CA and Linux
Filling in the gaps
There are two reasons why it is worth taking note of CA with respect to Linux, writes Robin Bloor of Bloor Research. The first is that CA believes, as I do, that Linux is going to become the standard OS. I know this because I heard Yogesh Gupta, the CTO of CA, say so at the last CA World. The second is that CA believes that it can generate a respectable revenue stream from Linux.
CA's prime spokesperson on Linux is Sam Greenblatt, SVP & Chief Architect of CA's Linux Technology Group. Sam, also an influential voice in the Linux community, briefed me last week on CA's Linux strategy and, by way of digression, one or two other Linux-related things.
According to Sam, CA's approach has been and will continue to be to add the management capabilities that are missing from the popular Open Source community products - ones that are unlikely to be provided by them. He commented on a chat he had had with Linus Torvalds, saying that "Linus thinks of the developer as the Linux customer, rather than the corporations that use Linux. Most of the Open Source is like that."
This is not surprising given that the Open Source community is, almost by definition made up of developers. However, it affects the priorities for what is added to Open Source products and hence presents an opportunity to CA to fill in the gaps, that won't be filled quickly or perhaps won't be filled at all. Working in this way is a collaborative activity with Linux developers, getting to know the detailed working of Linux and intervening at the right point.
Sam was particularly bullish about the capabilities that Linux now has. He noted that Linux didn't scale well until it altered its threading model to conform with Posix threads (in about version 2.4.14) "Windows has a similar problem with threads, but Microsoft has not yet changed the threading model" he said "After the change, scalability ceased to be as limited and the about-to-be released version of Linux (v2.6) has added many scalability features".
Sam pointed out that six of the ten largest Supercomputers in the world are Linux based (in effect they are clusters of Linux servers) and that IBM's Blue Gene project, which aims to build the largest Supercomputer in the world for the Lawrence Livermore National Laboratory in California (6 times faster than the current fastest machine) is based on Linux.
For the record, Linux Version 2.6 scales up to support 32-processor servers and uses processors more effectively. The kernel includes improved support for NUMA (non-uniform memory access) and there are file-system improvements that will make systems go faster. Linux 2.6 will be more appropriate for both large transaction processing systems and data-intensive applications. According to Sam, it also has a virtualization capability built into it. In effect, it will be possible to run multiple Linux instances on the same machine. (Note that this is not the same as the VM Ware capability which can virtualize Windows and also mix Windows and Linux instances.)
CA is clearly serious about to Linux, having taken the trouble to get deeply involved in the Linux community in a co-operative manner with a clear commercial motivation. In Sam Greenblat, it has an unashamed Linux enthusiast and advocate. Whether CA can capitalize on its Linux enthusiasm remains to be seen, but Sam is confident. CA already claims to be among the five top providers of software for Linux. ®
Sponsored: Benefits from the lessons learned in HPC