Dave Simpson's Storage Blog Archives for April 2009
April 27, 2009 -- The crown jewels in the Oracle-Sun acquisition are Java and, to a lesser degree, the Solaris/SPARC platform and, to a lesser degree, the MySQL database. But what about Sun’s storage lineup?
Here’s a clip from Ellison’s statements on the acquisition: “We will be able to tightly integrate the Oracle database to some of the unique high-end features of Solaris, engineering them to work together and for the first time to deliver complete, integrated computer systems – database to disk – optimized for” yada yada yada, but what disk system(s) will carry the load?
Although Sun has some interesting storage-related software technology (the ZFS file system and Open Storage code/initiative) that Oracle could capitalize on, what about the hardware?
For the most part, Sun is a storage reseller, OEMing its high-end arrays from Hitachi Data Systems and its mid-range gear from LSI’s Engenio division. But complicating the picture, Oracle has its own branded storage array via an OEM deal with HP. Not to mention Ellison’s pet project, Pillar Data Systems (which is funded by Ellison, but separate from Oracle).
And then there’s the Sun StorageTek tape business.
Some pundits have opined that Oracle will jettison all of the storage hardware elements. No way, because they’re already in the hardware business via Sun’s servers and why pass on some decent (although not as decent as software) margin business. However, the company definitely can’t go forward with all of the storage lines. There’s just too much clash.
Should I stay or should I go now?
I predict Oracle goes with two disk array lines. Factoring in Ellison’s ego, one would argue that Pillar stays. However, the HDS and LSI gear gives him a reasonable reason to pull the plug on Pillar and cut his (pocket change) losses. Then again, did I mention that he has an ego? It’s a tough call because Pillar definitely overlaps with Sun’s storage hardware.
Since Sun’s storage business is largely a reseller business, why not just continue to let HDS and LSI do all the work and throw some hardware bones to Oracle’s sales reps? Sure, Oracle sales reps don’t know how to sell hardware, but you wouldn’t have to retain all that many Sun sales reps to keep that business rolling. The downside to this is that it puts Oracle in direct competition with partners such as EMC, NetApp, etc.
Another option: Jettison the HDS, and perhaps LSI, relationship and extend the HP relationship and just concentrate on purpose-built, highly-optimized storage systems for Oracle databases. That option eliminates, or at least diminishes, the prospect of competing with partners such as EMC and NetApp in the more broad-based storage landscape. Then again, the HDS and LSI products are very solid and represent a lot of revenue if Oracle is willing to butt heads with partners such as EMC and NetApp.
And then there’s the Sun StorageTek tape business. Selling that off doesn’t seem to be an option. Who would/could buy it? IBM? Talk about monopoly. I never understood why Sun bought STK in the first place (access to mainframe accounts?), but it’ll be much harder for Oracle to figure out what to do with this business. Then again, it’s a profitable business with relatively low maintenance.
Yet another option would be to essentially spin-off Sun’s storage business.
What would you do with Sun’s storage hardware business if you were Larry Ellison?
posted by: Dave Simpson
April 21, 2009 -- Although overshadowed by the Oracle-Sun acquisition, things got hot in The OC today with Broadcom’s $764 million bid for fellow Orange County, CA resident Emulex.
Actually, this isn’t new. Broadcom approached Emulex in late December with an acquisition offer. Emulex told them to take a hike and subsequently, in January, concocted a poison pill and amended its bylaws (which subsequently led to Broadcom filing a lawsuit) to thwart Broadcom’s overture. What’s new is that Broadcom is taking its case public, or least to Emulex shareholders.
I have no idea if Broadcom will be successful in this bid. I suppose it depends on how Emulex execs and shareholders see the company’s future in the context of competition with QLogic and Brocade on the HBA and FCoE CNA battlefields. Is QLogic ahead of Emulex on the FCoE front? Can Brocade upset the Emulex-QLogic duopoly on the HBA front? If the answer to either of those questions is ‘yes,’ then now may be the time for a sale. And the price -- $9.25 per share -- seems to be sweet: about a 40% premium over yesterday’s share price of Emulex stock.
Obviously, Broadcom wants Emulex primarily for its Fibre Channel expertise and technology; more specifically, perhaps, its FCoE technology as Broadcom moves into the so-called converged networking space (Ethernet + Fibre Channel). As such, Broadcom is probably also eyeing QLogic. And it’s possible that the deal could turn as rapidly as the IBM-Sun to Oracle-Sun deal.
It’s also possible that Emulex could find a white knight or, more likely, push the acquisition price closer to the $1 billion mark.
If the Broadcom-Emulex deal does go through (financial analyst speculation is running about 50-50), it would have repercussions and a ripple effect across the storage industry, and would lead to more acquisition speculation. QLogic would be in play (with Brocade as a possible suitor, or maybe Cisco). Maybe even Brocade would be in play.
Given the Oracle-Sun acquisition, anything can happen.
Related articles:QLogic delivers single-chip CNAsEmulex CNAs support 10GbE, iSCSI, FCoEBrocade launches FCoE switch, adapters
posted by: Dave Simpson
April 20, 2009 – The channel bid adieu to the first quarter of this year with a collective “good riddance.” That’s one takeaway from a recent survey of 51 IT resellers conducted by Robert W. Baird & Co.
How bad was it? In terms of expected revenues vs. actual revenues, 60% of the VARs were below plan, 26% were on plan, and only 14% were above plan. That’s the worst quarter since RW Baird began its quarterly server/storage VAR surveys in Q3 2004. In fact, it was almost twice as bad as the previous worst quarters (Q2 2006 and Q1 2008, when about 32% of the resellers were below plan).
Once again, storage was cited as the strongest IT sector by resellers, with virtualization and data de-duplication cited as the hottest technologies.
It’s tough to predict how VARs will fare in the second quarter. About 43% of the survey respondents expect the market to remain about the same, while 30% are more negative and 16% are more positive (and 11% say it’s too early to tell). On the positive side, many VARs said that, although many purchases were delayed from Q1 to Q2, relatively few purchases were cancelled.
When asked which vendors’ products they expect to sell more of in the next 12 months, VARs cited VMware, NetApp, Data Domain and Cisco as the top four performers, with CommVault and LeftHand rounding out the top six.
When asked about specific vendors in the short term (“How do you perceive the following storage vendors in the next 3 to 6 months?”), Data Domain, NetApp and EMC topped the list, with LeftHand moving into the fourth position from seventh in the previous quarter. (LeftHand was also ranked the most channel friendly vendor, slightly above NetApp.)
Sun ranked consistently low in the survey, and it will be interesting to see how the channel reacts to the Oracle-Sun announcement. I”ll blog about that news as soon as I get done scratching my head.
posted by: Dave Simpson
April 14, 2009 – EMC’s announcement today of its Virtual Matrix Architecture and Symmetrix V-Max
systems certainly didn’t lack hyperbole. EMC chairman, president and CEO Joe Tucci called it the most significant Symmetrix innovation since EMC debuted the platform 18 years ago.
Not surprisingly, Hitachi Data Systems vice president and CTO Hu Yoshida was quick to weigh in on EMC’s announcement in his “Don’t Confuse Symmetrix V-Max with Storage Virtualization”
blog (even though EMC didn’t in fact equate V-Max with storage virtualization). Referring to the V-Max as “just another big monolithic storage system,” Yoshida makes the point that V-Max is as locked in as vendor lock-in gets. Not only can you not put other vendors’ disk arrays in the Virtual Matrix Architecture, you can’t even put EMC’s Clariion or DMX arrays in it. Which was a good springboard for Yoshida to sing the praises of HDS’ USP V system, which virtualizes heterogeneous storage platforms.
Not that anybody expected EMC to embrace heterogeneity in its new architecture.
Even the analyst community couldn’t resist hyperbole in covering this announcement. For example, Pund-IT Research’s
Charles King and the Mesabi Group’s David Hill wrote that the Virtual Matrix Architecture and V-Max systems “are likely to incite a tectonic shift in the enterprise storage market” and that the improvements (relative to DMX-4) “range from the simply impressive to the seriously jaw-dropping.”
And, sending me agoogling, King and Hill noted that the Symmetrix V-Max systems are “based on highly integrated ‘layers,’ resulting in a delectably enterprise-class equivalent of a Sachertorte.”
In the past, every time EMC made a momentous move (say with the acquisitions of Documentum, RSA and VMware), I questioned their strategy and even predicted that some of their acquisition follies would end in failure. I was always wrong.
I quit nay-saying the Hopkinton honchos a long time ago. Sure, there are a few chinks in their armor, but they have more arrows in their quiver than a few of their competitors combined.
The V-Max systems and Virtual Matrix Architecture will no doubt prove successful, despite the vendor lock-in and the starting price of $250,000 – not exactly the antidote for what’s ailing the IT economy (although the systems do provide a lot more bang for the buck than DMX-4s; e.g., 3x the performance, 3x more capacity and 2x the connectivity).
The architecture may not be storage virtualization per se, but it will play well in huge server virtualization environments, particularly in managed service provider environments trying to capitalize on cloud-based computing/storage/virtualization.
For EMC exec Chuck Hollis’ blog on the announcement, see “Symmetrix V-Max: Storage Architecture Redefined.”
It’s biased, but interesting.
posted by: Dave Simpson
April 8, 2009 – At the only press conference at this week’s Storage Networking World confab, Brocade tossed its hat into the FCoE ring with the announcement of a switch and converged network adapters (CNAs). See “Brocade launches FCoE switch, adapters.”
That announcement seemingly officially kicked off the FCoE battles between Cisco and Brocade on the switch front, and Emulex, QLogic and Brocade on the CNA front. (As storage bigots, we’re ignoring the traditional NIC and switch vendors which may, or may not, eventually be contenders in the FCoE ring.)
Brocade’s CNA announcement came on the heels of similar announcements from Emulex and QLogic:“QLogic delivers single-chip CNAs”
“Emulex CNAs support 10GbE, iSCSI, FCoE”
Upon each announcement, competitors offered their services in bad-mouthing their rivals. Since most of the claims and counterclaims are at this point are impossible to verify, we chose a Joe Friday approach to reporting the announcements – “Just the facts, ma’am.”
The FCoE battles may have begun in the OEM qualification arena (raising the question of how many CNAs or switches the big server/storage vendors will qualify/resell), but the real battles won’t begin until end-user adoption of FCoE begins. At yesterday’s press conference, Brocade CTO Dave Stevens predicted that we won’t see end-user adoption of FCoE in any volume until late 2010 (about the same time that 16Gbps Fibre Channel products are supposed to appear). And that might be optimistic.
posted by: Dave Simpson
April 7, 2009 – At a show where it wasn’t hard to headline, Brocade topped today’s new product announcements with its Fibre Channel over Ethernet (FCoE) and Converged Enhanced Ethernet (CEE) switch and an accompanying set of FCoE-based converged network adapters (CNAs). See Kevin Komiega’s news post here.
Brocade’s announcement officially set off the FCoE wars. Or not. I’ll blog about that tomorrow.
In other news at the Storage Networking World show today, LSI announced that it has signed a definitive agreement to acquire the assets and intellectual property of AMCC’s 3ware RAID adapter business for approximately $20 million in cash. 3ware products include SAS and SATA RAID adapters. Upon closing, the 3ware business will be integrated into LSI’s Engenio Storage Group.
Also at the show, LSI demonstrated performance of more than one million I/Os per second (IOPS) in a server with three of the company’s 6Gb/s SAS chips (LSISAS2108 RAID-On-Chip, or ROC, ICs) connected to 12 6Gb/s SAS disk drives. LSI officials predict that 6Gb/s SAS servers and storage systems will hit the streets late this year, with end-user adoption commencing early next year.
We’ll wrap up our SNW product coverage tomorrow on InfoStor’s home page.
posted by: Dave Simpson
April 6, 2009 – Granted, it’s only the first day of the Storage Networking World show, and there’s a storm, and the economy’s in the tank, and the exhibits don’t open until tomorrow, but things are sloooow here in Orlando.
No waiting in line to register, no background noise (or people) in the press room, only a spattering of product announcements in my inbox, and even the bar was empty at 5:00 (a sure sign that the show isn’t what it used to be). The only outfit doing brisk business here is Starbuck’s.
The biggest product announcements today were 3PAR’s mid-range disk arrays (see “3PAR ships quad-controller array”
) and FalconStor’s Backup Accelerator option for its VTLs (see “FalconStor speeds up backups”
), which is only major because FalconStor is the leader in VTLs and VTLs are hot (but don’t solve all the problems associated with tape-based backup environments).
Oh, yes, and you can add yet another cloud-based storage services vendor to the rapidly growing list of players in this space (see "Zetta launches cloud storage service"
You know that a technology is hot when the Storage Industry Networking Association (SNIA, which co-sponsors SNW) forms a group dedicated to it. SNIA today announced the Cloud Storage Technical Work Group.
In other announcements today . . .
Xiotech, which a year ago at SNW launched its Intelligent Storage Element (ISE) technology with great fanfare, today announced an energy-efficient feature -- dubbed PowerNAP – for its Emprise 5000 disk arrays. As its name implies, PowerNAP powers down drives so that they consume less than 24 watts (compared to 500 to 600 watts in power-up mode). The technology is based on Wake-On-LAN (WOL) technology, and company officials are quick to point out that it’s not massive array of idle disk (MAID) technology.
STORServer used SNW to showcase its STORServer Agent for VMware Consolidated Backup (VCB), which integrates VCB with IBM’s Tivoli Storage Manager (the engine behind STORServer’s Business Continuity Appliances). The agent provides centralized management, reporting and scheduling of virtual machine (VM) backups. STORServer also sells an appliance-based approach for integrating VCB and TSM.
I also sat down with Rob Stevenson, director of storage research at TheInfoPro
, which recently completed Wave 12 of its ongoing surveys of Fortune 1000 storage professionals. Although it’s no surprise that data de-duplication (in a disk-to-disk backup context) topped the charts in TIP’s Technology Heat Index, it’s a little surprising that data de-duplication for primary storage was so high on users’ implementation plans.
posted by: Dave Simpson
April 3, 2009 – Well, Storage Networking World at the Rosen Shingle Creek resort in Orlando is close
to the happiest place on earth, albeit only from a geographic perspective.
Senior Editor Kevin Komiega and I are heading to SNW next week and will be reporting on major product announcements at the show (if there are any), as well as blogging about the show in general.
So far, this is looking like one of the slowest SNWs since the early years of the show. I’m judging that partly from the number of briefing requests we get from vendors, and this show is off by about 50% compared to previous SNWs.
Although the show organizers claim
a record number of end-user attendees, the exhibitor list is way down. Is that because the show organizers have followed the Comdex model and priced themselves out? Or is it just a reflection of the down economy, budget cuts and/or restricted travel? We’ll collect opinions from show-goers and let you know.
So check into our home page
and blogs next week if you’re interested in what is still the storage industry’s largest trade show and love fest (and job searching venue).
posted by: Dave Simpson