The Pervasive Data Center

There's been a lot of speculation that Oracle purchased Sun for its software assets like Java, Solaris, and--although this point has seen more debate--MySQL. Even those of us who viewed the acquisition as a serious play by Oracle to become a full-fledged system vendor figured those systems would be mostly x86. That's not to say Oracle would kill SPARC processor development and servers outright--the installed base is too large and profitable--but it would be a business to milk, not to invest in.

However, Oracle CEO Larry Ellison, writing in an e-mail interview with Reuters, claims to have big plans for Sun's server business--including its in-house processor design capabilities.

Ellison begins by stating that "we are definitely not going to exit the hardware business." It doesn't get much more definitive than that as to Oracle's overall strategy of being a systems company.

What Ellison has in mind here is integration. He goes on to write that:

While most hardware businesses are low-margin, companies like Apple and Cisco enjoy very high-margins because they do a good job of designing their hardware and software to work together. If a company designs both hardware and software, it can build much better systems than if they only design the software. That's why Apple's iPhone is so much better than Microsoft phones.

Those are fair points. And Oracle has itself experimented with hardware/software integration such as the Exadata Storage Server that uses HP hardware.

At the same time, the idea that you can be in the server business and only sell into the profitable niches strikes me as a notion that Oracle may not want to depend upon too much. (Cisco has made similar statements with respect to its Unified Computing System.) The history of the system vendor business going back at least a decade suggests that the most successful companies have supply chains and partner networks that allow them to sell pallets of small servers in addition to a smaller number of highly profitable large ones.

Ellison then goes on to make it equally clear that he's not interested in just bundling software and hardware but deeply optimizing the hardware when he writes: "Once we own Sun we're going to increase the investment in SPARC. We think designing our own chips is very, very important... Right now, SPARC chips do some things better than Intel chips and vice-versa."

By way of background, Sun's CMT SPARC chips are designed around a philosophy of handling many tasks in parallel even if it means that individual tasks may run somewhat slower than on a chip with fewer but more powerful cores. This approach lends itself well to workloads that involve a lot of relatively independent activities--such as Web and application servers. It also lends itself to very power-efficient designs.

But Ellison isn't just arguing that SPARC is good for some things and x86 is good for others. He's arguing for hardware that is truly optimized for Oracle software.

Some system features work much better if they are implemented in silicon rather than software. Once we own Sun, we'll be able to plan and synchronize new features from silicon to software, just like IBM and the other big system suppliers. We want to work with Fujitsu to design advanced features into the SPARC microprocessor aimed at improving Oracle database performance.

There remain plenty of questions about how large Oracle's investments will be and how much it will tilt toward its own processor-server-operating system-middleware-applications stack. It will, of course, continue to sell software to run on HP, IBM, Dell, or wherever else it can garner license revenue from.

However, on the face of it Oracle has grand visions for its Sun acquisition that go well beyond selectively mining some key software assets and milking the rest. Oracle's purchase of Sun was the latest example of the general shift back to a more vertically-integrated computer industry going on. This latest interview with Ellison makes that point again--with exclamation points.

One of the favorite water-cooler games of the enterprise computing set over the past month or so has been, "Whither Sun Microsystems?" Now the first phase of that game is over. The answer, as of this morning, is: Oracle--subject to the usual approvals, of course.

But the next phase of the the game is just beginning.

It's obviously very early in the process but here are a few initial reactions:

At first blush, this acquisition may seem odd. Oracle is a software company. Sun tends to be viewed as a hardware company. Why would Oracle CEO Larry Ellison want to get into the hardware business? That's the standard "Huh???" about this purchase. But this misses a number of important points.

Sun is not a hardware company. It is a systems company. And, in fact, Sun has steadily ramped up its software business in recent years. Sun Solaris and Java were instrumental in Oracle's decision to acquire Sun. So this isn't really a software company buying a hardware one.

To get the bigger picture here you have to view it in the context of what's going on within the system vendor landscape more broadly. At the risk of overstating things, the system vendor landscape is being reconstituted into big, highly integrated companies that can do it all.

This is how essentially all computer companies used to be, but that way of business gave way to the horizontal industry structure epitomized by the likes of Microsoft and Intel.

Oracle has poked at this sort of thing before. Unbreakable Linux and in-house virtualization work were early efforts. But the purchase of Sun lets Oracle take this to the next level. Consider these sound bites from the press conference: "Tightly integrate the Oracle database to some of the unique high-end features of Solaris," Sun's operating system; "for the first time deliver complete integrated computer systems, applications to disk;" and deliver "complete industry-in-a-box."

This is not to say that Oracle may not divest or shutter segments of Sun's portfolio that don't post the right kind of financial return. But this looks to me like a very serious play to vertically integrate. With their applications portfolio, it's actually a more vertical integration than even IBM offers directly, for the most part. (IBM does have some industry-specific solutions but not at the same scale as Oracle Financials and Manufacturing.)

If there were any doubt that the pendulum is in full swing back to large, integrated systems companies, this should erase it. We had IBM and Hewlett-Packard (most recently with its EDS acquisition). Now we have Oracle. And Cisco Systems is easing over that way.

I just ran across this video from Sun Microsystems that demonstrates a source of disk latency that we usually don't think about. It's short and quite amusing. Take a look.

But being the analyst that I am, I also wanted to highlight some of the pieces that make the demonstration possible.

The technology that's under the covers, measuring the disk latency and so forth, is DTrace, a component of Solaris, originally developed by Bryan Cantrill. By way of background, here's what I wrote upon its introduction in Solaris 10:

DTrace stands for "Dynamic Tracing" and, indeed, it's that dynamism that most distinguishes it from other approaches. A developer, administrator, or performance tuner uses the DTrace scripting language to dynamically establish monitoring points of interest, whether in the OS kernel or user processes.

A probe is a location or activity to which DTrace can bind a request to perform a set of actions, like recording a stack trace, a time stamp, or a function argument. Think of them as programmable software sensors that gather interesting information about the system, and report it.

DTrace in Solaris 10 comes with something like 37,000 predefined probes; users can also define their own. Probes come from a variety of kernel modules that Sun calls providers, each of which creates a unique type of instrumentation.

For example, there's a provider that creates a simple time-based counter (profile) and another to understand lock contention and other sorts of locking behavior (lockstat). But essentially any function entry or exit is a potential probe location; DTrace hot-patches the function entry point in memory to insert the probe.

DTrace is an incredibly powerful tool in that it can do all this monitoring and measuring while minimizing its impact on a running system. (SystemTap is a Linux analog that's considerably less mature.) This allows DTrace to be used in conjunction with production systems--and thereby look for performance bottlenecks under real loads rather than synthetic test cases. The downside of DTrace is that you have to be a bit of a Solaris performance guru to actually make effective use of it.

Enter Fishworks, which Bryan calls DTrace-based appliance analytics.

With analytics, we sought to harness the great power of DTrace: its ability to answer ad hoc questions that are phrased in terms of the system's abstractions instead of its implementation. We saw an acute need for this in network storage, where even market-leading products cannot answer the most basic of questions: "what am I serving and to whom?" The key, of course, was to capture the strength of DTrace visually--and the trick was to give up enough of the arbitrary latitude of DTrace to allow for strictly visual interactions without giving up so much as to unnecessarily limit the power of the facility.

You can also find a more detailed overview of the analytics in this Sun presentation (PDF).

Fishworks is a big part of the secret sauce and value-add of Sun's recently introduced Storage 7000 Unified Storage Systems. It also reflects Sun's new, or at least "tweaked," software strategy. While DTrace is open source (under the CDDL license) along with the rest of OpenSolaris, the GUI dashboard, and other analytics software that makes use of DTrace, is not. This is consistent with a more pragmatic approach to open source on the part of Sun that allows for keeping proprietary modules and components mostly aimed at large-scale production deployments.

Fishworks is one of those--and an impressive one at that. It will know if you shout at your disks!

One of the interesting threads within this year's O'Reilly Open Source Conference (OSCON) was a variety of collaboration tools and platforms aimed at what might be called "mid-weight" collaboration. Which is to say, collaboration that is something more than mailing lists or user forums but something less than the code repositories that primarily target a core group of developers.

Here's the basic issue. Some popular projects, such as the Linux kernel, have a pretty broad range of contributors. However, many other projects--even successful ones--are the product of a much more constrained group of people. With open-source software specifically, a big part of the problem is that it usually takes considerable effort even for experienced developers to understand a code base well enough to make useful contributions, And, of course, not many people are experienced developers.

Even beyond software development, as shown in the Gartner Group graphic in this post, only a small minority of any community is highly active. To use Gartner's terminology, "Contributors" and "Opportunists" significantly outnumber "Creators." (Unsurprisingly, the largest group are "Lurkers," which is to say basically users.)

Thus, the challenge is to harness the diffuse energy of a community in addition to the energy concentrated in a relatively small number of core individuals--the "Long Tail" of content creation if you would.

Here are a few examples of the work in progress.

JasperForge is perhaps the project that's most explicitly about harnessing the contributions of nondevelopers. The community here are those associated with Jaspersoft (Open Source business intelligence) and related projects. (There are currently 227 public projects and 70 private projects.) Individual projects setup a portal that includes only the components they need. For example, a documentation project could have a wiki and shared files but not a code repository.

Launchpad is the creation of Canonical, the commercial entity behind Ubuntu. It recently got a major redesign for its 2.0 release. For our purposes here, one novel feature is a Web-based translation and review component that allows contributors to provide translations of strings in the program into any of up to 265 languages. It also includes a community support component that lets users create FAQs and build a searchable knowledge base of previous answers.

Finally, Sun's Zembly--currently in private beta--is a collaborative tool for building applications for social media, such as Facebook. A lot of these applications are pretty small and simple and they often lend themselves to reusing code from other similar applications. For example, if you have a widget that displays a group of photos from Flickr, another widget that displays a different set of photos, but in a similar way, can pretty much just cut and paste a lot of the first widget's code. In short, it's oriented for the lightweight code sharing and reuse that characterizes many of this type of application.

The need for sophisticated version control and code repository systems isn't going away. However, we're starting to see those systems complemented by new techniques that either layer tools for more casual contributors on top or that explicitly target lighter-weight collaboration from the get go.

Earlier this week, Sun Microsystems launched a family of new servers based on the SPARC64 VII processor. In contrast to Sun's "CMT" (Chip Multithreading) UltraSPARC T1 and T2 designs that deliver aggregate performance using a large number of threads, SPARC64 takes a more conventional approach that is more rooted in parallelism and performance at the level of a single thread. This design is more attuned with the performance requirements of typical enterprise back-end applications and databases, whereas CMT has more of a network-facing orientation.

SPARC64 comes from Sun's partner Fujitsu, which also designs and builds the midrange and high-end servers that use the chip; these systems went by the "APL" codename while they were under development. Fujitsu and Sun jointly sell these servers--as well as the CMT "Niagara' boxes for which Sun does the processor and server development.

The new processor and servers are solid upgrades. Although not as multi-threaded as Niagara, the SPARC64 VII bumps the number of cores per chip to four, and adds the ability to run two threads on each of those cores--a technique that helps mask delays associated with waiting for data to arrive from memory. Frequency is also up from the prior generation to 2.4 GHz and 2.52 GHz.

Sun pegs the performance boost over the prior generation at up to about 80 percent for commercial applications, and up to 2x on apps that are floating point-intensive. That's a nice increment, considering that upgrades from the SPARC64 VI servers require only CPU board upgrades. While I find that vendors often overplay the issues associated with competitors' "forklift" hardware upgrades and other supposed gotchas, there's no doubt that less is more when it comes to making infrastructure changes.

Overall, there's little to fault in this announcement from a product perspective. It's a solid, nondisruptive bump to a product line that--although Sun doesn't break out numbers--must contribute a substantial chunk of its server revenue.

My critique instead relates to how Sun (again) seemed almost bored by this announcement. Yes, there was a press release--it wasn't exactly a stealth launch--but there was certainly none of the mass marketing air cover that Sun (for better or worse) is wont to darken the skies with when it comes to something that it's genuinely excited about. No blog postings from its pony-tailed Blogger-in-Chief. No glitzy roll-out.

Don't get me wrong, many of the things that get Sun's corporate blood flowing such as open storage, OpenSolaris, Project BlackBox, ZFS and solid state disk, and Niagara are genuinely exciting. But many are also speculative. It would behoove Sun to at least make the old college try to display some comparable enthusiasm about products that are proven and bringing in real revenues.

Niagara 2 (formally the UltraSparc T2) was a big step forward for Sun Microsystems' chip multithreading (CMT) efforts. It's not that there was anything really wrong with its Niagara 1 predecessor, but 90-nanometer process technology imposed some fairly severe restrictions on what could be crammed into each of the eight cores. As a result, Niagara 1 was well-suited for a relatively narrow range of network-facing workloads. Niagara 2, by contrast, was able to leverage 65nm process technology to spread its wings considerably--which it did by significantly beefing up the threading, floating point, and other capabilities of the individual cores, in addition to adding on-chip I/O. (Our full report: Niagara 2: More Heft in the Weft.)

Now Sun has rolled out its promised dual-socket version of the UltraSparc T2--aka "Victoria Falls" or the UltraSparc T2 Plus--in 1U (Sun Sparc Enterprise T5140) and 2U (T5240) server flavors. In essence, it replaces the two on-chip 10-gigabit Ethernet ports with four coherence channels that tie together two UltraSparc T2 Plus chips into a single SMP system. So you get twice as many cores and threads, and about twice the processing power. Sun has also tweaked other server capacities a bit higher. For example, the 2U box now supports up to 128Gb of FB-DIMM memory (using 32 DIMMs and a new memory riser card) and up to 16 disk drives.

Most everything else remains unchanged from the UltraSparc T2-based servers. As before, there's embedded cryptography, a floating point unit for each core, eight threads per core, redundant hot-swap power supplies and fans, and integrated virtualization (LDOMs).

The big thing that the new servers bring is, not surprisingly, performance. While not all workloads scale with core count, those that these servers target mostly do. After all, if an application doesn't do well in a multithreaded environment, it's not a very good candidate for Sun's CMT line in general. And Sun has released a nice passel of benchmark results to back up its performance, price/performance, and performance per watt claims. Interpreting benchmark claims in as apples-to-apples a manner as possible is always a tricky undertaking (and it's not like vendors make the task any easier), but Sun's overall numbers look strong. For example, the T5240 outperforms a quad-socket Xeon system on a two-tier SAP Standard Application SD benchmark, and turns in a SPECjbb result over 2x another 4-socket x86 system. While one can always quibble with individual comparisons (and we frequently do), the overall picture is an impressive one.

The only notable tradeoff is that the T2 Plus systems move networking from the chip to the board, "slightly lowering" (in Sun's words) networking performance. Networking should otherwise work identically; this includes the virtualization-savvy packet classification and routing that falls under the overall "Project Crossbow" architecture. All things considered, this is a reasonable tradeoff for what has become a more general-purpose system.

For utility applications and network-facing systems, much of the IT universe has long since adopted distributed x86 servers as their platforms of choice--especially as you move out of mission-critical datacenter backends. That's the big challenge faced by Sun with the Niagara line (and indeed anyone selling anything other than scale-out x86). But these are solid systems that can handle big loads with aplomb.

Writing in Computerworld, Eric Lai notes that:

Despite the popularity of .Net within companies and other employers, Microsoft has seen its standing among students continue to be eroded by a combination of open-source programming tools and Adobe Systems Inc.'s Web design software. Now, after years of using half-measures to try to beat those technologies on college campuses, Microsoft is taking a bolder step by making four pillars of the .Net platform available free of charge to tens of millions of students in the U.S., Canada, China and eight European countries.

A few observations here.

If you're a follower of Sun and Solaris as I have been for many years, there's a familiar thread here. A major cause of Sun's financial problems--which the company is still working to put behind it--was that it lost a goodly chunk of the core developer constituency that gave it market relevance. And, in CEO Jonathan Schwartz' words: "To establish a high-integrity relationship with a broad and participative community is really the principal objective of bringing Solaris into the open-source world." Yes, there are still many developers for Windows and other Microsoft software platforms, but it often seems a dutiful and passionless crowd.

The analogy between Microsoft/Windows and Sun/Solaris is not a perfect one. Perhaps the most notable distinction is that Microsoft has a broad presence in both consumer and SMB markets that Sun did not (and does not). The inertia this provides insulates Microsoft to at least some degree from the shifting breezes of developer fashion. Nonetheless, when you add in that Microsoft also has to contend with the shift of computing into the network cloud (and the corresponding diminution of Microsoft's incumbent advantages that implies), a weakening connection to developers can't be viewed as anything but bad.

Finally, while giving away software may well be a reasonable step for Microsoft to take, it's hardly a sufficient strategy to counter the rise in Open Source (and programming to application programming interfaces in a Web 2.0 or Software as a Service context). Yes, students like free. Who doesn't? But they also want to work on projects that they consider interesting, relevant, and--yes--cool. And that has very little to do with Microsoft making products available under an academic license.

As with other companies serious about being viewed as virtualization players, Sun has found that one size cannot be made to fit all. Thus, Sun's early religion around the operating system containers built into Solaris 10 has given way to more of a toolbox approach that most notably also includes its Xen-based xVM Server.

Thus, Sun's most recent addition--the Open Source VirtualBox (by way of acquiring Innotek) isn't particularly surprising. But it makes a nice addition all the same.

VirtualBox is a "Type 2" hypervisor product, which is to say that it runs on top of an existing operating system rather than directly on top of the underlying hardware as does a native hypervisor. Thus, from a technical perspective, it has similarities to VMware Server and Microsoft's Virtual Server, which are also hosted virtualization products. The disadvantage of this approach is that there's more overhead involved--especially for I/O-intensive workloads. However, in exchange for this overhead, the hypervisor gets to piggyback on all the device drivers and other interfaces already built into the OS. Hosted virtualization products can thereby support whatever peripherals and I/O cards are supported by the native operating system right out of the box.

Some other facts and figures about VirtualBox and Innotek:

• It's about a 17 MB download from virtualbox.org; Sun reports four million downloads since January 2007
• It will run on top of Windows, Linux, Mac, and Solaris host operating systems
• It will support guest operating systems that include all versions of Windows from 3.1 to Vista, Linux 2.2, 2.4 and 2.6 kernels, Solaris x86, OS/2, Netware and DOS
• The financial details are as follows: "The stock purchase agreement to acquire Innotek is subject to customary closing conditions and is expected to be completed during the third quarter of Sun's 2008 fiscal year. The terms of the deal were not disclosed as the transaction is immaterial to Sun's earnings per share."

This acquisition is a positive move for Sun. One thing that Sun's virtualization portfolio was missing was a true developer on-ramp that offers a no-cost mechanism to "play around" with (specially) Solaris on an existing laptop or workstation. Like the MySQL acquisition and aspects of "Project Indiana," this is all about providing easy entry path into the Sun ecosystem that developers want to head down.

Based on our very preliminary tire-kicking, the product still needs some work. For example, when my colleague Jonathan Eunice tried to load a VMDK (VMware VM disk image) for the latest Solaris Express Developer Edition (SXDE) using VirtualBox, the VM just endlessly rebooted. Senior Sun technical people took a close look at VirtualBox before the acquisition and liked what they saw. I've also started to hear positive buzz online about VirtualBox of late. So I have little doubt that Sun get get VirtualBox up to full snuff. But it will take some development.

It's a different approach from VMware. In VMware's case, it's their original server product (VMware Server, nee GSX Server) that's free; VMware Workstation still carries a charge. To be sure, in practice, lots of people try out VMware on their desktops using VMware Server. Nonetheless, we see here a difference in philosophy. Sun, especially CEO Jonathan Schwartz, has had a mantra over the past couple of years about how "developers don't buy things; they join things." In fact, the roots of this thinking at Sun go back almost a decade; Sun Open Sourced Netbeans in 2000--before Eclipse came on the scene.

VirtualBox also provides an on-ramp for other Sun properties. When Sun pre-briefed us on this announcement, they made considerable ado about using pre-packaged virtual appliances to get developers up and running quickly with products like GlassFish, the Open Source application server that Sun is backing. Virtual appliances are a promising idea--although in practice they've been laggardly. If Sun truly wants to leverage the appliance idea here, it will have to get aggressive in kickstarting a library of up-to-date instances.

In short, this is very much part and parcel of what is probably Sun's single greatest thrust--to reconnect with the development community that it once essentially owned but who abandoned it en masse for Linux. VirtualBox isn't a heavy gun in this effort like MySQL is. But it's a nice little bodkin.

Over on The Open Road, Matt Asay analyzes the price paid for three open-source companies: MySQL (bought by Sun Microsystems earlier this week), JBoss (Red Hat), and Zimbra (Yahoo). He concludes that depending upon the revenue assumptions, whether you use trailing or forward-looking revenue numbers, and whether one looks at bookings rather than revenue, the valuations for all three were somewhere in the 15 to 20 times annual revenue range.

This is a big multiple. By contrast, Oracle is paying a multiple of about 4 times for BEA Systems--and some analysts are saying that's too high.

So is this just another bubble in which companies that are considered in the forefront of the Web or open source or whatever get snatched up for unjustifiable sums?

It is true that all of these companies could be considered category leaders. It's clearly so in the case of MySQL (open-source database) and JBoss (open-source application server), so some premium might reasonably attach to their post position. Yet, one would think clear leadership would already be reflected in their revenue numbers, so that can't be the whole story. Is there any other explanation--especially one that doesn't require irrational exuberance?

I think so. As I wrote in the context of Sun's acquisition of MySQL a few days ago, it's hard for standalone, narrowly focused open-source companies to profit. A financial analyst on the Sun/MySQL call estimated that MySQL had annual revenues of $60 million to $80 million in 2007 and operated at about breakeven. Not bad, but considering that MySQL is widely regarded as one of the true open-source success stories, it's hard to view those financial results as better than modest. At issue is that even with an enterprise version and value-add services--in addition to basic support--MySQL converts a small proportion of users into paying customers. That might be OK, but even when it does monetize users, it's pretty much limited to selling them a subscription for its enterprise version--which is still a great bargain by historical proprietary database standards.

However, plug MySQL or some other open-source company into a larger organization and the opportunities increase enormously. In the case of Sun, each MySQL customer that is willing to pay for the Enterprise database is now also a potential customer for Sun professional services, servers, and other software. The same logic applies to JBoss and Zimbra with their respective owners although those paths to incremental monetization may be less clear--and, indeed, Red Hat has publicly admitted that, so far, it hasn't leveraged its JBoss acquisition as well as it might have.

Although there are any number of small and profitable independent software vendors (ISV) in the proprietary software world, small software companies get gobbled up by larger vendors all the time of course. There's often more value in integrated offerings than in point products. Enterprise are also more comfortable sourcing some types of products, such as high-level management tools, from large ISVs or system vendors.

But over and above all the reasons why it's hard to make a profit as a standalone ISV, a look at the market suggests that it's even harder in some ways for standalone open-source ISVs. It's not that their product is any less valuable and it's certainly not less desired. But it's hard to monetize in a standalone way.

That could well be a reason for these high valuations. The value is already there but it takes a larger and more diverse organization to supply the leverage that makes money off that value.