Oracle made it clear that it was Sun Microsystem's software business that made it want to buy the company, raising significant questions around the fate of Sun's hardware business, analysts said on Monday.
Oracle said Monday it planned to buy Sun for $7.4 billion, and that it was most interested in the company's Solaris OS and Java software. It said it would build soup-to-nuts systems that combine Oracle's database, middleware and applications with Sun's operating system and hardware.
"Engineering the Oracle database and the Solaris OS to work together allows us for the first time to deliver complete integrated computer systems, from database to disc," said Larry Ellison, Oracle's CEO during a conference call.
Whether the hardware it offers will include Sun's proprietary Sparc-based systems was unclear, however, and analysts wondered whether Oracle would support them over Sun's x86 servers in the long term. Sun's Solaris OS is available for both Sparc and x86-based systems.
The industry will have to wait and see what Oracle does with Sun's hardware business as the deal unfolds, said Jim McGregor, chief technology strategist at In-Stat.
Sparc's market share has been falling to rival Unix server vendors like Hewlett-Packard and IBM, giving Oracle more incentive to cut the development of Sparc or sell off the business, he said.
"A lot of these [Unix] products have been withdrawing to smaller and smaller areas like high-performance computing. It doesn't make sense for too many companies to continue long with those architectures. Look at companies like Silicon Graphics and Cray, all these guys have just moved to what's off the shelf," McGregor said.
Oracle executives have declined interview requests to discuss the deal. In a conference call Monday, CEO Larry Ellison implied that Solaris is more important to Oracle than Sun's Sparc chip. "Sun's largest business is its Sparc Solaris computer business, and the heart of that business is the Solaris OS," he said.
Oracle has already built high-performance servers with Hewlett-Packard for data warehouses based on industry-standard hardware. With Sun, it can see other opportunities to build such systems.
"Sun's open storage platform is very similar to Oracle's Exadata storage platform. Both use standard servers, commodity disks and InfiniBand interconnects to lower the cost of storage while improving performance," said Charles Phillips, president at Oracle.
Oracle is a big company and can afford to jump into the hardware business, but Sun's most valuable asset to Oracle is its software, McGregor said. "I won't say it's not a good opportunity, but being in the hardware segment is probably not the best opportunity for Oracle," he said.
The question is whether Sparc offers enough advantages to Oracle to make continuing its development worthwhile, he said. The market for specialized chips like Sparc has been slowing as companies move to industry-standard servers running on chips from Intel and Advanced Micro Devices, McGregor noted.
Oracle would be unlikely to stop developing the Sparc chips right away, and may continue to offer systems based on the processor, said Dan Olds, principal analyst with Gabriel Consulting Group. Sun has already slowed its own Sparc development as it lacked resources to keep up with rivals Intel and IBM, he said. Sun has already outsourced some of Sparc's development to system vendor Fujitsu.
"There's not as much chip development as there was at one time. It's been hard for Sun to get more performance out of Sparc. Customers believe IBM and HP systems offer more performance," Olds said. As Sun loses ground to IBM and HP, Oracle would need to maintain a volume of shipments to justify continued development of the chips.
Oracle does not have much incentive to continue Sparc development as it goes against its model of building a standards-based platform, said Jack Gold, principal analyst with J. Gold Associates. Oracle's software is mostly hardware-agnostic, and it may not want to offer a specialized system to sell its software.
Oracle might slow
Killing SPARC would be the smartest thing Sun has done in 10 years. It's a shame that it will have taken an acquisition and probably a partial dismantling of the company to make it happen, assuming it does. Sun needed to embrace and open source Solaris/x86 in the early part of this decade rather than declaring it dead. That was when the major ISV's were looking to jump from SPARC to x86 hardware. When Sun wasn't there for these ISV's, RedHat took up the slack. Sun needed to start EOL'ing SPARC 4 years ago instead of investing all of their R&D into a next generation version of it that is of interest to no one but TPC geeks.
Sun has a promising software portfolio -- including some potential game changers like thin client computing. They just lack the focus to bring any of that software to market in a way that addresses the needs of their customers, and part of the reason why is that they're spending too much energy preaching the doctrine of SPARC and Solaris. This is not the 90's, and vendor lock-in is no longer a viable business strategy. If Oracle manages to force them out of the hardware market, it will be better for both companies.
Post new comment