DCIG rates top-selling EMC flash array as survey bottom-dweller In its latest All-Flash Array Buyer's Guide, DCIG plonks EMC's market-leading XtremIO array in the bottom group of suppliers, with HP's 3PAR StoreServ 20000 Series ranked best-in-class. The DCIG folks use an 8-step process to score and rank 28 array products from 18 suppliers, surveying each vendor's products and assigning …
Sounds like most of you are naive - how do you think Gartner rates vendors in their magic quadrants? Do you see any customer surveys?
Seriously, all analysts play the same game including Gartner. At least DCIG's selection criteria is transparent so a user reading it knows how to interpret the data and apply it to their own requirements.
I'm certainly not going to defend DCIG but...
DCIG published a report ranking HP's 3PAR first and XtremeIO 11th. DCIG has a history of being enamored with the 3PAR product. So their credibility is to be questioned, right?
The Register then has a problem with EMC's XtremeIO being ranked 11th. Doesn't The Register have the same history of being enamored with EMC and all their products? So their credibility is also to be questioned, right?
Religion aside, the issue here is with the counter argument. At no time does scientific data enter into the discussion. The Register's point is that "How can XtremeIO be bad when it sells so much?". There's no technical validation, no technical questioning, and no logical reasoning to that argument.
The historical data is on the side of DCIG here. In modern Storage Array history no other product has required a full data wipe of its array for a firmware upgrade except XtremeIO. No other flash product declines concurrent drive rebuilds except XtremeIO. All flash products are extremely immature in terms of software offering--and EMC's answer is to bolt on VPLEX & Recover Point which isn't part of XtremeIO and therefore does count against them--like it or not.
XtremeIO's best points are performance and the EMC Sales force. BUT flash negates performance as a criteria because every flash array--with the exception of Pure (Pure is "fast enough")--is going to be faster than most customer's actually need. XtremeIO is left with EMC's superior Sales force as their best feature. I'm pretty sure DCIG wasn't looking at "sales force" when evaluating the product itself. AND EMC Sales does in fact have a very long history of "pulling the wool" over their customer's eyes.
Take a look at the 10 products that ranked ahead of XtremeIO and see if you can't make a scientific argument based on feature-set and availability that they're not better than XtremeIO and publish a thoughtful counter-argument. When I look at the 10 ahead of EMC I see at least 8 that are actually better.
Listen as the EMC sales force collectively loses their minds over this. Just points out that their are a LOT of arrays out there better than XtremeIO when you pull away al the marketing $$ and FUD. XtremeIO is really expensive....and really not that good. Wait until EMC customers are told they will have to forklift their initial investments down the road when EMC no longer sells the 5 and 10Tb Xbricks.and customer wants to add capacity. :-)
What was the point of your post there? The link just tells you that DCIG will evaluate anyone's product--and they did.
Not only that but DCIG gives each vendor their personal review--not the collective--before publishing. If there was anything wrong with what they had to say about XtremeIO then EMC should have caught it. They certainly had enough time and ability to do so.
Isn't it interesting how no one will actually technically challenge DCIG's assertion BUT everyone who doesn't like it is so quick to bash DCIG? Tells you they're on to something here...
And that vendor gets called out in the comments section...
Are people really that brainwashed by EMC that they fail to see DCIG actually has a valid point? XtremeIO doesn't have anything but speed. The more I look at it the more I think it's a fair ranking. Might even be a little generous...
"XtremeIO doesn't have anything but speed"
And we think its DCIG that's brainwashed! Could you expand a little on that? Sure, we all have our favourites, but (hopefully) we have the sense to see beyond the rhetoric. The vast majority can see this article for what it is - a thinly veiled ad for a particular vendors product. All it really does it discredit DCIG further, if that's possible.
DCIG has been on HP's payroll for years. This isn't news. Often their reviews are so flawed, it's comical and downright false.
Some companies like EMC do not allow you to publish benchmarks without prior permission. So they do the next best thing, pull down a virtual edition of the product and try to test it in production in their little so called lab. Often that VM is a major revision behind. Or pull down every whitepaper and see if the answer is there to match their little feature matrix.
On the Symantec front, they use out of the box settings without tuning buffers, etc. You know, NBU basics...
What I wonder is, how many people really take DCIG as serious? Gartner and IDC are viewed much more favorable by CIO's I've known. DCIG is a small time rag leveraged by few vendors.
True but Gartner's reputation is not very high anymore. Most CIO's reference them but very few of the good ones give it any additional weight over DCIG, ESG or Forrester. Basing a buying decision on an analyst firm anymore is the sure sign of a lazy CIO and the vendors play them like a fiddle.
Is it really news that EMC's portfolio is currently pretty weak? You shouldn't need (most likely flawed) surveys to tell you that. VNX is so dated and limited now, XtremeIO is a real slow learner and still way behind the other flash arrays. There isn't a lot to defend about EMC arrays right now. If EMC weren't so good at locking out other vendors, they would really be suffering right now.
Seems EMC has a fairly cosy relationship with DCIG when it suites, just google DCIG site:emc.com.
Same for IDC, Gartner, ESG, Forrester, Silverton etc as do all of the big players and many of the smaller. If you visit the DCIG webpage they have a couple of sections listing sponsored and unsponsored content and the buyers guides are unsponsored.
If you come top in the rankings then you'll buy the report to distribute to Customers and publish on your website. So yes the reports for sale but only after the rankings are completed and I'm certain any vendor with a top ranking in one of these would do the same and be crowing from the rooftops regardless of the validity of the tick box process.
Not suggesting this is anywhere near perfect, but all those rated had an opportunity to supply information and correct mistakes before going to press. So if there's something technically inaccurate in DCIG's content then it's in the vendors interest to wade in early and correct. But then again if there's not much to correct, well then your only option is to cry foul.
Some very interesting views bouncing around in here, EMC portfolio is weak?...... #1 across virtually every data platform software and hardware?
All XtremIO has is speed, thats like what have the romans ever done for us?....Inline de-dupe, compression, 8% Raid overhead, no write amplification VSI and ESI to integrate with VMware, Oracle and SAP...... yeah......it is pretty basic!!!!!! oh and quick too!
If you know anything about flash, then you know that trying to run it through a platform designed as a "disk" array is like putting propellers on a 747.
Ah, so you're still in the outdated mindset that flash is just for stuff that needs super high performance so it's basically a drag car and you don't need bells and whistles. Wake up, flash is great for just about every workload these days and the market for general purpose arrays full of flash and benefiting from the performance is way bigger than the market for super fast storage without any real functionality! XtremeIO is lagging behind - Disruptive firmware, 2M IOPS if you take 8 bricks (that isn't even that special when 'disk arrays' full of flash can do 3.2M), poor scalability, has it even got replication yet (if it has, it's not mature). Those 747's with propellers can go just as fast as those jet ones, and carry a load more passengers these days. Stop throwing outdated FUD around and open your eyes.
Yes you are quite correct, it is a weak and basic product. EMC should be embarrassed to be associated with this poor implementation of SSD in a storage system.
They may be able to improve it over the next few years but in the meantime it should be kept out of production environments.
Terrible report, however, they do have the XtremIO in the correct place. It is a very poor implementation of an SSD array and as usual, customers only buy the product because of the EMC badge on the front. Chris, you need to remove your "blue" tinted glasses and speak to real customers on real world experiences with XtremIO. Having an X in the name and an EMC badge on the front does not make it good product.
Well, my EMC bashing friends, it seems that Gartner, IDC, Wikibon and many more have got it all wrong on XtremIO and its status in the AFA marketplace and this should be fixed using cold hard facts!
So I hope that you can help me by explaining why it is not as good as ANother technology?
Inline de-dupe, compression - Just about everyone has these and where haven't they make up the efficiencies in other ways.
8% Raid overhead - Everyone has there own special flavour of raid, but customers don't buy systems based on backend features unless they enable other more useful front end features.
no write amplification - This is a given for any Flash array
VSI and ESI to integrate with VMware, Oracle and SAP - Again everyone's got them and where they haven't third party backup solution can provide even broader support.
Qos = No
VVols = No
Replication = No not without external appliances or host based software
Mixed clusters = No
Granular upgrades = No
Competitive cost per GB = No
Online everything = Questionable
Most basic enterprise features are still missing unless you bolt something else into the data path, so where is the secret sauce, snapshots.....oh yeah everyone has those also ?
Gartner, IDC and Wikibon et al are only interested in market share and feet on the street, not the technology or even it's capabilities. The swing kit EMC services had to purchase to manage the data destructive upgrades must have given factory revenues and subsequent market share a nice boost not to mention the usual Stockholm syndrome brigade.
More importantly, the TCO of XtremeIO is MUCH higher than the immediate competition. Adding capacity is insanely expensive given I have to add a whole brick. 2-3 yrs down the road when I cannot add a "like" brick, it's a forklift (again). Data efficiency rates are not what the EMC salesforce promises (many customers backing this up, EMC is giving away a lot of bricks!). Plus the switches, cabling, UPS's needed to expand....ugh. I had a customer recently rip out Vplex & RecoverPoint for his VM environment because is simply did not work. Too bad he had to overlay all that on top of XtremeIO just to get replication.
Yeah, it's fast, but I have not located one customer that needs 2 million IOPS at <1ms. All the other "stuff" has proven more valuable. All the arrays available give breakneck speed!
Yes they do have it wrong. Gartner, for example, have had to publicly appologies several times just recently for incorrect reporting.
http://www.storagereview.com/gartner_issues_apology_for_pure_storage_s_wildly_inaccurate_sales_data
All the facts are in this discussion, you just need to read the comments.
If it wasn't for the EMC marketing machine and funding, Xtremio would not be a player in the AFA space. If it was a product for the future why did EMC buy DSSD?
Remember the "Emperors clothing" fairy tale, this is EMC to a tee, if they say it is so then it must be, no questions asked.
Shame on any truly technical person that allow themselves to be caught in the headlights of these juggernaut organizations.
Just to be clear, from a technical perspective the Inline de-dupe and compression should be table stakes for most AFA vendors, but how many of them actually guarantee that those services are always delivered that way.......very few. How many vendors have re-written RAID so that you get rid of duplicate writes and don't generate write amplification on the back-end array.......very few. If you don't get rid of traditional RAID how can you deliver the full IOPS capability of your entire array to a very small physical environment.....you can't! Everyone has replication capability for their AFA's.......most do and it is less than enterprise class. Funny that someone brought DSSD into the discussion........DSSD is something different again.......not meant as an AFA with full data services, looks like some homework is required!
This thread just got my attention again. How funny.
If XtremIO has reduced the write amplification on the backend, why is it the only product in the market still dependent on eMLC?
Upgrading from 1 x-brick to 2 is still an offline process.
Want to crash the system? Pull the power cables off one of the x-bricks. when it reboots metadata will rebuild and slow down your array for an hour or so. if something else goes wrong you'll actually crash the system and lose data.
that system was never designed to work. it was designed to be sold by emc and nothing else.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
