Will The Sun Rise Tomorrow?
It Industry's Maverick Facing Tough Obstacles In Winning Back Customers
By Greg Goth, Bank Technology News
3 May 2005
Sun Microsystems has gained a reputation through the years as the feisty underdog, the scrapping outfit with the different idea good enough to battle the dominance of Microsoft on the enterprise desktop and that of IBM in the data center.
The willingness to do things its own way has earned Sun plenty of respect and admiration among competitors, technology analysts, and customers. Sun's combination of powerful servers and highly capable workstations captured a significant slice of financial services customers as electronic trading became mainstream a decade ago.
But since the bursting of the Internet bubble five years ago, the respect has not been partnered with the willingness to buy what Sun has been selling-the company lost $244 million in 2002, $1.8 billion in 2003 and $1.1 billion in 2004. And presently, as IT executives are finally beginning to authorize major investments to upgrade data centers last overhauled for the Y2K problem, Sun might find itself in further distressing straits. "The problem Sun is going to run into is that organizations have to make decisions at some point in time," says John Kershner, CTO for Waterbury, CT-based Webster Bank. "When all is said and done, if you're investing millions of dollars in an infrastructure, you have to ask yourself, 'What have they been able to deliver to date, and what can they deliver in the future?'"
For Kershner, the answer to that question, about six months ago, was to change the company's core infrastructure from Sun and its hallmark Solaris Unix operating system to IBM's AIX Unix. Kershner says Webster executives looked closely at Sun, Hewlett-Packard and IBM products before deciding on IBM.
"When you're making big, long-term decisions, you need to be confident they will remain in the market," Kershner says. "We couldn't get that comfort with Sun."
Kershner says Webster chose IBM for several reasons beyond Big Blue's stability. The platform chosen will allow Webster to run either AIX or Linux. The mid-range pSeries 670 and 690 database servers will allow for partitioning and more flexible allocation of resources, a feature available only on Sun's more expensive machines. Kershner was also impressed with the IBM machines' price-to-performance ratio and failover capability. He also says several large enterprises near Waterbury, including ESPN and a large insurer in Hartford, also invested in IBM equipment, and he believes the proximity of other enterprise IBM customers is beneficial.
Not that Kershner has any complaints about the Sun installation that predates his 2004 arrival at Webster.
"We've been very pleased with the support we've gotten from Sun. Their products have performed wonderfully," he says. "We obviously hope Sun recovers and does well. We all like to see competition in the marketplace."
Gordon Haff, senior analyst at technology consulting firm Illuminata, says Sun must leave the red ink far behind if it has any hope of regaining the trust of its core financial services constituency. "The bottom line is, Sun has managed to keep themselves from imploding," Haff says. "They have a lot of interesting ideas out there and interesting projects going on, but if a year from now they're still basically not making money and they're saying, 'We're working on improving our situation,' I think those who have not lost faith or patience with them yet are going to be getting pretty close to the edge. I can't point to any magical deadline at which they'll need to make a certain amount of profit, but the foundation should be set. This is clearly a year where they have their plans in place, products are rolling out, and there is spending going on in the industry as a whole. They have to show some real concrete evidence of forward progress."
Yet Stuart Wells, Sun's svp for worldwide financial services, remains fully confident the company's new offerings and vision will capture market share just as its powerfully networked workstations captured Wall Street in the last decade.
"There's a sense of passion," Wells says. "There's a sense of urgency, there's a sense that we have competitive weapons like Solaris 10. There's not a sense of make-or-break. We have $7 billion-plus sitting in the bank. We've got the cost structure for the company well in order. I've been at Sun 17 years and I'm extremely bullish on the company."
Wells's optimism was borne out by a major deal announced on March 31, when JP Morgan, a Sun customer for more than 20 years, revealed it was going to use Solaris 10 to build new applications, including grid computing, data archiving and virtualized data center solutions. The applications, which are now in testing and development, will utilize the multi-platform power of Solaris on both Sun's SPARC and x64 systems. Full independent software vendor (ISV) portfolio support to port applications to Solaris on x86 systems also will be available.
Solaris 10 is at the center of Sun's new offerings. It's the newest version of its robust and reliable Unix operating system. Solaris 10 offers features such as guaranteed backward compatibility with older versions of Solaris, a completely rewritten TCP/IP stack for faster performance, and special optimizations for multithreaded processors and x86 architectures.
Wells says the improved performance of Solaris 10 is attractive to financial services customers looking to consolidate data center functions. "We see a huge amount of business in that area because all these financial services customers have space problems, power problems and heat problems. Data centers are full, and they are trying to see if new technology will allow them to use existing data center floor space and then give them back floor space, and at the same time give them power savings. In many cases, we go in there and give them 40 percent or sometimes more floor space, and then also save them 15 to 25 percent on power."
In late January, Sun also finalized its plans to make Solaris an open-source operating system to compete with Linux, under a different licensing scheme than Linux's General Public License. Wells says the move will allow Sun's internal developers as well as external developers to craft new features for Solaris functions at price-to-performance ratios competitive with Linux, plus a wider array of industry-specific utilities and management applications. In addition, he says the internal Sun development process will deliver a tried-and-true pipeline missing from the ad hoc nature of Linux development. "With Solaris 10 on Opteron or Intel x86 systems, we are lower cost and higher performance in many cases than Red Hat. Our support costs compared to Red Hat are two-thirds the cost over a three year period."
Others, however, say Sun has entered the race too late. Guillermo Kopp, vp at TowerGroup, says Linux has matured enough over the past couple of years to make it perfectly acceptable as the de facto open source operating system even in a conservative industry such as financial services.
"It's a niche play," he says. "As good as Solaris is from an operating software perspective, it still requires developers to learn and feel comfortable in that environment, and I would contend developers have already done that with Unix and I would contend they are beyond that. It's a me-too play. They may capture some niche opportunities, but I don't see Solaris emerging as the big open source software. Linux is already there."
Indicative of Kopp's observations is the experience of Steve Rubinow, CTO at Archipelago Holdings, which operates the first totally open all-electronic stock exchange in the United States. (At press time, the New York Stock Exchange had announced it would acquire Archipelago and rename itself the NYSE Group. It is unclear how this could affect Sun's relationship with the merged entity.) Rubinow has been a Sun customer since 2002, when Archipelago bought GlobeNet, an electronic communications network that was already a Sun shop.
Archipelago chose Sun over Linux at the time because the supporting industry infrastructure around Linux had not matured to the point Rubinow felt comfortable relying on it. While Archipelago will continue to be a predominantly Sun shop, Rubinow is hedging his bets-he says his staff has begun some pilot work on Linux. "If Solaris 10 had come out earlier we probably would have skipped the Linux stuff, but we had some infrastructure running Windows and wanted better performance, and Solaris 10 was in beta, still too early for production. We decided here's an opportunity for us to get some firsthand experience in Linux. So, I have a few running in production right now, and I'm going to be rolling out more that are Linux based. When when we feel more comfortable with Solaris, we're going to have to make a decision about which operating system is going to be rolled out, whether it's Solaris 10 running on x86 or Linux."
Yet other Sun customers seem to be convinced Solaris 10 has already proven itself to be competitive in multiple deployments. The JPMorgan deal, for instance, was announced after a series of pilot projects, including utility computing, provisioning and a trading data archive for use by JPMorgan and its clients, were conducted in New York, London and Hong Kong over the last year.
In addition to being a customer, Archipelago also is embarking on a partnership with Sun to deliver computing power over a grid. The Sun utility computing model, launched in February, 2005, offers customers power at $1 per CPU per hour and storage at $1 per gigabyte per month. The Archipelago partnership will enable companies to bid dynamically for open compute cycles. Both Wells and Rubinow are eager to see how the concept proves itself.
"Without disclosing customer names, we've already had multiple financial services customers that use the grid," Wells says, and expects others to be intrigued by the concept as well, such as "big banks who need to do Monte Carlo simulations, or small hedge funds who don't have the time or inclination to build up data centers with 10,000 CPUs."
"It's more of a theory now than reality, because we have to see how the market shakes out and if people really come to play," Rubinow says. "But from a utilization standpoint, the last time I saw a number I remembered, the average server utilization in a corporation is about five percent, so that means they have terribly underutilized assets. This is a way to improve the ROI on those. Of course, how you enable this, how you deal with the security concerns, needs to be worked out. Sun will tell you their grid technology is the closest thing to being able to make this real."
Illuminata's Haff says the grid concept will probably take some time to gain momentum. "In the near term, for regulatory reasons if nothing else, that is sort of a tough sell," he says. "In financial services, there are a lot of interesting grid styles of computing, but I would expect them to be more internally hosted for a lot of reasons, such as compliance and risk mitigation. These are sophisticated customers, but they have a lot of issues with going to sharing and being shared outside their organizational model. That's not to say that down the road something like that couldn't happen, but it just doesn't seem too likely in the near term."
Webster Bank's Kershner confirms Haff's sentiments. "We really haven't looked that closely at the grid model," he says. "We need to make sure we have the blocking and tackling done correctly. We're focused on reliability."
TowerGroup's Kopp minces no words in his estimation of where Sun should focus its work if it wishes to recapture its past strength in financial services.
"Java is the one and only valuable asset Sun has today," Kopp says. "Yes, they have a brand name, they have engineering prowess, they have a portfolio of clients, but all those things are at risk strategically in the next five years unless Sun builds on a key strength, and to my mind, that key strength is Java."
From data center code portability to the coming wave of pervasive computing that will enable the most mobile of devices to perform highly sophisticated operations, Sun's famous "write once, run anywhere" software is either already firmly entrenched as the de facto standard or is likely to become one.
Webster's Kershner says the bank's entire Internet banking infrastructure is written in Java and has "performed magnificently," and the new IBM equipment will also include WebSphere middleware written in J2EE.
Archipelago's Rubinow says many of the old perceptions of Java as too slow to run fast financial services applications are receding. "The two primary languages for us are C and Java," he says. "We probably have a legacy barrier to overcome mentally here because we use C for any of our applications that require the highest performance, and that's probably true of a lot of Wall Street firms as well. However, depending on who you talk to, some people will tell you Java can perform just about as well as C, and so good you probably couldn't tell the difference-and yet you'd have all the benefits of Java portability, productivity, etc. We're actually going to try and prove that to ourselves here as well soon."
Java is also poised worldwide to power the mobile banking revolution, according to Wells, who says the other pieces of the technology puzzle, such as security protocols and broadband-capable mobile devices, are coming into place.
"You've got the pieces you need to provide high-performance, high-quality services to consumers," he says, "and for the financial services community, you've got multiple-factor authentication available to you so you can actually do trading or you can provide mission-critical business information on mobile devices. The Europeans and Asians seem to be leading in terms of adoption of these sophisticated handsets and mobile devices. The technology is there."
The technology may be there, but Sun must also combine that with strategic vision to emerge from the doldrums, Kopp says. "They need to be selective on the verticals they target," he says. "They need to learn from the past. I think they won Wall Street years ago by accident. Now they have to win Wall Street, or whatever verticals they choose, by knowing them by heart. The company is taking some right steps, but it's not there yet."
© 2005 Bank Technology News and SourceMedia, Inc.