Sun's vision to standardize IT

By giving away software that might otherwise be used to shackle customers, Sun earned customers' trust -- and pointed to a more meaningful way of delivering technology

There will always be solutions that IT has to cook up itself, but in the main, your business' needs and problems are not unique. Your plumbing is certainly not special. IT is filled with people who are continuously buying, upgrading, configuring, deploying, and maintaining largely identical machinery (software and hardware) for different employers. The amount of busywork invested in keeping the trade in commodity technology alive is mindboggling. We need to walk down the line of information technologies that haven't changed appreciably (or needed to) in a decade, call the finished projects finished, document them as standards, and stick them in the IT equivalent of the public domain. Then let's all innovate on top of that stack instead of wasting time on uncoordinated improvements to it.

The contribution of an operating system, programming language, database manager, or  development toolset to a company's bottom line or an end-user's productivity is zero. The value of any of these as a service in a cloud is the same. Below a high water mark of essential modernization, and excepting real innovation in their categories, such enabling capabilities are commodities that should just ship with hardware. That the modern x86 server BIOS still vectors to ROM BASIC whenever it can't find a boot record speaks to the sorry state of IT evolution.

[ Earlier this week, Oracle agreed to buy Sun for $7.4B. InfoWorld's Neil McAllister correctly predicted the acquisition. What does the deal mean for Java? For Sun hardware? For MySQL? For Microsoft? ]

I blame lawyers. Commoditizing core information technologies is a very nice idea. It's exactly what we need to equip an inherently interoperable cloud of clouds. This sounds in tune with the song of open source, but open source is often about reinventing the mature technology that owners won't part with. In order to do what I've described, Microsoft or Novell would need to have a serious change of heart. Failing that, another company would have to obtain a catalog of unencumbered properties to commoditize for the overall good. We don't really want orphaned software. IT demands accountability from vendors, so code needs to be kept in repair. We don't want to eat a face full of dust when we boot our servers; even a commodity, standard server OS needs to know how to make good use of AMD's 6-core Istanbul CPU. No, these free and public commodities need to be commercially maintained and supported. Volunteers?

Sun set out to become the de facto source for enterprise solutions, the place that any company, organization, or agency looking to use technology to make problems go away would ask first before trying to do it themselves. Sun wants customers to move away from expressing requirements in deeply technical, heavily qualified terms and engage in shared brainstorming at a higher level. IT gets better solutions when it leaves issues of platform, language, system, tools, infrastructure, and middleware open to a trustworthy solution provider, but trust is the sticky point. UNIX's heyday was marred by vendor lock-in, with Oracle and Sun being the poster duo for that practice. Whenever a vendor says "We'll set this up for you," I still hear the jingling of handcuffs.

The brilliance of Sun CEO Jonathon Schwartz's method for turning the page on that era, for earning the trust of IT customers -- giving away the very assets that might otherwise be used to lock customers into a solution -- will be appreciated only retrospectively by most. The value that Oracle sees in Sun is a pipeline primed with enterprise projects, and a huge base of customers and prospects whose trust Sun has earned by patiently disarming its own capability to lock its customers in. The way to do that is to hire lawyers once to write lawyers out of a license for good.

By drawing on free of charge, free for unlimited use software from its own catalog, Sun can afford to set up trials for prospective customers -- and it can afford to walk away from a project that goes south after a few months. Sun saves itself the awkwardness of having to go back to a customer site to get its stuff if a relationship doesn't work out. It doesn't have to put prospects in collection for enterprise software bought to kickstart a project that seemed like a good idea at the time, but never ramped up. Those bills never get paid. They only accrue bad blood.

Sun started with Java. In stepwise fashion, Java was transformed from proprietary, to open, to free, and finally to open source. The integrity of Java standards is ensured by binary distributions, so the reliable operation of commercial software that relies on Java as a foundation is protected. The source is there to carry Java to the few places where it isn't.

Solaris is undergoing a similar transformation. At present, the Solaris 10 operating system is the only freeware certified UNIX. Solaris 10 is free of charge, free for unlimited commercial use by any registered user who downloads it. If you build your own UNIX server by combining a generic x86 rack box with a downloaded Solaris binary obtained from Sun, you can activate commercial support for that server with a phone call. Sun has open-sourced unencumbered swaths of Solaris as OpenSolaris. The Solaris binary is not OpenSolaris. This method of becoming an after-the-fact Sun customer is common to the assets that Sun releases for free licensing.

Some commodities are worth keeping secret even if they aren't worth charging for. Sun Studio is a free, supportable commercial C, C++, and Fortran compiler suite for Solaris and Linux that brings the power of advanced performance optimization, including profile-based optimization (i.e., optimization based on run-time trace recording), and auto-parallelization to free software. Sun's recently acquired MySQL gives the cloud a mature, unencumbered database manager that can be swapped out with something capable of heavier lifting, where that proves necessary.

The rest of Sun's portfolio of enterprise software assets is worth a cruise. Appropriately, that which rises above the level of commodity or experimental project is not free. Oracle will have its hands full sorting through and classifying Sun's assets, but I hope it won't feel rushed in doing this. The freedom that Sun cut itself to create relaxed, ad-hoc, collaborative relationships with IT, academia, sci/tech, and even competitors has reaped rewards. When there is money to spend again, there is money waiting to be spent on Sun's enterprise solutions. Oracle wasn't on a mission to keep Java and Solaris from falling into the wrong hands. Even Sun, with its penchant for giving software away, was in business to make money, and it was on the right track.

Copyright © 2009 IDG Communications, Inc.