Nobody knows if Isilon Systems will be the next storage vendor to get acquired, but it’s clear that Isilon executives want to be the next. The scale-out NAS vendor has hired a firm to explore a sale, and has been linked with just about every large storage vendor since Hewlett-Packard gobbled up 3PAR last month.
The New York Post last Friday reported EMC is in exclusive talks to buy Isilon for more than $2 billion. I don’t know about the exclusive and $2 billion parts, but it’s likely that EMC is interested in Isilon. So are other storage vendors, mainly because EMC and others have failed to either develop their own clustered NAS or successfully integrate the technology after acquiring it from others.
EMC took a shot with its Hulk and Maui projects, which turned into its Atmos cloud platform but not the type of scale-out NAS system that media and Web companies crave for their storage needs. NetApp acquired Spinnaker in 2003 and still hasn’t fully integrated its clustered IP into the DataOntap operating system. IBM created its own, brining out SONAS last February. It’s too early to say if that product is a success, but it figures prominently whenever IBM presents its storage roadmap. HP acquired PolyServe in 2007and when that didn’t work out, it bought Ibrix last year. HP also talks up its XP9000 platform based on Ibrix a lot these days. Dell acquired the IP of Exanet earlier this year, and plans to use it for clustered NAS and multiprotocol storage in combination with EqualLogic but has yet to bring it to market.
In a research note today, RBC Capital Markets analyst Amit Daryanani, suggested that much of the interest in Isilon could be defensive. In other words, vendors want to keep their rivals from buying it.
“We suspect EMC would be the most likely acquirer, as it lacks a storage platform with a single management platform and its Celerra-NAS platform doesn’t scale as efficiently as Isilon,” Daryanani wrote. “Plus, the harm this could do to NetApp would make the deal even more appealing.”
As for NetApp, he wrote: “We would not be shocked to see NetApp interested in acquiring Isilon as a defensive move to keep its unified storage advantage versus EMC. Also, an Isilon acquisition by any of the other suitors would harm NetApp’s revenue stream.”
Daryanani also mentioned Cisco, Dell, HP, IBM, and Oracle as possibilities but admitted Cisco and Oracle were unlikely to make a bid.
The feeling here is that HP and IBM are willing to go with their relatively new scale-out products rather than throw money at another option. NetApp probably won’t get into another bidding war with EMC after losing out for Data Domain last year. Dell might find integrating Exanet more difficult than originally anticipated, but Isilon’s technology is proprietary and does not work with other vendors’ storage. That would make it even tougher to integrate with EqualLogic.
That leaves EMC, which could position Isilon at its clustered NAS for high performance, rich media and web customers while clearing the way to consolidate its current Celerra NAS with its Clariion SAN platform.
One thing is clear: Whoever buys Isilon will be overpaying. That’s because any large vendor could have picked it up for a lot less money a couple of years ago instead of exploring other options.
Iomega is getting into high-performance Flash. The first step is a portable drive for consumers and SMBs it will begin shipping next month. The next step is adding solid state drives (SSDs) to its NAS platform.
That’s the plan laid out by Jonathan Huberman, president of EMC-owned Iomega. The vendor today launched the Iomega External USB 3.0 SSD Flash Drive that is about the size of an iPhone. The SSD device comes in capacities of 64 GB, 128 GB and 256 GB.
Huberman said the same form factor will be used for Iomega’s NAS soon. He doesn’t have a timeframe because the company has to make tweaks to optimize it for the SMB NAS devices.
“It will be the same flash and the same form factor,” he said. “We’re architecting our NAS stack now.”
Huberman said USB 3.0 support is the key driver for the external drive, as well as decreasing Flash prices. “Without USB 3, you miss a lot of the benefits of Flash because you’re capped by USB 2 performance,” he said. “And price points continue to come down. Historically, the price/value equation wasn’t attractive. Now it makes sense for our customer base.”
The External SSD Flash drive costs $229 for 64 GB, $399 for 128 GB and $749 for 256 GB.
EMC and Oracle shared a lot of headlines this week. Two days ago, EMC brought out its Greenplum Data Computing Appliance that will take on Oracle’s Exadata system. Then Thursday, the companies were linked by rampant rumors that Oracle is looking to buy EMC in a blockbuster deal.
Despite the speculation, it’s a lot more likely they will go on as competitors and in manay cases partners than they will merge into one company.
It’s hard to say exactly how the rumor got started. There was a note from a Wall Street analyst Wednesday speculating on companies that Oracle might buy. EMC was on that list of 12 companies, listed as a potential long-shot. Still, that touched off a spark among investors, who had probably been hearing whispers that Oracle wants more storage and is intereted in VMware. EMC owns more than 80% of VMware, but is unlikely to sell that valuable asset on its own.
“I don’t know who started this talk,” a Wall Street analyst who covers storage told me. “I am sure there are bankers pitching EMC to Oracle, even if that deal doesn’t make any sense. Usually the target leaks these things, but I’m pretty sure EMC did not leak or start the rumor.”
After all the big deals we’ve seen this year in technology, more credence is given to rumors than usual. But this deal doesn’t seem credible, for several reasons:
Too expensive – Most analysts say EMC plus VMware would be worth more than $50 billion. Oracle would have to finance at least two-thirds of that, and that might not be so easy to do these days.
Too complex – Integrating large companies always brings about a transition period where product development and sales take a hit. The bigger the company that gets integrated, the longer this period usually lasts. Oracle is just about finished with its Sun integration, is this a good time to start another?
Too much technology – Oracle has shown with Exadata and its recent ZFS Storage Appliance upgrades that its strategy is to sell storage specifically to make its databases run better. While EMC would make a case for its storage doing that, what about all of its other products? Does Oracle want EMC’s full backup, security, and content management platforms? It would have to pay for all those pieces whether they fit or not. It could sell off the parts it doesn’t want, but that just adds to the complexity of the deal.
If Oracle CEO Larry Ellison wants a broader storage portfolio than he got from Sun, he’s better off going for a smaller company such as NetApp. If he wants VMware, he’s probably out of luck.
Overland Storage today picked up the intellectual property of failed clustered file system startup MaxiScale, and Overland CTO Geoff Barrall said the technology will enable the vendor to deliver scale-out versions of its SnapServer NAS platform.
Barrall said about five to 10 of Maxiscale’s engineers will join Overland, which acquired the Snap portfolio from Adaptec in 2008 for $3.6 million in an attempt to become a storage systems vendor instead of only selling tape. The Snap line is a key piece of Overland’s turnaround plan under CEO Eric Kelley, who was Snap’s CEO when Adaptec bought it in 2004.
MaxiScale’s Flex software was developed to run with commodity hardware. Barrall said he’s hoping to release Snap clustered NAS systems by mid-2011. Overland has already brought out Snap iSCSI SAN and unified storage products.
“Adding scale-out feature to SnapServer line was definitely on the list of things we wanted to achieve,” he said. “This is a great opportunity for us.”
Overland did not disclose the purchase price, but MaxiScale had $25 million in venture funding and no paying customers. MaxiScale launched its first product in Sept. 2009 and was preparing to bring out a second-generation product when it ran out of money.
Barrall said Overland would not market MaxiScale’s existing or planned products, which were targeted at customers such as cloud providers and web companies with large amounts of small files and millions of concurrent users. He said he Overland will tailor MaxiScale’s technology intended for high-end products to the lower-end and midrange NAS markets that Snap addresses.
“They were targeting large server clusters,” he said of MaxiScale. “Their technology scales well. You really don’t have this technology in products in the price range we sell into.”
He didn’t rule out using the technology for a new product platform down the road, though. “A file system comes with the client that you can access through the NAS,” he said. “You can install the client on a server for direct access to the cluster. You can bring Exchange and SQL databases onto the NAS. There’s definitely interesting things we can do there.”
But the first priority is to make MaxiScale’s technology fit with Snap products. “Our goal is to make the technology straightforward to use,” Barrall said. “We want to make it easy for customers to scale and keep all their data in one area.”
Barrall said there were connections between the companies because several Maxiscale executives were former Snap employees before Overland bought Snap. There is another connection – Barrall and Maxiscale CEO Gianlucca Rattazzi were founders of NAS vendor BlueArc.
MaxiScale raised just under $8 million in a B funding round over the last year, but it wasn’t enough to keep it going.
“MaxiScale had a great story, but needed execution,” Server and StorageIO analyst Greg Schulz said. “In some ways it was a company with IP focused for a market that had not evolved enough to be commercially viable. In the end, great stories and strategies need execution, time, money and patience – at least two of which investors often are not comfortable with.”
QLogic this week sought to raise the bar for network convergence products. The vendor launched its third-generation 10-Gigabit Ethernet converged network adapter (CNA) and NIC, and its first converged LAN-on-motherboard (cLOM) chip. The products can help nudge the industry toward Fibre Channel over Ethernet (FCoE) adoption, or at least introduce more Ethernet into storage networks.
QLogic’s new products allow organizations to run FCoE, iSCIS and Ethernet traffic concurrently and use GigE and 10-GigE from one chip with its FlexLOM technology. John Spencer, QLogic’s manager of product marketing for its host solutions group, said he expects OEM and channel partners to sell the new products by the end of the year. Spencer didn’t name any partners, but QLogic’s press release quoted Dell, EMC, Hewlett-Packard and IBM.
QLogic also added features it calls ConvergeFlex, VMflex, FlexOffload, and SecureFlex. Ovbiously, the theme here is flexibility.
ConvergeFlex lets customers change protocols without taking down their servers.
VMFlex lets organizations carve physical ports into virtual ports and assign guaranteed bandwidth to each port. Customers can create four virtual ports per physical port and works with any 10-GigE Ethernet switch.
FlexOffload moves the workload from the processor to the adapter, freeing up bandwidth for virtual machines and applications.
SecureFlex encrypts data in flight over the network, and leaves encryption of data at rest to the storage array.
Wikibon analyst Stu Miniman said the best thing QLogic has going for it is expertise with all the protocols required for convergence while its competitors have had to add pieces of the stack.
“The biggest difference I see is QLogic has flexibility across Fibre Channel, Ethernet with FCoE and isCSI, and InfiniBand,” he said. “Today QLogic is the only vendor with the ability to run FCoE, iSCSI and TCP/IP concurrently. QLogic has more options than the competition.”
In a blog on QLogic’s new products, Miniman wrote the pieces for an all-Ethernet data center are emerging. However, he warned that convergence may be less of a money-saver than a process of merely keeping up with growing bandwidth requirements.
“Many people think that convergence will reduce spending but in fact it will only allow us to keep pace with the escalating server bandwidth requirements,” he wrote. “Wikibon estimates that in the next 72 months, servers will need close to 100X the bandwidth of today but traditional methods of delivering that capacity will fall far short of requirements. We believe that the market will need 6X what Moore’s Law alone will deliver. As a result, users should plan on buying more, not less equipment and they need to endeavor to find ways to reduce I/O consumption using techniques such as compression.”
IBM launched two new storage systems today. The Storwize V7000 is a brand new midrange array and the DS8800 is an upgrade from its DS8700 high-end enterprise SAN system.
IBM will hold a live event and webcast this morning to formally announce the systems, and we’ll have more on SearchStorage.com following that event, but the vendor has already released some information on the products.
Interestingly, IBM is using the Storwize brand for its new midrange array without any of Storewize’s primary data compression technology in the product. IBM acquired Storwize for around $140 million in July.
The Storwize V7000 is IBM’s only internally developed midrange system. IBM also sells midrange storage through its OEM deal with LSI, but recently discontinued the internally developed DS6800 that it had positioned as a high-end midrange/low-end enterprise product. IBM positions the V7000 as competitive with EMC Clariion CX4, HP EVA and Hitachi Data Systems AMS storage.
According to a blog by IBM Master Inventor Barry White today, the V7000 has the same SCSI front-end and back-end as the IBM’s SAN Volume Controller (SVC) storage virtualization product, and can virtualize the same storage from competitors as the SVC. “Why not stick your Clariion behind one and see a performance boost,” Whyte wrote.
The V7000 includes Easy Tier, thin provisioning, dynamic migration and FlashCopy built in.
The Storwize V7000 system has 2U controller and expansion enclosures. Controller enclosures consist of dual controllers and drives, and expansion controllers only contain drives. The system supports 24 2.5-inch drives or 12 3.5-inch drives, and each controller enclosure can connect to nine expansion enclosures. The V7000 holds 6 Gbps SAS drives and 300 GB solid state drives (SSDs), and a system can contain 2.5-inch and 3.5-inch enclosures.
Each enclosure can hold 24 TB with 12 2 TB nearline SAS drives or 14.4 TB with 24 600 GB performance SAS drives. A 20U half-rack system can hold 480 TB of raw capacity. The V7000 supports 300 GB 2.5-inch E-MLC (enterprise-grade multi-level cell) SSDs, and can hold 72 TB of SSD capacity in one system.
A dual-controller system has eight 8 Gbps Fibre Channel and four Gigabit Ethernet iSCSI host ports.
The DS8800 consists of incremental changes from the IBM’s DS8700 high-end enterprise SAN system, with most of the microcode remaining the same. The major design change is the DS8800 has front-to-back cooling instead of the DS8700’s chimney design of taking cold air from front and back and sending hot air out the top.
The DS8800 scales higher than the DS8700, and uses faster Power6-plus processors with dual-core 8 Gbps Fibre Channel and FICON host adapters. It also supports 2.5-inch 6 Gbps SAS drives. The denser DS8800 holds 1,000 drives in three frames, compared to five frames required for the same amount of drives on the DS8700.
IBM said the DS8800 won’t support Easy Tier sub-LUN automated tiering software that moves data between SSDs and hard drives until next year.
Hewlett-Packard picked up a new storage leader as well as a new lead storage platform from its $2.35 billion 3PAR acquisition.
HP today named former 3PAR CEO Dave Scott senior vice president and general manager of its StorageWorks division, reporting to EVP of enterprise servers, storage and networking Dave Donatelli. Scott replaces Dave Roberson, the former Hitachi Data Systems CEO who will be re-assigned inside of HP.
Scott ran HP’s XP enterprise storage division before taking the 3PAR CEO job.
Scott will oversee a group that will continue to sell HP’s current storage platforms, but 3PAR’s InServ family becomes the new flagship. “We’re clearly positioning 3PAR front and center in the midrange and enterprise markets,” HP StorageWorks marketing VP Tom Joyce told StorageSoup.com today in a phone conversation from Barcelona, where HP execs are meeting with European media. “We will lead with the 3PAR F Series and T Series, and if a customer wants an EVA or XP we’ll sell them.”
Joyce’s comments echoed those made by Donatelli last week at HP’s analyst day. Although HP refreshed its high-end P9500 (formerly XP) family last week and Joyce said the vendor plans to upgrade the EVA, he made it clear that HP sees 3PAR as the key to increasing its storage market share.
HP had 11% external storage market share in the second quarter of this year according to IDC. That placed HP tied for third with NetApp behind EMC and IBM. Those numbers show there are a lot of SAN customers who haven’t bought EVA or XP, and HP will chase them with 3PAR. But Joyce said HP won’t push customers off its current flagship, the EVA.
“EVA has well over 100,000 customers and a lot of loyalty,” Joyce said. “We’ll continue to drive that product and deliver feature functionality on our roadmap for next year. If customers want to stay on EVA and buy more, we’ll sell them.”
Joyce said it would be difficult to port 3PAR technologies such as its thin provisioning and Adaptive Optimization automated tiering to other HP platforms, but he pointed out the iSCSI SAN family HP acquired from LeftHand Networks has similar features for the lower end of the midrange.
“LeftHand is definitely below where 3PAR hits, but they have things in common: they run on industry standard gear, they’re built to scale out with thin provisioning, and they’re software management is all inclusive so you don’t have to buy all the options like you do on other arrays,” he said.
That suggests HP’s future storage portfolio will rely heavily on scale-out architectures, with 3PAR, LeftHand and the X9000 NAS – built on technology acquired from Ibrix – as its major platforms.
Like other large storage vendors, IBM gets criticized for having too many storage platforms and for not making it clear which products are best for certain markets. But IBM is looking to clarify the position of several of its systems by combining them as part of its cloud storage strategy.
Todd Neville, product development lead for IBM Cloud Storage, told StorageSoup that its XIV and Scale-out Network Attached Storage (SONAS) systems are key pieces of Big Blue’s cloud strategy both separately and in combination. He said Storwize data reduction software will also play a role in the cloud when IBM integrates it with its hardware.
“Going forward, you will see various aspects of the IBM portfolio coming together and it will be obvious as to how they complement each other,” he said.
Neville said there are four key components of IBM’s cloud storage strategy. Two of them are services — Smart Business Cloud Storage (SBCS) and Storage Cloud Services (SCS). IBM also launched SONAS based on its General Parallel File System (GPFS) with the cloud in mind, Neville said. He referred to SBCS as a managed SONAS.
Then there is XIV. It’s been almost three years since IBM bought XIV, and Big Blue hasn’t done a good job convincing people where the product fits in its enterprise storage strategy. Neville said XIV’s main role is as a cloud product.
“People are confused, they say ‘How does that position with these other [cloud] things?’” he said. “Well, it’s really complementary. The cool things in XIV are the nuts and bolts in the storage itself – the RAID, the provisioning, and the way you’re laying out and retrieving data from disk drives. Now look at the really cool intellectual property on SONAS, a lot of it’s the presentation of data to the end user – the namespace, file system, clustered CIFS, clustered NFS. These two are very complementary. They’re not walking on top of each other at all. They’re also both very scalable. Today we ship SONAS two ways – as native SONAS or we can put an XIV behind SONAS as the storage device.”
Neville said the two platforms will integrate more. He also said IBM has no plans to discontinue or halt development of XIV now that XIV founder – and EMC Symmetrix creator – Moshe Yanai has left IBM.
“We are absolutely committed to XIV,” he said.
IBM also looks to SONAS to help XIV scale beyond its single-box 79 TB usable capacity limit, Neville said.
“Any product has limit on the individual box,” he said. “You have the ability to put multiple XIV boxes under a single SONAS. The native SONAS software scales to something like 14 TB. And we have the GPFS layer on top of it. We have implementations in the field that are in the multiple PBs. There’s no reason an XIV-SONAS combination couldn’t be very large. We can do it today with multiple XIVs under the SONAS.”
“I can’t comment on what we’re going to do to the XIV hardware to make it more scalable, that’s a whole other discussion.”
Neville said Storwize will also play in IBM’s cloud picture, although that wasn’t the main reason IBM acquired the data reduction vendor.
“Storwize is an interesting asset,” he said. “I’ve been looking at them a lot more since we acquired them. I think they will play a prominent role across a number of different products. There’s potential for them immediately in SONAS, which has a good use case for compression and deduplication and the kind of things the Storwize algorithm can do. I’m talking generically, not just what it can do today. The technology’s good for a lot of things. I see it applied toward these cloud items like toward SONAS, SBSC, SCS, but it is also applicable toward a different type of compression and dedupe level at the block level. I envision it with some of our block products as well.”
Now that Hewlett-Packard has closed its $2.35 billion 3PAR acquisition, HP executives are looking for 3PAR’s InServ storage systems to fill three spots in its product lineup.
HP’s GM of enterprise storage, servers and networking David Donatelli said Tuesday during HP analyst day that 3PAR is a technology leader in the storage space and a key to his plans to revive HP’s storage business. Pointing to HP’s 11% market share in external storage, Donatelli said “storage is an area we really want to improve on. We want to go after the 89 percent of the market we don’t have.”
And HP is counting on its shiny new toy to chase that market share.
“3PAR has the new storage software architecture on the market, and it has all the features customers want,” Donatelli said, mentioning thin provisioning and sub-volume automated tiering among those features.
Donatelli said 3PAR’s products will complement HP’s current storage lineup without replacing any platform. The vendor plans to continue its P4000 (Left-Hand iSCSI), P6500 and P8500 (EVA) and P9500 (XP) families, as well as its scale-out NAS (Ibrix). But he said EVA would become a lower mid-tier platform with 3PAR taking over the high end of the midrange, open systems enterprise, and cloud storage segments.
“With 3PAR, we can cover multiple markets with a single product,” Donatelli said. “Everybody has unique products by market segment, even we do. But for the service provider and cloud space we had nothing. 3Par covers mid-tier, the high end and the cloud. You can start it small [with two controllers] and grow it [to eight controllers]. It supports three distinct markets.”
HP launched its P9500 enterprise system Monday based on the same architecture as Hitachi Data Systems’ Virtual Storage Platfrom (VSP). Donatelli said that product “is great for our installed base and mainframes.”
Hewlett-Packard is launching is version of Hitachi Data System’s new VSP system today, which should end speculation that HP will sever its OEM deal with Hitachi’s parent company for its enterprise storage platform.
HP is dropping the XP brand for its enterprise system, instead calling it the HP StorageWorks P9500 Disk Array. That’s part of HP’s new naming structure, where the MSA became the StorageWorks P2000, LeftHand iSCSI SANs became P4000 and the EVA turned into the P6400 and P8400. But the P9500 is a continuation of HP’s OEM deal with Hitachi Japan (not HDS).
“It’s still XP, but it’s called the P9500,” HP StorageWorks VP Tom Joyce said.
HP’s acquisition of 3PAR for $2.35 billion raised the question of whether it would end its Hitachi OEM deal. 3PAR positions its InServ T-Class arrays as a direct competitor to the XP line, as well as enterprise systems from HDS, EMC and IBM. But 3PAR lacks mainframe connectivity found in the XP/P9500, USP/VSP, EMC Symmetrix, and IBM DS8000, and that’s reason enough for HP to maintain its Hitachi OEM deal.
“It’s premature for us to talk publicly about 3PAR because we haven’t closed it [the deal closed today], and we don’t want to get ahead of ourselves,” Joyce said. “But a lot of customers need mainframe capability and this [P9500] has it. A lot of customers have made an investment in XP and we want to keep those customers happy.”
What else about the P9500can keep customers happy? Joyce pointed to its use of Intel architecture as a key to HP’s converged infrastructure strategy and the automatic sub-LUN tiering capability built into the new architecture. HP, which sells its own management software with the XP/P9500, also has new metered licensing software for the P9500.
HP will eventually have to answer questions about where it will position its 3PAR products. When HP first bid for 3PAR, HP storage chief Dave Donatelli said it could fit in the midrange. HP’s midrange EVA has been losing market share and people in the industry say it lacks the features of its competitors, most notably the EMC Clariion. Still, the EVA has a large installed base and HP wants to keep those customers happy, too.
David Floyer, CTO of analyst firm Wikibon, said 3PAR’s arrays are not sufficient to replace HP’s enterprise platform.
“3PAR is not tier one,” he said. “It makes everything work well, but it’s not the box you want if you have a huge database that is temperamental and you have to throw everything at it to work well.”