Linux - ServerThis forum is for the discussion of Linux Software used in a server related context.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
I have been reading a lot about High Performance Clustering (HPC) but it appears only to be relevant to application specifically designed to take advantage of it, is there a distro that can treat each node in a cluster as a seperate process or processor so tha any application can use the cluster? Or am I being a bit to fanciful here?
Well, yes but you are in such a wide field we can't say for sure what you want or need.
We use hundreds of clusters each day for a specialized industrial task. A master node passes a task to nodes as they become free. The node performs a task and sends results back to the master node. In our use the nodes are 8-way standalone servers so we don't send to a core as such but to a nfs node. The software on each node uses the entire server for it's task generally.
Other common HPC setup's so similar things but they break up a single large task into smaller parts then send them to a node. This differs from our use which is thousands of tasks to be worked.
Sun company had a neat trick where they were able to install a client application to almost any workstation anywhere in the world. The idea was some big company could use the power of every workstation they owned to do processing. Each node could be set for amount of task or times or such to allow the normal user to continue to work.
For the most part, you can't easily use an HPC. You can set them up in a virtual machine or a small cluster of systems in your home if you wish but to use it for some work is difficult unless you are using some scientific program or very technical program that was made to run in a HPC engine.
Common cluster distros may be Rocks, Cluster knoppix. I think something like openmosix (I forget). There are a few others. Some of the enterprise level distro's offer a choice.
Thanks for taking the time to reply, its appreciated and kind of confirmed what I thought ... you have to admit though it would be cool and useful to make the power of the cluster available to the a normal distro front end.
What fueled this was when I was thinking about virtualisation, normal technologies distribution VM's to idividual nodes but what I was thinking is having a mutinode cluster running virtual box and then VM's woulod then be spread accross the cluster. Thus adding harware to the cluster to improver performance and capacity would be seemless.
There are some flavors of clusters for HPC: one is to submit jobs with a batch queuing system to route tasks to a node in the cluster (like ROCKS), but you can also set it up by hand. Even this could be parallelized if your software supports it to use more than one node for each job by PVM or MPI. What you are looking for sounds like a single system image which was offered by OpenMOSIX, but IIRC each process was still bound to a node and could only use local memory. It’s now LinuxPMI. There is Kerrighed to offer a single system image across a bunch of machines too.
I'm a final year diploma student in cse and i'm trying to make and implement an infrastructure which would serve as file server, web server, mail server, news server, database server on old systems on linux-ubuntu based platform.I researched a quite bit and then came to know about this Kerrighed cluster. Would it be of any use and fisible if i build a Kerrighed cluster and deploy these services on them. Any i/p would be helpful for the project.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.