Thursday, June 24, 2010

PMC girds for channel battles, opportunities

June 24, 2010 – PMC-Sierra completed its acquisition of Adaptec earlier this month, and is now fine-tuning its game plan for entering the channel. PMC acquired Adaptec’s channel storage business, including its RAID hardware and software products.

PMC-Sierra paid approximately $34 million for Adaptec, which at one time (about ten years ago) had annual revenues in the $800-million ballpark.

In case you’re not on the chip side of the storage industry: PMC-Sierra is a fabless semiconductor company that originally focused on telcom chipsets but has since branched out into many other industries. The company entered the storage market about nine years ago with Fibre Channel chips, but has more recently focused on SAS chipsets. PMC’s enterprise storage business is now the company’s largest business segment.

Until the Adaptec acquisition, PMC-Sierra focused primarily on semiconductors for OEMs. But in addition to Adaptec’s RAID product line, PMC now has access to Adaptec’s extensive channel partner network.

“We’ll sell Adaptec’s products through their channel, but the acquisition also gives us an avenue to sell PMC products more broadly, beyond our OEMs,” says Mark Stibitz, vice president and general manager of PMC-Sierra’s enterprise storage business.

On the storage semiconductor front, PMC competes primarily with LSI and, to a lesser degree, vendors such as Marvell, Emulex, Intel and others.

On the RAID adapter/controller front in the channel, PMC will continue to compete most directly with LSI (which is the #1 vendor in the x86 RAID channel market, followed by Adaptec at #2), as well as other vendors such as HighPoint, Promise Technology, Areca and Atto Technology.

In other words, the PMC-LSI battle will turn into a war.

Meanwhile, LSI has expanded its channel presence by offering its white box partners external RAID building blocks (see “LSI expands in channel with external RAID”).

The channel-based x86 RAID market is estimated at $200 million/year.

Adaptec has been primarily using Intel’s RAID-On-Chip (ROC) ASICs in its 3Gbps SAS products, and will continue to develop and support the Intel-based products. But when the company makes the transition to 6Gbps SAS – its next major product introduction – it will use PMC-Sierra’s 6Gbps SAS chips, according to Jared Peters, formerly with Adaptec and now vice president and general manager of PMC’s channel storage division. The company is expected to release the PMC-based 6Gbps SAS products in the first quarter of 2011.

As a result of PMC’s acquisition, the vendor formerly known as Adaptec has changed its name to ADPT Corp.

Related articles from InfoStor:
PMC-Sierra to buy Adaptec’s channel storage business
LSI ships 6Gbps RAID array
Adaptec enables SSD + HDD RAID [MaxIQ SSD caching products]

Tuesday, June 22, 2010

Ocarina: 4 dedupe predictions, and an OEM strategy

June 22, 2010 – A couple of recent news announcements/rumors have focused attention on data reduction (compression/deduplication) for primary storage, also referred to as primary storage optimization, capacity optimization, storage efficiency, and many other terms that may be confusing end users.

First, Permabit announced a software-only version of its data deduplication technology designed for storage hardware and software OEMs (see “Permabit deduplicates primary storage”). If Permabit succeeds in nailing a major OEM deal or two (and the company claims to have at least one in the works) this could be a game changer. For a second opinion on that, see Jeff Boles’ blog post.

Second, the rumors are still swirling around IBM’s alleged negotiations to buy primary storage compression specialist Storwize (see “IBM to acquire Storwize for $140 million?”).

I recently spoke to Murli Thirumale, CEO at Ocarina Networks, which is a Storwize competitor and soon-to-be Permabit competitor. Leading up to a product/strategy announcement that Ocarina made today (see below), Murli offered four predictions on data reduction for primary storage:

--“Every major storage vendor will have primary storage deduplication in their portfolio by 2011, whether by building it or buying it. And by 2012 every major host vendor – including servers and virtualization platforms – will have data reduction built in.”

--“The compression vs. deduplication argument will go away as users realize that you need both. Each technology has advantages/disadvantages for different data types and access patterns.” [Ocarina’s technology combines compression and deduplication.]

--“The industry will eventually have solutions integrated across all tiers of storage, rather than point solutions for different tiers. You shouldn’t have to re-hydrate across tiers or workflows. Once you shrink the data, keep it shrunk.”

--“Deduplication is not a feature; it’s a business. The data deduplication market will exceed $3 billion within five years. And I think it will ramp faster on primary storage than it did in the backup market.”

Which sort of leads into the announcement that Ocarina made today. In a strategy similar to Permabit’s, Ocarina announced a software-only, ‘embeddable’ version of its ECOsystem data reduction technology targeted at OEMs. (Ocarina’s existing data reduction technology is delivered as a software-hardware appliance.)

Ocarina may not be a household name, but note that its partners/resellers include, among others, HP, EMC, Hitachi Data Systems and BlueArc, although Ocarina didn’t make any OEM announcements for its embeddable software today. (It takes a long time to qualify and integrate technology like this into a vendor’s existing stack.)

One differentiator that Ocarina may have is a content-aware data reduction technology with a combination of data compression and deduplication. In addition, Ocarina CEO Thirumale says that the OEM version of the company’s technology can be applied anywhere in the IT infrastructure – from primary storage devices to archiving platforms – and on any type of data (e.g., files or blocks), enabling data to remain in compressed/deduped form throughout its lifecycle. In addition, the software can be implemented in a post-process, in-band or hybrid approach.

In other data deduplication news today, GreenBytes announced its entry into the European market with a variety of reseller partnerships. GreenBytes packages its data deduplication software in inline GB-X appliances that combine solid-state disk (SSD) drives and 2.5-inch SATA drives in what the company refers to as a Hybrid Storage Architecture (HSA). GreenBytes’ block-level deduplication can be used in NAS or SAN environments.

Friday, June 18, 2010

"SSD week" recap

June 18, 2010 – I don’t think it’s purely coincidental, but there was an uncharacteristic amount of rapid-fire announcements related to solid-state disk (SSD) drives this week. Three SSD startups launched (with two based in Israel). There was one SSD-related acquisition. And one vendor trumpeted the now familiar tune, “MLC SSDs are as good as SLC SSDs.”

In short, four interesting SSD announcements in the course of three days. And all four articles covering those announcements made the Most Popular Articles list on this week. The industry analysts must be right about increased end-user interest in SSDs, although 39% of visitors have no plans for using SSDs because they’re still too expensive (see below).

A quick recap:

At the high end of the announcements, Kaminario launched with the introduction of the K2 (meant to conjure up an image of the mountain, not the skis). See “Startup claims 1.5 million IOPS on RAID array.” An entry-level configuration starts at $200,000 for 300,000 IOPS and 1TB of DRAM, but the system can scale to millions of dollars and IOPS. Since its system is based DRAM technology, Kaminario will compete most directly with vendors such as Texas Memory Systems (which has been selling DRAM devices since disco was popular) and Violin Memory.

Speaking of which, Violin bought Gear6 this week. You might (but probably not) know Gear6 as a supplier of a souped-up Memcached distribution, but that’s not why Violin plucked them. Gear6 also has NFS caching software. If Violin can orchestrate that code into its SSD arrays it will give Violin an entry into the lucrative NAS acceleration market (see “Violin to acquire Gear6 for caching software”).

Anobit (add another bit) was another SSD startup that launched this week, with an MLC-based array that the company says provides the performance and reliability (“endurance” in SSD parlance) of more expensive SLC SSDs. You’ll see a lot of similar claims in the months ahead, as the mathematicians at SSD vendors have come up with a variety of tricks to mask the inherent drawbacks of inexpensive MLC flash NAND technology. Anobit’s secret sauce is its Memory Signal Processing technology (see “Anobit claims MLC SSDs rival SLC SSDs”).

And on the PCIe front, Virident launched its tachIOn SSDs. (Weirdly enough, when you type “tachIOn” in Word it automatically gets changed to “tachyon” which, by the way, is a hypothetical subatomic particle that moves faster than light, although most of us old-timers think of it as a Fibre Channel controller chip). Virident is going head-to-head with Fusion-io, essentially trying to beat them at their own game (see "Virident ships SLC SSDs for PCIe").

Despite the rapidly falling prices of flash memory and SSDs, many end users still say SSDs are too expensive. In an QuickVote survey conducted earlier this year, 24% of the respondents had planned to deploy enterprise-class SSDs in the first half of this year; another 26% planned to buy them in the second half of the year; but 49% had no plans to deploy SSDs, with 10% citing insufficient reliability and 39% saying SSDs were still too expensive.

If you’re really interested in SSDs, consider attending the Flash Memory Summit, August 17 – 19 in Santa Clara.

Monday, June 14, 2010

IBM to acquire Storwize for $140 million?

June 14, 2010 (Updated June 15) – Rumors swirled this week around speculation that IBM is in talks to acquire startup Storwize, which specializes in data compression/reduction for primary storage, for a reported $140 million. I wasn’t able to confirm the rumor, but I have no doubt it’s true (the talking part, although not necessarily the acquisition part).

The potential deal makes sense for a number of reasons:

One: Data reduction for primary storage is going to take off big time this year. Heavyweights EMC and, more so, NetApp have capitalized on this trend with data deduplication techniques for data reduction. IBM has data deduplication for secondary storage via its acquisition of Diligent Technologies, but has been out of the picture when it comes to data reduction for primary storage. And buying Storwize (already an IBM partner) will get Big Blue into the game faster than developing the technology on its own.

Although it’s based on compression, not data deduplication, Storwize’s technology (Random Access Compression Engine, or RACE) can be used in conjunction with data deduplication products such as those from EMC and NetApp.

Two: IBM has plenty of experience acquiring Israeli companies. In addition to Diligent, IBM acquired Israel-based XIV in 2008. (Storwize’s R&D facility is in Israel, although the company now has U.S. headquarters.)

Three: IBM has already been reselling Storwize’s STN appliances, and has jointly presented Webcasts with Storwize and produced white papers legitimizing Storwize’s various claims, including the claim that its compression appliances can reduce capacity by 50% to 90% without negatively affecting performance.

Four: It hasn’t been a secret that Storwize has been shopping itself for some time. Getting acquired makes more sense than trying to go it alone in the primary storage optimization market, which will be ruled by the big boys – not startups (just my opinion, but I’m right.)

This acquisition would be interesting on a number of fronts.

--Would it lead to HP acquiring partner Ocarina Networks, a Storwize competitor?

--And let's not forget about other players in the primary storage optimization space. Exar (via its acquisition of Hifn) has data reduction technology that combines data deduplication and compression and handles those functions at the ASIC level. And GreenBytes has inline deduplication ttechnology that can be applied to NAS or SAN environments.

--Could IBM’s offer lead to a bidding war? Perhaps between Big Blue, EMC, NetApp and/or HP?

Further fueling the flames in this market was Permabit’s announcement of its Albireo technology last week (see “Permabit deduplicates primary storage”). Albireo is a software-only deduplication technology designed specifically for primary storage, and the company claims to have a number of design wins, with at least one rumored to be with a big vendor. Who could that be?

This could get even more interesting than last year’s NetApp-EMC bidding war for Data Domain.

Related articles:
Data reduction for primary storage: Benefits and options
Storwize upgrades compression appliances
Infineta unveils deduplication for inter-data-center traffic
GreenByes launches entry-level dedupe appliance
Nexenta debuts ZFS-based in-line deduplication

Startup claims 1.5 million IOPS on RAID array

June 14, 2010 – Not so long ago, performance claims of one million IOPS seemed impressive (and suspect, and irrelevant). But 1.5 million IOPS?!

Startup Kaminario launched today with the introduction of its aptly-named K2 storage system. An entry-level configuration is spec’d at 300,000 IOPS and a high-end configuration delivers 1.5 million IOPS, or 16GBps of throughput, according to company officials.

There’s only one way to get that level of performance, and it ain’t short-stroking. In fact, it’s not even SLC NAND flash SSDs. You have to go with the fastest, most expensive type of storage device – DRAM.

Kaminario’s K2 RAID array is officially dubbed the K2 Performance Enterprise Data Appliance. The blade-based, scale-out architecture has two key building blocks: ioDirectors and Data Nodes.

The ioDirectors interface with hosts via 8Gbps Fibre Channel ports and, together with kOS software, are responsible for directing all read/write I/O requests. The Data Nodes store all data in DRAM (up to 288GB per node), although each node also includes two mirrored SAS disk drives, which are used only for data backup. The Data Nodes appear as a single logical unit. Users scale performance by adding ioDirectors.

The entry-level configuration that provides 300,000 IOPS and 3.2GB of throughput includes two ioDirectors. The 1.5-million IOPS configuration includes eight ioDirectors. The system has equal performance for both random and sequential I/O operations, according to Kaminario product manager Arik Kol.

Not surprisingly, all that performance comes at a steep price. The entry-level configuration (300,000 IOPS) with 1TB of DRAM is priced at $200,000. Yes, $200,000 per TB.

However, why go with a startup? Kaminario isn’t the first company to build super-fast storage systems with DRAM. Texas Memory Systems, for example, has been selling DRAM SSDs for decades.

If you’re in the market for systems of this caliber and cost, I’m sure you have the wherewithal to do in-depth evaluations, and there are many factors to consider, but let’s take a cursory look at some performance-price-capacity comparisons:

Kaminario is charging $200,000 for 1TB of DRAM capacity and 300,000 IOPS of performance.
You can buy two Texas Memory RamSan-440 DRAM arrays with 1TB of capacity and 1.2 million IOPS for $280,000. That’s 4X the performance for an extra $80k.

Kaminario isn’t the only vendor making solid-state memory news this week. Stay tuned to’s News & Analysis section on Tuesday for info about two startups with interesting SSD plays – one with an SLC-based PCIe device, and another with an MLC-based SSD that the company claims rivals SLC SSDs in endurance – at a fraction of the cost.

Friday, June 11, 2010

Storage software market: CommVault grows revenues 25% y-o-y

June 11, 2010 – Revenue performance in the storage software market was lackluster in the first quarter of this year, although the market did exceed $3 billion – a 7.2% growth over the same quarter last year. Given the sorry state of IT spending a year ago, that qualifies as lackluster.

In contrast, the external disk array market posted year-over-year growth of 18.8% in the first quarter (see “Disk array SmackDown: NetApp vs. IBM, Dell vs. HP”).

With a few exceptions, storage software vendors posted modest revenue growth in Q1 2010 vs. Q1 2009: EMC (13.7% revenue growth), IBM (11%), NetApp (8.1%), CA (7.2%) and HP (8.7%). The only vendor posting negative growth was Symantec, with a slight 0.5% drop.

EMC topped the list with a 23% market share on Q1 revenue of $696 million, followed by Symantec (17.5% market share), IBM (14.2%), NetApp (8.3%), and CA and HP in a virtual tie with 3.9% and 3.5% market shares, respectively.

But the most impressive numbers came in the data protection and recovery segment of the overall storage software market, where EMC racked up revenue growth of 46.6% and CommVault, although it didn’t crack the top six vendors list, increased revenue by an impressive 24.9% in the first quarter vs. the same period a year ago.

For IDC’s full press release and more stats, see “Storage Software Market Delivers Good Growth and Continues Signs of Recovery.”

Wednesday, June 9, 2010

Disk array SmackDown: NetApp vs. IBM, Dell vs. HP

June 9, 2010 – In the external disk array market, it’s surprising how little changes in terms of the top five vendors’ market shares quarter-to-quarter or even year-to-year. The leading vendors almost always occupy the same rungs on the ladder. However, based on IDC’s Q1 2010 stats, market shares appear to be shifting.

Of course, EMC held on to its #1 spot in the first quarter with a 24.6% market share on revenue of $1.22 billion. But the rest of the race is tightening.

For example, for the first time NetApp pulled into a virtual dead heat with IBM for the #2 spot in the external disk systems market. IBM had an 11.7% market share, while NetApp had an 11.1% share. Statistically, IDC considers that to be a tie. Compared to Q1 2009, IBM slipped 0.5% in terms of market share while NetApp gained 2.3 points, leapfrogging both Dell and HP, which were in the #3 and #4 spots in the previous quarter.

If NetApp’s ascendancy continues at this clip, the company could eclipse Big Blue. And if NetApp’s first quarter and year-end earnings are any indication (see “NetApp wows Wall Street, doubles quarterly profits”), I think they will.

Rounding out the top five, HP slipped a bit in the first quarter, to a 10.2% market share, followed closely by Dell with a 10.1% share. Dead heat. And I predict that Dell will overtake HP in the next quarter because the revenue differential between the two was only $6 million in the first quarter ($506 million vs. $500 million).

Hitachi Data Systems and Oracle/Sun failed to crack the top five.

NetApp also gained market share in the NAS arena over the last quarter with a 26.9% share (vs. 20.2% in the last quarter), coming in second to EMC’s 45.1% market share (down from a 50.5% chunk in the last quarter).

Overall, the NAS market grew a whopping 44.6% year-over-year.

Equally impressive, the iSCSI SAN market posted 45.7% revenue growth in Q1 2010 vs. Q1 2009. Dell led the iSCSI market with a 36.9% revenue share, followed by NetApp with a 14.4% slice.

For IDC’s press release and more stats, see “Disk Storage Systems Market Rebounds to Double-Digit Growth Across All Segments in First Quarter.”

Friday, June 4, 2010

Cloud storage and unified computing: Beware vendor lock-in

June 4, 2010 – Vendor lock-in is always a bad thing, but in the case of cloud computing, cloud storage and so-called unified computing, it could be much worse.

Vendor lock-in with, for example, RAID arrays or backup software is a mild form of lock-in. You’re not exactly shackled and chained in an iron-clad tank underwater. Although it might be painful, time-consuming and costly, IT shops migrate data from Vendor A’s disk array to Vendor B’s disk array all the time.

Migrating from one cloud provider (whether they’re providing cloud computing or cloud storage services) to another may be much more difficult. There are options, however. Some vendors offer tools for migrating from one cloud provider to another. Nasuni is one example, and here’s the Nasuni team’s (not exactly commercial-free) blog on this subject: “The Cloud’s Little Secret: Vendor Lock-In.”

And the Storage Networking Industry Association (SNIA) is working on standards that promise portability between cloud storage providers (see “SNIA develops standards for cloud storage.”) However, it’s unclear whether SNIA’s standards will succeed and whether the cloud market share leaders will support those standards.

But these potential solutions to cloud storage lock-in beg the question: Why would you want to move from one cloud storage provider to another?

Nobody’s rushing to put their production data in a public cloud; rather, they’re contemplating putting their backup/secondary/archive data in the cloud. And if cloud storage providers meet their SLAs (and that’s a good bet with the large providers because those SLAs are relatively non-demanding) and they keep their prices in the pennies-per-GB range (which they will due to intense competition in this space), then it would be an extremely rare case where you would have any need to switch providers.

So the whole issue of migrating between cloud storage providers (which apparently has become a hot topic in the cloud community) seems to be a moot point, at least until companies start putting production data in the cloud.

Now, vendor lock-in in the context of unified computing is a different matter. If unified computing takes off, lock-in will not just be something to avoid; it will be a way of life in IT land.

But I think it’s quite possible that unified computing may backfire on the vendors that are pushing it. In an age of unified computing lock-in, IT shops may be more inclined toward non-unified-computing vendors. That means the “other” vendors in the market share pie charts.

The potential for vendor lock-in in unified computing may drive IT away from their existing vendors. Instead of cementing the market leaders’ positions, unified computing could finally be the trend that works against the big guys.

Maybe I should have titled this post “Lock, stack and barrel.”

Anyway, here’s some more info on unified computing and the potential of vendor lock-in: “Weighing the pros and cons of unified computing.”