If you enjoyed this article, please consider sharing it!
14 Responses to Enterprise Computing: Is the Solid State Drive Hype Over?
Leave a Reply
You must be logged in to post a comment.
- Use Symantec and know your sensitive data is protected with industry-leading backup & recovery software.
Experience Symantec Backup Software
Chris,
you are completely right!
Vendors are adding SSDs to actual arrays not designed to support that technology.
We need to wait next generation arrays and hope better implementations.
BTW, (as you know i’m little bit biased) i think that Compellent’s approach is the best to achieve a good balancing of performance and usability of SSDs.
ciao,
Enrico
Having spoken to EMC recently they are pushing SSD and SATA drives very heavily for in array tiering. It looks like they are trying to prepare customers for FAST2. That’s what they said, not FAST but FAST2. FAST2 will provide block level migration on the fly compared to drive level migrations.
With FAST2 it then starts to make sense using SSD’s as the heaviest used blocks can very quickly start writing to very fastest drives.
If you are thinking of buying SSD’s then negotiate hard as it appears EMC committed to buying an minimum number of SSD’s but overestimated the number that would be sold so they need to move the surplus quickly.
Right on Chris. SSD is a major change. It takes time for such things to transpire.
Seagate has been getting a lot of comments about being ‘late’ to this market. My take: Seagate’s entry this quarter seems to right about on time: http://tinyurl.com/yjm52y2
Pete
I read your article on Seagate’s SSD release earlier. When can we expect to see them in Enterprise products and have you any performance/specs available to read?
Chris
Paul
I guess we need to see exactly whether FAST2 lives up to the hype. I think even with block-based tiering, all that’s achieved is workload balancing at a more granular level. FAST2 needs to offer policy-based migration. It will be interesting to see how it works.
Chris
Enrico,
I think for Compellent the next piece of evolution is to reduce the re-org interval to less than a day and make things more policy based. Hoping to get my hands on some hardware soon to write a proper review.
Chris
We started shipping in September. As you know, it takes a while to get from there to our customers shipping revenue systems. No dates yet. Don’t have specs to share today, but stay tuned. It won’t be too long.
[...] SSD) non sono molto buone e hano scatenato tutta una serie di commenti nei blog (esempi: qui e qui ) sull’adozione della tecnologia SSD. EMC è il cliente più importante per STEC ed è stata [...]
Competition always enters new markets where money can be made. Unfortunately, however, not all competitors products are equal. There is a lot of room for marketing nonsense around SSD reliability and performance. Reality is that flash is a media, and all media has warts. How you manage those warts is key in managing drive useful life. If you do not own the fundamental technology (the SSD controller in this case) you will never be able to make a drive that performs consistently and reliably in the Enterprise. This controller technology is in very few hands, and as such, will become the differentiator among products and competitors. Only two companies today are building drives using their own controller IP. The industry will only see the value of this IP four or five years after the system is deployed and once the media starts to degrade significantly. Eventually SSD’s will be mandatory in most Enterprise appliances simply due to the performance gains they bring. If an SSD were at cost parity with HDD’s today, the argument would already be over. Until then, it will be open season on outlandish claims and specmanship.
Chris,
Your comment that “many customers are struggling to identify use cases for solid state drives and have no easy way to measure and justify the potential performance improvements – other than to install the drives and see what happens.” This is not true. Granted, I believe customers are struggling with how to quantify and measure the benefits before they purchase and install these drives, but the methods do exist.
As a performance engineer at EMC, we have been taking a scientific approach to helping the customers to understand what volumes are showing up as good candidates from array statistics, correlating those volumes with application/database data to make sure the match will produce an application improvement. We even go so far as to model the performance and showing lower utilization on the spinning disk drives. This does not mean that the process is somewhat involved. It does require some cooperation and effort on the part of the customer. It also requires that customers examine their application/database data to do the correlation part. But the process works and it has been demonstrated at several customer sites to improve performance, lower drive utilization and drive overall disk array solution costs oveall.
Chris,
My last portion of my last sentence in my previous comment should have read, “lower disk utilization and lower overall costs for the disk array.”
[...] SSDs manufacturer) aren’t very good and raised comments by many bloggers (i.e.: here and here) about adoption of SSD technology. EMC is the most important customer for STEC and was the first [...]
[...] Oscar … Il Divo (MPI Media Group through Music Box) Aldo Signoretti and Vittorio Sodano …The Storage Architect Blog Archive Enterprise Computing …Storage and Virtualisation … Name (required) Mail (will not be published) (required) Website. [...]
[...] hard drive isn’t going anywhere overnight (witness the crash of STEC shares and EMC’s EFD stockpile), however storage technology is changing as we move [...]