Solaris 2.6 ushers in the millennium
What's Sun's new, long-term strategy for Solaris? How will Sun improve on everything from ease of use to Solaris-on-Intel sales?
By Rick Cook
In conjunction with the company's recent official introduction of Solaris 2.6, Sun outlined the goals of its "Millennium Project." Where is Solaris headed? We tell you what markets Sun will pinpoint, how it plans to make Solaris easier to use (especially for PC users), and how its two-year subscription model could work to its advantage. (2,500 words)
Solaris 2.6 is the first major result of what Sun informally calls its Millennium Project, a plan to remake Sun's entire operating system effort by the year 2000.
"It's actually less than a three-year plan," says Brian Croll, Sun's director of marketing for Solaris servers. "We're very close on all this." By the end of next year or the year after Sun will probably have all the pieces in place for its new look.
The thrust of the Millennium Project goes beyond the features in Solaris 2.6. There are also the longer-range questions of what Sun is trying to accomplish with its operating system and how it intends to do it.
What Sun needs to do is no mystery. The challenges Solaris faces are pretty much the same ones faced by all its competitors. Talk to Microsoft, Hewlett-Packard, or the Santa Cruz Operation and you'll hear about scalability, ease of use, and delivering new features in a timely manner. But each company is starting from a different mixture of strengths and weaknesses and pursuing these goals in a different way.
In Sun's case the starting point includes such strengths as scalability and major positions in workstations, enterprise servers, and Internet servers. Like all the companies offering modern Unix, Sun also has a well-developed operating system to build on.
Sun's weaknesses include ease of use, ease of system administration, a very minor position in the PC market, strong competition from Microsoft in the departmental and enterprise server markets, and a lack of reach for the very high end. Sun's liabilities also include a reputation for arrogance and a style that has focused on competing against itself rather than against other companies in the market.
"You don't refer to how much better your product is in contrast to its previous release, you refer to it versus its competitors," says Chris Le Tocq, a software analyst who follows operating systems for Dataquest. "Your focus should not be on fixing bugs the customers have identified. Your focus should be on `what does it take to make sure this product stays competitive in the marketplace?'"
In announcing Solaris 2.6, Sun also announced a new business strategy with three main components. Taken together they show Sun's strategy for capitalizing on its strengths and shoring up its weaknesses. As announced, Sun's plans look good. Of course execution is about 90 percent of any strategy, and only time can tell how well Sun will execute on its plans. Still, Solaris 2.6 and what's come out with it is hopeful.
Multiple markets, one (more or less) OS
The most obvious change in the Solaris software is one of the most expected -- further moves toward a full 64-bit Solaris. This is important for the enterprise market, where Sun is taking territory traditionally belonging to mainframe operating systems. But it is only part of the story. Solaris is intended to be an operating system that will work across a full range of computer systems.
Sun is moving toward 64-bit Solaris at a measured pace, giving its customers plenty of time to upgrade. Sun does not intend to repeat the confusion and incompatibilities that marked the switch from SunOS to Solaris. "That was one of those never-again type situations," says Croll. "I think we learned our lesson very well off that."
Solaris 2.6 includes new features to enhance clustering, multiprocessing, and reliability with automatic failover. In 1998 the enterprise version of Solaris will be able to handle 16 nodes of 64 processors for a total of 1024 processors. Like the 64-bit features in Solaris 2.6, this is aimed at the high end of the market, where Sun's competition is big iron.
At the high end, Sun is well ahead of its traditional competition, especially Windows NT. Indeed, Microsoft's recent Scalability Day only served to focus on what NT lacks as an enterprise-wide operating system. However, Microsoft and the other competitors are not standing still. As the economics of using Unix or NT as an enterprise operating system become increasingly compelling this area will be more strongly competitive.
Enterprise computing isn't just a matter of processing power. At the company-wide level, management, security, and performance tuning issues become critical. These have been areas where operating systems like Solaris have lagged compared to systems like VM and MVS. As part of the enterprise effort Sun plans to offer enhanced security, better management, and other features.
The first release of Solaris 2.6, however, is the Web-enhanced version intended to drive Web servers. There are several reasons for this, not the least of which is that the Web market is white hot now, and Sun has a major piece of the action.
Ultimately Solaris is to be a universal operating system that will run everything from a multinational corporation down to a subnotebook computer. This is less because it's a theoretically neat idea than because Sun's customers want it. Having the same operating system everywhere greatly eases the management and integration problems which are the bane of modern information technologists.
An operating system that runs everything from a palmtop to a Fortune 500 company has been an IT manager's dream for a long time. It has remained a dream because it is so difficult to do, partly because such an operating system would be bloated with unnecessary features.
"Like a number of other people, I think Sun understands that the more functionality you put in an operating system, the more potent it becomes, the more risk that it will become too monolithic," says Jean Bozman, research manager for International Data Corp., a market research firm. "In the last couple of years they all came to the same conclusion, and they are all moving to a modular approach. There are certain core functions you have to have to support the file system and hardware and such. But then you have to recognize you're serving a wide array of users."
One solution is different versions of the operating system. In Sun's case that means versions of Solaris optimized for its four areas of market focus: enterprise, departmental servers, Web servers, and desktops.
Sun is careful to note that no one loses any functionality in this development. All the versions of the OS will at least have all the capabilities of Solaris 2.5.1. However, each version will probably become increasingly specialized as new functions are added.
Sun isn't the only one taking this approach. Microsoft has announced that it is splitting Windows NT into at least three editions as it stretches to serve the range of the market and it further extends Windows' reach down to the palmtop and embedded systems market with Windows CE.
It will take time to see how well Sun's strategy works -- and how well the company can implement it.
The operating system as process, not product
Under the new marketing plan, Sun talks about selling subscriptions instead of operating systems. Solaris customers will purchase a two-year subscription and automatically get all OS updates and patches for that period. Delivery of these updates will be in the form of CDs.
As a marketing strategy, Sun's new approach to Solaris is especially interesting. It could solve problems for Sun and its customers, but unless Sun moves carefully it is going to confuse a lot of customers at first.
Essentially the subscription concept moves from the idea of operating system as product to operating system as process -- a work in process that is never finished. Perhaps more correctly, Sun's strategy merely recognizes the "never-ending OS." It has been a fact of life at most installations for years. Between bug fixes and major upgrades, most operating systems are extensively reworked about every two years anyway.
Microsoft's solution has been to simply sell the latest-and- greatest version of the operating system. Of course the result is that Microsoft is notorious for slipping release dates and then offering incomplete and buggy versions of its operating systems in its initial releases. Windows didn't take off until Windows 3.0., NT 1.0 was a mess, and it's rare for the first release of a Microsoft OS product to include all the features originally touted.
This also has a serious impact on customers since they tend to hold off on buying decisions when a new release of an operating system approaches. That affects cash flow for the company and leaves customers putting off upgrades, often for months, while they wait for the new version of the OS.
The subscription plan sidesteps all this. Sun avoids the pressure of explicit deadlines to release new versions of Solaris while still promising timely upgrades and generating a consistent revenue stream. The customers don't have to worry about paying for upgrades since they automatically get any new versions that come out during their subscription.
This still requires some adjustments for Sun customers. For one thing, Sun says Solaris 2.6 will be the last "version" of Solaris. From here on out the subscription approach will take care of upgrades and compatabilities, so there will be no need for versions. Then there's the simple recognition of change. Computer users, at least outside the mainframe world, aren't used to the notion that their operating system will require a continuing investment to keep current. Of course most of them have been doing it for years, but they're not used to thinking of it in that way.
One of the major advantages is that everything in a Solaris release will be tested together. This should greatly reduce the kind of introduced incompatibilities that are all too common with patches.
Patches and their incompatibilities have been a particularly vexatious problem for Sun. One of the most common discussions on the Internet Solaris newsgroups is which patches are needed for what and which ones are incompatible. This is the logical result of a big, complex operating system (as a check of the NT newsgroups will show). Still it is frustrating for Sun's customers and has led to such anomalies as "private patches" (patches developed for a specific customer and not generally released). Considering the amount of testing needed for a "full" patch, this makes sense from Sun's point of view, but customers resent it.
If a.) everything works as advertised and b.) the subscription price stays reasonable, Sun could well have a winning strategy in subscriptions. Eliminating patch incompatibilities is especially important and only time will tell if that part of the plan works.
Ease of use
The dirty little secret is that any server-level operating system needs to be administered. With NT, Solaris, HP-UX, or Netware, you have to have an administrator. However that doesn't mean that a server-level OS has to be either difficult to administer or require the constant ministrations of someone with a lot of training.
This is one area where Microsoft's Windows NT has a significant advantage over Solaris. It is seen as much easer to set up and manage. The truth is that NT isn't as easy to administer as reports have it, and Solaris isn't as hard. But the perception has enough roots in reality that it has hurt Solaris, especially in the departmental server market.
Sun is aiming to make major improvements in ease of use and ease of administration, and Solaris 2.6 lays out the first steps. The company stresses this doesn't mean "dumbing down" the system but instead providing an alternative, easy to use, method for handling many jobs.
The alternative is, of course, browser technology. Not only does a browser provide a consistent, easily recognizable interface for administration, it lets a remote expert handle the more complicated parts of the administrator's job via an intranet.
"We're going to be relying on browser technology to get the ease of use," Croll says. "For things like installation and management that's really wonderful. The browser is self-documenting, and it's a nice way to mix actions with documentation very easily. It really has the potential to leapfrog technology that's being done today."
"To administer and manage Unix you had to know quite a bit," Croll says. "The browser just kind of dropped into our laps."
The plan is not to replace everything with a browser interface. Croll points out that for many kinds of jobs a conventional application interface is a much better choice. But for a lot of common tasks, such as looking something up in the documentation or installing new software, the browser interface is superior.
Ideally there should be no visible administration, and for Web servers that's at least theoretically possible.
"When we say `Web server of choice' it has to be all the way up and down the line," says Dataquest's Le Tocq. "You should be able to bring in a reasonably-priced box up and down the [size] spectrum with Solaris embedded, plug it in, power up, and you've got a Web server."
"The notion that there's an installation required, the notion that there is administration, the tolerance of administration has to go away. Sun's challenge is to make it go away, and they can do it."
Increased focus in four market areas
Finally, Sun's decision to focus Solaris on four areas -- desktops, intranets (for workgroups), Internet service providers (ISPs), and the enterprise -- is more interesting for what it implies than for the areas Sun chose.
Sun has traditionally been a technology driven company that tended to divide its efforts by technology rather than customer base. By deciding to focus on the customers' needs rather than the technology it offers, Sun seems to be signalling a shift in attitude.
The company is now officially targeting sales to ISPs with this breakdown of four market segments. On the x86 front, Croll says another focus will be on deals with major Intel system providers. This is a change from Sun's heavy emphasis on value-added reseller (VAR) channels or selling directly to customers in this market.
Making it work
Can Sun make its new Solaris strategy work? On paper there's no reason why it can't. There are, however, some major challenges, notably in the areas of ease of use and customer focus, but the company seems to be headed in the right direction.
At this point the question is less the strategy than the execution. Solaris 2.6 represents a major change, and that's never easy.
"Sun absolutely has the ability to succeed in its grasp," Le Tocq says. "If they find themselves with a knife in their back, they may well find their own hand on the knife."