see much mention of a VMware that's hungry to not lose customers and relevancy in that space. (Or Microsoft, with its plans to use Azure connectivity to trump others.)
"Fun" times.
I find it silly that AFA is even considered a market or a "strength". The storage array venodors spend 40 years taming spinning rust which takes a lot of experience. E.g. How to lay down data in the most efficient, reliable way cobsidering the physical limitations of a disk drive.
Flash is exciting, because startups have to have one less skill (taming the spinning rust) so they can spend more time on developping front-end gadgetry - while having seemingly endless IOPS in the back.
That's great news for customers with poor solution architecture, because their AFA can mop up the mess the previous Solution architect made...
While the startups grow and try to impress customers with their All Flash from the ground up pedigree, there's still a lot of knowledge the old storage guard can bring to the table. Especially when the time comes when the array will drive the 'the next cool storage medium" very hard. Startups donot have that experience.
And please dont get all technical on me, I know that there's other factors besides IOPS...
The storage industry (old and new guard) is full of people trying to sell snakeoil to confused customers, but at the end of the day they sell buckets for data...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2026
