The headline is a little ludicrous, because almost nobody serious uses Windows on clusters. Take a look at the beowulf mailing list, or the Top 500 -- 1% of them run Windows.
As the largest supplier of commercialized Linux operating systems, it is a bit embarrassing for Red Hat that in the supercomputer market that gave Linux legs, other distributions own the high performance computing (HPC) space. That includes Novell's SUSE Linux Enterprise Server, but also a smattering of other Linuxes. Enter, …
You do not pay >$100 dollar per node if you can get it for free. Most free distributions out there fit just as good, since the OS will most likely be stripped down and optimized for running a few applications and some communication stacks.
If you have a global site license, like most universities, it doesn't bother what you run though.
You rather put money into more nodes, applications, queuemanagers (like LSB, PBS) and cluster management software (like Scali,RLX etc if it is a cluster, not a sh-mem machine).
If you run commercial binary only versions, and they are only optimized for RedHat, then you run CentOS. This was one of the main issues for that distribution to be created.
In most cases commercial HPC applications vendors are flexible and optimize the software for your OS of choice, or they provide header files and generic objects so you can compile yourselves.
Look, if you're going to fill an article with acronyms and expand on some of them in a pseudo-helpful way, at least get them RIGHT, otherwise you just look daft.
Platform's LSF is not "load service facility", it's load SHARING facility, and has been since before Windows 2000 was knee high to a grasshopper. Anybody with a real clue in the HPC sector would know this.
The rest of the article wasn't much better.
Have a nice weekend
Biting the hand that feeds IT © 1998–2022