Impacts for Sparc, Power and Java
An IBM/Sun combination raises several interesting technology questions underneath the business and cultural issues.
It is easy to imagine IBM might let the Sparc microprocessor architecture die a gradual death, articulating a road map to its Power and X86 systems. That's just part of the consolidation game in an industry that has winnowed out dozens of interesting but relatively low volume CPUs.
But the Niagara version of Sparc for Web servers has pioneered work in multicore processing. It sports the most threads and integration of any chip in the server space with embedded crypto and networking cores.
An overdue Rock chip planned to push the boundaries in multicore for database processing. It was described as leading the way in news areas such as transactional memory and scout threads.
In the practical hands of a Palmisano, those pioneering efforts will probably wither on the vine. But the ideas they proved out could well appear in a Power8 or Power9 design with new marketing names attached to them. IBM might even consider using a Sparc core as an accelerator for legacy Sun workloads in a future Power chip.
An IBM/Sun combination might finally free up Java. Despite all its talk about openness, Sun over-managed and controlled the technology around its programming language seeking ways to make money from it up to the recent release of its JavaFX. That often caused clashes with some of the big OEMs who wanted to use Java as a wedge against Microsoft in everything from cellphones to set-top boxes and servers.
If the merger goes through, IBM should find a home for Java in a neutral standards body such as IEEE. The language and the technology around it will have a clearer, cleaner role to play if it is truly open.
If Sun is acquired, no doubt many engineers will be out on the street at a difficult time. But some of Sun's top talent has already moved on. Last year, co-founder Andy Bechtolsheim left for the startup world and David Yen who ran many of Sun's pioneering efforts joined Juniper Networks.
Perhaps the biggest lesson Sun may have taught the industry is how far you can go in open source—and how little it apparently means to customers. Under Schwartz, Sun made its Solaris operating system and even its Sparc architecture available for anyone to adopt, adapt and drive forward. The company got virtually no takers.
In some ways, an IBM/Sun merger would mark the end of the independent, vertically integrated computer company. In other ways it would mark the beginning of a new and very different wave of consolidation.
Cisco's entry into the server business puts pressure on networking companies such as Brocade and Juniper to deliver a full set of data center systems, including server blades. That could send them running to the doors of some of the second tier server makers such as Fujitsu-Siemens. And so, the consolidation continues.