« 0.044: not dead yet, I guess | Main | 0.046: spc-1 results for symmetrix dmx »

October 23, 2007

0.045: pushing daisies, ds8000 style

Well, it seems I spoke too soon. And it turns out I was wrong. Actually, I was right first...THEN I was mistaken. I'm sure some of you were fooled as well.

I'm not a doctor or coroner, but let me be the first to declare that despite the best attempts of IBM PR department and their paid cadre of industry analysts, it turns out that the rumours of the rebirth of the DS8000 were in fact premature.

The DS8000 officially passed away today, Tuesday, October 23, 2007.

Dead. Finished. Kaput. Expired. Done. Departed. Kicked the bucket. All but buried.

Go ahead - Google today's IBM announcements and coverage and see for yourself.

Now there are still some press and analysts out there who haven't seen the obituary. There are a few analysts-for-hire who are hungry enough for income to sing the praises of the DS8000 and to publicly declare that assertions of the DS8000's demise (like my prior "not dead yet" post) are nothing but FUD. But I think its becoming clear that these analysts aren't doing customers (or the industry they serve) any benefit by trying to help IBM hide the fact that they haven't been investing in the DS8000 for quite some time now.

I'm reminded of a line from the movie The Sixth Sense:

I see dead storage products. They just don't know they're dead yet.
 

So thankfully, as in the new sit-com "Pushing Daisies," someone temporarily revived the deceased, just long enough to identify the killer.

your honor, the evidence will show beyond any reasonable doubt...

...that the dearly departed is in fact dead, and that it died a quiet, peaceful, almost unnoticed death, brought on by a total lack of investment by IBM attention from its parents.

Now granted, minimizing their investment may have been prudent and wise, given the fact that most DS8300's are to this day most frequently bundled as "free" storage with IBM mainframe and server sales. But in a day when OPEX and energy efficiency are of increasing importance to IT, even free storage just isn't good enough anymore.

After today's IBM announcements, customers seeking enterprise-class storage should now know what many of us have suspected for a couple of years - that IBM is getting out of the enterprise-class storage business, as evidenced by the following observations...

...after introducing the DS8000 over 3 years ago, IBM today announced:
  • No new hardware, in fact - no real architectural enhancements for more than 3 years - even though the P6-based servers have been available for over a year. Still  the same old two-node cluster, complete with all the split-brained availability and reliability issues that have plagued the product for years. Meanwhile Hitachi has done 1 new spin (USP-V), and EMC has done two on Symmetrix (DMX-3 and DMX-4)
  • No thin provisioning, or even a commitment to thin provisioning. Just crickets. (Celerra support since Jan 2006, EMC announced availability Q1’08 for both DMX-4 and the prior generation DMX-3, and for CLARiiON later in 2008 - and the USP-V has thin provisioning too, although their implementation doesn't yet support external storage or remote replication, nor prior generations of the USP)
  • No SATA drives (only DMX-4 supports native SATA-II drives, since Aug’07)
  • No drives larger than 500GB (DMX-3&4 support 500GB today, DMX-4 support for 750GB is announced for this quarter and the 1TB is on the way)
  • No RAID 6 (DMX-3 has supported multi-dimensional RAID since Q1’07, DMX-4 since Aug'07, USP-V also supports)
  • No 4Gb back-end (USP-V since May '07, DMX-4 since Aug’07)
  • Still only 1024 maximum disk drives (DMX-3 & 4 support up to 2400 drives, USP-V supports 1152)
  • Space-saving Snaps for FlashCopy, almost 3 years after EMC delivered TimeFinder/Snap on Symmetrix DMX (since 5671 code, Hitachi slipped this out sometime in the last 12 months or so)
  • Dynamic volume expansion (Symmetrix has supported on-line meta expansion for at least a decade)
  • Internal Striping for performance (Symmetrix has offered striped metas for all storage tiers for more than a decade as well , while USP-V supports internal striping only with Thin Provisioning)
  • Unified Management (Control Center’s been the premier enterprise-wide SRM platform since 1999, with support for third party storage since 2002. Meanwhile, EMC's SMI-S support remains second to none, recognized by third party developers as the most complete and easy-to-use implementation in the industry.
  • Still only two HARD LPARs (partitions) – DMX has 8 SOFT/Dynamic cache partitions and QoS priority queues, and even IBM’s mid-tier products support more than 2 storage partitions (in this same announcement)
  • New pre-fetch algorithms for sequential I/O – ho hum, Symm’s been doing this for years as well
  • No native Ethernet replication or iSCSI support (Symmetrix has had since 2002)

And absolutely no mention of thin provisioning. Now seriously, how can ANY storage company be serious about ANY segment of the storage market if they can’t commit to thin provisioning on the platforms they are shipping in this day and age?

So with all this, can there be any doubt that IBM has given up on the enterprise market and (more importantly) the needs and requirements of the customers in this space?

And I have to ask: Why isn't IBM being called on this by the press and analyst community? I'll bet that if EMC were to have put up such a lame announcement, these same press and analysts would be beating the heck out of EMC, and publicly. Sure, there have been a few slaps on the wrists about the lack of any thin provisioning commitment, but one analyst [name redacted] actually took the stance on a call today that IBM customers can now safely invest in the DS8000 again, as a result of this announcement (he must not have gotten the memo).

I see extinct analysts. They don't know nobody listens anymore, either. 

what this means to you

OK, if you're an IBM customer/prospect and they're pushing the DS8000 Turbo at you, after today's announcements, you have really got to be asking yourself: how much longer is IBM going to keep pushing me to use this same old, outdated, expensive storage technology, while their competition innovates, scales and out-performs them in every dimension? Do I really want to use technology that arguably should already be end-of-life already?

Or to put it another way...

Why are they Pushing Daisies at me?

-----

P.S. Now we know why Fellow Bloggers TonyP and BarryW could take their vacations this week - There's Nothing Happening!


TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d834c659f269e200e54f1127b58834

Listed below are links to weblogs that reference 0.045: pushing daisies, ds8000 style:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Barry Whyte

So saying DS8000 is dead, based on an update to its features, is one giant misguided assumption, even for you Barry.

My comments on your previous post were related to our SVC 4.2.1 announce which also happened yesterday. See my blog for a quick high level update. Will post some more details when I return.

the storage anarchist

Actually, BarryW, I declared the DS8000 dead precisely because of the LACKLUSTER "update of features." The product is barely feature competitive with the prior generation DMX-2s or USPs of 2004, and is no longer even in the same league as current enterprise-class storage offerings. And when you don't invest to meet customer requirements, you wither and die - like life, the market is not kind.

And I have to admit that your SVC announcement went by without notice - didn't even cross my radar. With an installed base that averages less than 10TB per SVC, your total lifetime shipments of SVCs couldn't virtualize even half of the enterprise-class capacity that ships in a single quarter, nor has the SVC made any detectable dent in the high-end storage market. Plus, as I've noted before, those SVCs are being replaced in enterprise accounts on a routine (and frequent) basis. Scale, performance, complexity, reliability, data integrity - pick your poison, but you probably already know that the SVC is leaving a wake of dissatisfied customers around the world.

Sorry to be so harsh, but it is what it is...and yes, I know you'll haul out a lengthy list of reference customers. But the needs of the vast majority of them would be better satisfied with a single small CLARiiON, IMHO.

Dave Vellante

Let me try again Barry. I have to disagree with you on this one. While I concur with many of your assertions about the DS8000 lagging I come to different conclusions on this announcement. And in my dozens of interactions with SVC customers I'm hearing a dramatically different story.

Here's my take:
http://www.wikibon.org/IBM:_High_end_credibility_is_crucial_for_overall_storage_success

the storage anarchist

Well Dave, your perspective is definitely more centrist than mine. But at least we agree on the fundamental lack of IBM investment in enterprise-class storage :>

But Dave (and BarryW), please don't get me wrong. I know of numerous SVC successes as well, and I don't argue that it has its place.

But the SVC is just not appropriate in the same application spaces that today's enterprise-class storage serves. Given the inarguable (and admitted) data integrity and cache coherency risks of in-band, regenerated I/O approaches, prudent customers think twice before they trade simplicity for the potential cost of silent corruption of their data that SVC (and USP) introduce.

We'll all do customers a better service if we strive to clearly differentiate the various solutions, and accurately reflect the markets they serve. I know enough not to position Symmetrix in the mid-tier market, and I honestly don't think "SVC" should ever be confused with "enterprise-class storage."

Thanks!

Open Systems Guy

In response to your last comment, SVC bills itself as an aggregator of existing storage for open systems... I don't really even think of it as storage. It's nothing to scoff at though- I've been seeing more and more of them lately, and since they now support VMWare which is growing faster than other applications in the open systems space, I can't imagine them starting to shrink their market share.

In response to your post, the DS8000 has not had much in the way of new features, but a bunch of the ones you mention are not really commonly used in enterprise (read mainframe) storage: iSCSI, RAID 6, ethernet replication, and SATA in particular. iSCSI and SATA are mainly used to reduce costs, but iSCSI is not iESCON or iFICON, and a SATA drive would probably be more appropriately placed in another subsystem that doesn't cost as much per DDM slot. The cost savings per GB of SATA are completely lost if you put the drives into a subsystem that can cost as much as a small passenger jet. RAID 6 is usually only needed for high capacity, low spindle speed drives, and ethernet replication is only needed for small shops that don't have a fabric capable of doing FCIP or iFCP. These are not typical of a mainframe environment.

That said, their lack of thin provisioning and hardware updates is certainly not going to help them. One thing I've always hated about IBM is how they never release bleeding edge technology- they were the last of the big 3 server vendors to embrace quad core technology, and their storage always seems to run on generation n-2 internals.

the storage anarchist

OSSG - I'm not sure what segments of the market you work in, but I guess you and I have a very different definition of "enterprise" - in my world, mainframes are no longer the predominant platform for deploying business applications. I know of numerous banks, insurance companies, telecommunications companies, and manufacturers who run their entire enterprises on open systems platforms. The mainframe is long from dead, but "enterprise" and "mainframe" are clearly no longer synonymous.

And for many of these companies, things like RAID 6, IP-based host access and replication, and even SATA drives are a critical component of their enterprises' storage platforms. In a world where the amount of data that must be stored is doubling every 18 months, and an increasing amount of on-line storage is for digital retentation, SATA is no longer just for "cutting costs" - it is the appropriate primary storage for an increasing range of data, from the check images your bank stores for you to the email and IM respositories erquired by compliance rules.

And in fact, some of the most avid interest in SATA is in fact coming from the mainframe community, who for so long have only had 2 choices for storage: fast disk or (relatively) slow tape. SATA affords a new price-performance point for many mainframe applications.

But like I said, we may just be looking at different parts of the market.

Thanks for your perspective!

Open Systems Guy

It's true, the mainframe is certainly not the only enterprise solution, but it's squarely where the DS8000 has the most traction. If a company is lucky enough to have migrated away from the mainframe platform, or started recently enough that they never owned one, then I would agree that the DS8000 is becoming less attractive.

The thing with IBM is that I don't really think they're trying all that hard to innovate in the DS8000 series- it's a stable, mature line that is often used by glass-house shops. Places that use more of an open platform (like intel/windows) usually opt for the DS4000 series. It can't do mainframe, but it does most everything else, and there's been plenty of innovation recently from IBM there.

Barry Whyte

Some interesting thoughts and observations. Glad there were some people standing up against your onslaught while I was AFK.

I'd agree with a lot of what OSSG has said, especially after scan-reading this post while away and thinking about the SATA comments in particular. I'd agree that SATA in an enterprise controller is somewhat questionable, when it negates the cost savings it can bring. Thats where something like SVC wins hands down, the ability to buy a mid-range or low-end SATA controller and use ILM to move data down to that level when its no longer needed on Tier1.

You make 4Gbit FC disks sound like its something that everybody needs. Most shops are IOPs limited, not throughput limited and 4Gbit does very little to help that. As long as you have enough bandwidth into the box and onto the disks, 4Gbit disks give very little gain expect to the vendor thats getting the extra cash. With 8Gbit HBAs and switches not far away from general availability, I'd hedge a bet that says it will be 12-18 months before we even see or think about seeing 8Gbit disk interfaces.

Anyway, as always an interesting viewpoint, from a "Symm-dude" :)

the storage anarchist

BarryW - you and OSSG are making a rather brash assumption over the pricing for SATA drives in a DMX. Although I can't provide specific pricing, I can assure that cost to add SATA drives to a Tier 1 DMX4 configuration is only marginally more expensive than buying those same drives in a separate SATA storage array. There are even built-in discounts for array-based software on SATA-class storage. And in sharing the array infrastructure, the SATA storage enjoys higher performance & availability, and it is easier to manage, replicate and relocate.

And while I understand SVC affords some of these same values (one significant exception is guaranteed data integrity), mainframe customers really have NO REAL ALTERNATIVE to get low-cost/low-performance storage, other than DMX.

YOU two guys may question this, but the growing number of customers who are adding SATA to their DMX4 configs sure don't.

Regarding 4Gb - if it's not so important, why do you even bother supporting 4Gb connections to storage on the SVC? Fact is, 4Gb transcievers cost the same (or even less) than the older 2Gb ones, and the drive manufacturers are just about to shut down production of 2Gb FC drives.

So you have to ask yourself, why wouldn't IBM give customers this "free" benefit, even if it really doesn't add all that much (which is arguable, but irrelevant - 4GBb IS FREE!)?

The answer is obvious: The DS8000 is about to be end-of-lifed for good, and thus there is no justifiable business case for IBM to invest in the design and inventory of a 4Gb back-end for product that is already headed out to pasture.

You don't have to be Karnak to understand the implications: there really are only TWO viable enterprise storage platforms...

Open Systems Guy

I'm not as familiar with EMC technology, so I'm not going to question any statements made about SATA costs in an expensive array compared to a low priced one. I'll defer to your experience :)

SVC is really not part of the mainframe discussion- it only does open systems. You can put it in front of part of a DS8000, but I think it would be redundant, and it would mean that (correct me if I'm wrong Barry W) any replication for the DS8000 storage under the SVC would have to be done by SVC. That's not something likely to be appreciated by DS8000 admins...

As for 4Gb, I agree- the 8000 has always been at n-2 tech when it comes to the parts used to make it. Even the FATA drives offered are still only available in 500GB while most other lines have 750GBs available to them.

I don't know about them putting the DS8000 end of line though- it seems a little drastic...

Barry Whyte

I too cannot counter your cost arguments based on lack of pricing knowledge of EMC products.

SVC needs 4Gbit because it sits in the SAN and can make use of the latest and greatest port speeds. We've proved that you can drive the ports to as close as makes no difference to 100% utilisation. Normally most vendors don't recommend this, usually due to the hardware buses behind the ports being unable to keep up - this is not the case with SVC - while we wouldn't recommend driving the CPU's over 75%, the ports themselves are happy to be driven pretty much at 400 MB/s each.

As for DS8000 being end-of-life, that is FUD. I cannot obviously disclose roadmaps but there is one and it is being actively worked on - nothing more than EMC wishful thinking and an attempt at spinning doubt in customers eyes - which I can whole-heartedly refute.

Chris Mellor discusses this further :

http://www.techworld.com/storage/blogs/index.cfm?blogid=3&entryid=649

The comments to this entry are closed.

anarchy cannot be moderated

about
the storage anarchist


View Barry Burke's profile on LinkedIn Digg Facebook FriendFeed LinkedIn Ning Other... Other... Other... Pandora Technorati Twitter Typepad YouTube

disclaimer

I am unabashedly an employee of EMC, but the opinions expressed here are entirely my own. I am a blogger who works at EMC, not an EMC blogger. This is my blog, and not EMC's. Content published here is not read or approved in advance by EMC and does not necessarily reflect the views and opinions of EMC.

search & follow

search blogs by many emc employees:

search this blog only:

 posts feed
      Subscribe by Email
 
 comments feed
 

 visit the anarchist @home
 
follow me on twitter follow me on twitter

TwitterCounter for @storageanarchy

recommended reads

privacy policy

This blog uses Google Ads to serve relevant ads with posts & comments. Google may use DoubleClick cookies to collect information (not including your name, address, email address, or telephone number) about your visits to this and other websites in order to provide ads about goods and services of interest to you. If you would like more information about this practice and your options for not having this information used by Google, please visit the Google Privacy Center.

All comments and trackbacks are moderated. Courteous comments always welcomed.

Email addresses are requested for validation of comment submitters only, and will not be shared or sold.

Use OpenDNS