While Linux clusters can handle the demands of high performance computing, they’re still lacking in some key features, said Eric Pitcher, vice president of technical marketing for Linux Networx, a cluster systems provider in Bluffdale, Utah. In part one of this interview, he championed Linux clusters, citing their productivity and scalability. In part two, he discusses pricing and points out Linux clusters’ current shortcomings.
Filling in the gaps among Linux clusters, part 2
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
They are the ones who are building and administrating Linux clusters in all the physics, chemistry, math and computer science departments around the world. That fact represents a huge save of money for universities which can use thier resources in a better way. There are not reasons in academic institutions to buy the wannabe solution which MS is marketing during the last months.
It is all very well to talk about Cluster in Universities but, there are a number of things missing from Linux Clustering (and most Unix Clustering Implementation) that are needed to improve the USP of this technologies to mainstream business.
1) A Cluster Wide filesystem (as used for 20years in OpenVMS).
2) Single boot images for multiple system.
3) Rolling Upgrages to the Kernel & major O/S components
Implementing any two of the three above would really differentiate LINUX from the Microsoft world. It has been done before and is one of the reasons why many companies don’t want to migrate away from OpenVMS.
Just my 0.02Sum’s worth.
Linux Networx has all of those features you describe.
The clusterwide file system problem can be resolved with several file systems, gfs, lustre and your openvms. GFS recently became opensource and is available to any user.
As for the single boot images, Linux Networx uses a cluster administration tool called Clusterworx. Using this tool you can have an os image, kernel and different partition information including a tempfs option for diskless nodes. There is also a tool to keep track of the different os images and the changes made to them similar to a version control system. If you ever need to go to a “known good” state of the os, just check it out from the vcs server.
Upgrades are handled the same way with Clusterworx. If it is a kernel (modular or monolithic) create a new kernel in Clusterworx, and provision the nodes you want to update.
Clusterworx also has a instrumentation panel that shows at a quick glance the status of all nodes in the cluster (cpu load, network activity, disk space and memory).
There seems to be many solutions for your concerns.