Yeah that title got your attention right… For now it is just me writing about it and nothing has been announced or promised. At VMworld I believe it was Intel who demonstrated the possibilities in this space, an All Flash Virtual SAN. A couple of weeks back during my holiday someone pointed me to a couple of articles which were around SSD endurance. Typically these types of articles deal with the upper-end of the spectrum and as such are irrelevant to most of us, and some of the articles I have read in the past around endurance were disappointing to be honest.
TechReport.com however decided to look at consumer grade SSDs. We are talking about SSDs like the Intel 335, Samsung 840 series, Kingston Hyper-X and the Corsair Neutron. All of the SSDs used had a capacity of around 250GB and are priced anywhere between $175 and $275. Now if you look at the guarantees given in terms of endurance, we are talking about anything ranging from “20GB of writes per day for the length of its three-year warranty” for the Intel (22TB in total) to three-year and 192TB in total for the Kingston, and anything in between for the other SSDs.
Tech Report had set their first checkpoint at 22TB. After running through a series of tests, which are described in the article, they compare the results between the various SSDs after 22TB writes. Great to see that all SSDs did what they are supposed to do and promised. All of them passed the 22TB mark without any issues. They had another checkpoint at the 200TB mark, which showed the first signs of weakness. As expected the lower end SSDs dropped out first. The next checkpoint was set at the 300TB mark, they also added an unpowered retention test to see how well they retain data when unplugged. So far impressive results, and a blog series I will follow with interest. The articles clearly show that from an endurance perspective the SSDs perform a lot better than most had assumed in the past years. It is fair to say that the consumer grade SSDs are up to the challenge.
Considering the low price points of these flash devices, I can see how an All Flash Virtual SAN solution would be possible leveraging these consumer grade SSDs as the capacity tier (reads) and using enterprise grade SSDs to provide write performance (write buffer). Hopefully we will start to see the capacity increase even further of these types of devices, today some of them go up to 500GB others up to 800GB, wouldn’t it be nice to have a 1TB (or more) version?
Anyway, I am excited and definitely planning on running some test with an all flash Virtual SAN solution in the future… What about you?
** 500TB blog update! **
** 600TB blog update! **
** 1PB blog update! **
** 2PB blog update **
** Conclusion **
I’d be curious how the eMLC based drives perform given their classification as “enterprise” level. Hopefully they can do an analysis with those as well.
Kaminario is offering an all flash array with 800GB enterprise grade MLC SSD’s today with a 7 year warranty guaranteed. Today we support linked clones and in April we’ll be supporting full clones with volume based deduplication along with very attractive effective cost per GB pricing. Our largest installation to date is an Israeli Bank with 4,000 linked clones where we beat out the competition in a bake-off. Taneja Group Lab Validation white paper can be downloaded from our website at: http://kaminario.com/blog/press-releases/taneja-group-dubs-kaminario-a-force-to-be-reckoned-with-after-testing-newest-generation-k2-all-flash-storage-array/ It shows that the Kaminario K2 is about 50% faster than EMC XtremIO in starting up 1,000 VMware seats. The K2 took 3 minutes and 35 seconds and EMC took 7 minutes!
Definitely an emerging trend. Consumer discs are quite different than enterprise class, and tend toward SATA, rather than SAS as well, but the real issue has historically been one of quality control. If you take a look at what has happened to OCZ recently, and how they retooled their entire line because of QC issues, it does appear that they may not be dead in the water after all… Meanwhile, the SSD market continues to get more interesting. SanDisc, Intel and others coming along with new products to compete in both size and consistency well beyond their previous abilities.
On the enterprise side, capacities seem to be increasing, prices are dropping (albeit slowly) and durations are increasing.
Ever heard about ‘PernixData’? A startup which’s exactly into that already!
Is that a serious question / comment? Not sure what it has to do with this topic? (Yes I know about them, I wrote various articles…
Pernix based on their description seems to be a “Virtualized global cache”, not a virtual SAN.
I see All Flash SANs available, and then VMware’s vSAN available — then virtual SAN solutions that live on typical magnetic disk, or typical RAID arrays.
But so far, no Virtual SAN solution suitable for primary mission critical storage, specifically designed for all-flash — to manage, safelty, and efficiently place and retrieve all sectors from a large set of SSD-only drives; with the expected replication features, etc.
Seems like VMware could be close with vSAN.
Thanks Duncan for pointing out their article about the test of those SSDs. Very informative. Their latest update is for 500TB. Which is 140GB of writes per day for 10 years. They use those SSD plugged-in to a two single Windows systems.
In virtual environments under vSphere, if we plan for “full flash VSAN” cluster, how much data do we write daily? We could possibly know if we take traditional SAN’s data, from existing cluster’s (non vsan) daily writes. I wonder if there is a VMware utility allowing to monitor the daily disk writes. Right now I’m not aware of any. Perhaps via PowerCLI?
Another great article and i’ll mirror vladan in saying thank you as well for pointing out the test articles and I too will be watching this closely.
It’s apparent that the interest in moving the storage stack closer or into the Virtual Layer is going to reap tremendous benefits around SDS, especially when we get into Application/VM storage policies, vVOls..etc.
TIER “0” has really been about use case, right? I mean, in reality all SSD arrays etc are “purpose built” solutions to address a specific use case such as VDI, Analytics, Highly Transaction DB’s..etc…but looking at how manufacturers are looking at longevity and MTBF..and how we see those things moving to the consumer market change that outlook a bit. I mean, in reality, I think we’ll still need high cap spindles for unstructured data, archive..etc, but I believe at some point, SSD will not just be purpose built..right? For example..throw all application workloads on SSD, and all unstructured data on Spindle…or even SSD cache accelerated spindle.
Very exciting times, and I am very excited about VSAN. I think VMware has a compelling solution which will get better over time. It’s also exciting in the majority of spaces I work, the School districts, medium size businesses, etc where now they will have the opportunity to leverage “enterprise” technology where it’s much more affordable.
IMHO – Without an Smart Monitor Daemon (NOT crappy ESX 5.1 introduced S.M.A.R.T Feature) where provides intelligent SSD Lifetime and Wear Leaving Monitoring on ESX Side are not realizable . We allready realized and All Flash SAN with Inelligent Smart Monitoring for Customer with 15 TB of Samsung SSD 840 SSDs.
IBM DS8870 is available as all-flash enterprise disk system also, and that can be hundreds of flash drives.
Euhm yes?
Physical desktops today do not come with enterprise grade drives. All the major manufactures today put in what they get for next to nothing to squeeze an insignificant profit out of desktops. Why then does a VDI solution need to have eMLC or SLC? Why can’t we have an inexpensive SSD array with enough redundancy built into it to make it feasible for anyone to afford? I’d be interested in an all flash comsumer grade vSAN. I don’t mind putting in a extra spare drives on top of the raid that cost a fraction of an eMLC or SLC drive. I want my cake and want to eat it too! 🙂
Duncan, thx for the post. cMLC is moving fast, and software only stacks (VSAN being one) will ride this trend fast. A lot of people miss that the software stacks presume unreliable nodes and media and use copy distribution as the recovery mechanism.
IMO, where the architectures you are thinking (and there are others for other use cases – being biased – disclosure, EMCer here – like ScaleIO on Linux kernels) will come to play more is as they add some form of soft error check – but that’s a matter of short time.
Interesting, and very disruptive! We live in interesting times!
Duncan,
Will VSAN allow for disk groups of only local SSDs and no spinning disk?
Yes and No.
Yes: you can do this by tagging some SSDs as HDD using esxcli
No: It is not officially supported today, but it is an area of interest for VMware for sure