Sun Set . . .Or Sun Rise?

Sun Set…Or Sun Rise? by Charlie Martin

What to do with Sun Microsystems?

It’s not an easy question: Mike Malone, here at Edgelings, has suggested that Sun should just shut down.  Liquidate the assets, distribute the proceeds to the shareholders, and tell CEO Jonathan Schwartz to turn out the lights on his way out.

Since then, Sun’s stock has gone up a surprising 40 percent. It turns out that a Memphis-based investment firm, Southeast Asset Management, has bought significantly into the company — so that it now owns about 22 percent of Sun’s shares.  Just this week Southeast was rewarded with two seats on Sun’s Board of Directors.  

The normal interpretation of that kind of deal — with Sun’s book value being almost twice it’s market capitalization — would be that of a corporate raider prepared to break up the company and sell it piecemeal for a quick profit; the Wall Street version of euthanasia.  Southeast isn’t known for that, though: it’s a “value investment” firm that likes to buy undervalued businesses.  It certainly talks about Sun as a going concern, just one that has gone astray.

So, just as a thought experiment, let’s consider what Sun could do to become a strong and profitable company once again.  

First, let’s look at what Sun’s strengths really are.  For one thing, the company has some  astonishing technical assets in Solaris, MySQL, and a less-definite relationship with PostgreSQL.  And it has the entry into the storage market it acquired with StorageTek.  (Note that I didn’t include Java in that list.  More on that below.)

Software

Looking at the big picture, there are three empires in the operating system market:  a) z/OS — the latest name for the mainframe operating system family that goes back to the Mythical OS/360; b) Unix variants; and c) Windows.  

z/OS is tightly tied to the mainframe world, a hardware and software combination that is both more vital than most people credit, and even less exciting.

Windows, well, is Windows.  A massive market, a technical train wreck.  The Vista debacle is probably a harbinger of things to come, but Windows will hang on, especially on the desktop.

In the Web world, the name of the game is Unix: it’s dominated by a software stack of a Unix variant, Apache’s web server, MySQL, and PHP, with some fashionable up-and-coming packages (including Ruby on Rails) challenging PHP.  

Sun, with arguably the technically strongest and most secure Unix variant in Solaris, and with MySQL, potentially can dominate this entire market, but only if Solaris can compete with Linux.  As a software company, Sun can sell services and support against the xAMP stack that no other company can match.

Another advantage is that Solaris is also the only really commercial operating system that is fully certified under the government’s Common Criteria as a system that can be trusted for classified data.  I could write an entire article on that point alone, but for now it’s fair to say that there should be a lot of opportunities for Solaris Trusted Extensions in a world of cyberwar and nationally-sponsored cyber-espionage.

So it seems that there may be a software company hiding in Sun Microsystems; so it’s probably not a coincidence that Staley Cates of Southeast has been quoted saying that they think of Sun as a software company.

Storage

When Sun bought StorageTek, there was a lot of excitement, tempered by dark suspicions based on Sun’s legendary clumsiness handling acquisitions.  Those suspicions were justified: since then, Sun has lost storage market share and taken a $1.5 billion write-down.  

Still, the storage business overall is good, and as often as its death is predicted, tape as an archival medium isn’t going away soon.  But Sun has another advantage in storage, one that I don’t think the rest of the market realizes yet.  It is that the cost of providing IT services, either in-house to a big company, or hosted, is mostly driven by the costs of administration;  meanwhile, the hardware fixed cost is far more dominated by the cost of operations, and a lot of that cost is dominated by backup and archival.  (This is only getting worse as the combination of Sarbanes-Oxley and the courts’ interpretation of discovery rules make long term storage a legal necessity.)

Thus, Sun has a potential game-changer in storage with its new ZFS file system.  With ZFS, great masses of data can become one giant pool of storage, with efficient backup, transparent use of commodity hardware, and the potential for reducing the management cost of big storage “farms” tremendously.  Tie ZFS to storage, show that resulting operating cost benefits, and Sun can sell disks, tape, and Solaris.

Systems

Historically, Sun has thought of itself as a systems company, but this is a business that is slipping away.  For whatever reasons, the x86 architecture is dominating everything in computing.  Sun’s SPARC architecture has technical advantages, but the economics of chip manufacture are such that being able to ship many, many millions of units beats technical superiority every time.  Thus, with the exception of some specialized markets, keeping a proprietary chip architecture no longer makes sense.

But processor architecture isn’t the only thing that makes systems, and Sun has some other technologies up its sleeve.  Web-based applications tend to “scale” best by adding many relatively small, processors, each of them handling only a few computations but working together in parallel to provide mass computation.  This kind of system depends critically on fast communications amongst the processors, and to memory, and Sun has a very strong position there in Infiniband, a very fast, reliable network that can connect both processors and storage.  Sun is already experimenting with super-high-performance computing using Infiniband.  Combine a very fast “backplane” with fast commodity processors, and you can build very fast, cost-effective systems.  What’s more, since they are based around large collections of inexpensive individual processors, the combination becomes extremely scalable: you can start very small, and build out very big, without making massive changes.

It’s interesting to note at this point that Andy Bechtolsheim, legendary systems guy, has “left” for a startup, Arista, looking toward very high performance inter-system communications.  Bechtolsheim has done this before, and then sold his startups back to Sun to create new product lines.

Combined, these technologies would give Sun a very interesting product line: the dominant xAMP software stack; inexpensive, high performance commodity systems based on x86 and Infiniband, and all of it tied to massive systems storage with low cost of operations and management.

The Costs

But Sun won’t be able to do any of this without giving up some things.  The company needs to concentrate on things that it can dominate, things that will lead to sales and profits.  And so, there are some obvious businesses the company should drop:

The SPARC Architecture.  It’s a shame, but as noted above, when you can amortize the die costs across millions of chips, cheap common hardware will beat technical elegance.

Java and Open Office.  As a technology, Java has been very successful, but Sun has really never made a lot of money from it.  Javasoft and Sun made some big mistakes in developing java as an “enterprise” language, and with the release of JEE it’s only now beginning to recover from those errors.  No matter; Java is established now . . . and as every old COBOL programmer can tell you, widely used languages never die.  So, form a Foundation, and give it control of the Java Community Process.  Let OpenOffice go at the same time; it really was always just a way to insulate Sun desktops from Microsoft Office — and while it too is widely used, it also doesn’t generate a lot of revenue.  Give it up.

Desktop systems.  This one is a goner, anyway.  It’s clear Sun’s commitment to desktop systems is nearly gone.  The only desktop system Sun sells today is the Ultra-25, which is basically just a Sun-branded x86 desktop.  Sun would be better off devoting resources to making Open Solaris work on as broad a variety of hardware as Windows does.  Then someone who wanted a Solaris workstation could buy one from Dell.  Or, conversely, make an open alliance with Apple, and establish Macs as the “Solaris desktop.”

Current Company Management.   Look, I don’t know much about football, but I do know that when a team has gone 10 wins to 90 losses, it’s time for a new coach and staff.

So there’s my thought experiment.  Despite the nay-sayers, I still believe there is hope for Sun.  All the company has to do is abandon its past image of itself, and then figure out how to sell scalable software on a low-operating cost platform.  Best of all, it already has the products it needs to get there.

So, all it will really take to turn Sun Microsystems around is a management brave enough to take an honest appraisal of itself.