As EMC prepares for its 14 November launch of its all-flash XtremIO box, Dell has put out a analyst paper explaining that on price – which is what counts in the expensive all-flash storage world – it is the bossdog of flash arrays. Michael Dell’s privately owned biz points to an ESG technical lab validation (PDF) that compares …
Apples to Apples
Fill a Dell PowerEdge with SSDs (that's all an SC8000 is anyway, from a hardware perspective) and it'll be cheaper still. If I wrote ESG a check, I could get them to testify to John Wilkes Booth's innocence. After years in this business, I still fail to understand the credibility of checkbook research.
Both the Dell Compellent all-flash array and Pure Storage use the same Dell R720 servers as the "head"/controller unit and Xyratex 2U 24-bay SAS JBODs as the expansion enclosures. NetApp actually uses the same exact drive trays for 2.5 HDD's in the FAS6000. Not coincidentally, all of the above have virtually denitical performance (300K or so IOps), which is rather weak compared to the purpose-built machines like Violin, Nimbus Data, and TMS/IBM. But the cost is ultimately in the flash, not the chassis/controllers...and Dell's only advantage there may be a willingness to make less margin given its weaker software features (a typical strategy for Dell).
data integrity is everything in storage
'Anonymous Coward' I often read you posts but never challenged you as you appear to know what you're talking about, however, you are so off the mark on this one you have no credibility with me! Dell Compellent do not use Xyratex 12 or 24 bay enclosures and have not for a considerable amount of time. As for the use of R720, that's like giving your good self and Gordon Ramsey an identical list of ingredients, asking you to cook a meal and expecting both to turn out the same... the storage center software is Compellent "secret sauce" and how they can do more with less. As for your final point on Dell's willingness to make less margin/profit how would you know this, do you have inside knowledge on what margins they're retaining or are you saying that Dell are providing better value for money to its customer and that other vendors are making excessive profits? I don't expect you to answer that last question and will provide an answer for you, if you read the article Dell are mixing SLC and MLC drives in 2 tiers of SSD and using the disk characteristics of each tier to drive down the cost while maximising read and write performance through those aforementioned characteristics. I would say that this is a leap forward... but Compellent have been tiering data for an age and have just enhanced it so that it can be done in real time. Evolution rather than Revolution!
Ever wonder what an EMC system runs on? It's nothing fancy, in fact its standard x86 server architecture for many of their products. If you look closely at an EMC Data Domain or a RecoverPoint Appliance, they look very similar to the Dell PowerEdge 520.
That depends on the EMC system
-Disclosure NetApp Employee-
Some of the EMC kit that is not designed for 5 nines of uptime (e.g. Isilon, DataDomain) is based on stuff like super-micro servers, which is mostly pretty good, but there is a reason the hardware is cheap (and it's not massive economies of scale). Unlike servers designed for virtualisation which like cattle should be shot when they get sick, a device designed to maintain the health of critical data needs to be built out of stronger and more resilient stuff.
The kit with decent hardware engineering does have intel bits in it, but are designed to much higher standards than your usual commodity stuff. You'd be surprised how much engineering goes into making something you can really rely on. When you expect to ship tens of thousands of units, even very low annual failure rates can turn into lots of very very unhappy campers. Good companies with a decent sales volume take a lot of care to make sure they dont lose their customers data. Many startups on the other hand take advantage of that risk aversion and push products out the door with impressive stats, but less impressive resiliency as fast as they can before they run out of VC funding. If they have a 1% annual failure rate, then that translates to maybe 10 unhappy customers, which is a manageable risk.
In short, dont fall for the old "its all the same hardware" schtick that is being pushed by a number of relatively uninformed folken, and keep in mind that there are a bunch of very good reasons why vendors like EMC, and NetApp, and HDS have QA departments that dwarf the entire development teams of the startups.
Re: That depends on the EMC system
John that's an interesting take on things, but is perhaps slanted in a certain way. IBM have been using servers in their flagship range since the late 90' (VSS/ESS/DS8k), so using "off the shelf" components is not a new concept or one that you should disregard as inferior. If you look at IT in general virtualisation is a concept that not only allows higher utilisation of components but also has the ability to provide resilience, probably best understood by the wider IT community through features such as Vmotion/Live Migration and all the other functionality built from this in server world. What I would agree with you on is the QA part and this is normally signified with the five 9's label. Having seen the work that's required to allow a vendor to take the five 9's to market in the form of engineering time to satisfy the legal teams, although never a guarantee, allows customers the confidence that solutions are enterprise class.
- Fee fie Firefox: Mozilla's lawyers probe Dell over browser install charge
- 20 Freescale staff on vanished Malaysia Airlines flight MH370
- Neil Young touts MP3 player that's no Piece of Crap
- Review Distro diaspora: Four flavours of Ubuntu unpacked
- Sysadmins and devs: Do these job descriptions make any sense?