Running a single application on a cluster
 

Hello Friends,

I have come across an issue. We have an application for the biotechnology. The application is very heavy so we are trying to run it on cluster. We have four Dell workstation 7500. Each is having 32 GB RAM. But I am not getting the exact method of configuring cluster and node. I have tried Conga (Luci, Ricci). My questions are:

1) Is it possible that our application can run on cluster ?
2) If its possible then how shall i configure it?

I need the solution of this problem urgently. Please help me in this regard.

regards
Manoj Kumar

what kind of programming language are you using, you could setup a cluster of 4 machines running an application server supporting the language you use

Your application can probably run on a cluster, but you didn't tell us what problems you were having so I'm not sure how we can help you.

Hi,

If this specific application can be clustered depends a bit on the application itself, but most applications can be clustered. Be sure to check if session fail-overs are also needed (depends on the application).

But is clustering really necessary? Would "simple" load-balancing do? I.e.:

1 load-balancing machine handles all the incoming requests and reassigns these requests to different machines, each running the application.
The load-balancing machine could either be a server or a dedicated load-balancer. Both can be configured to reassign based on different criteria (round robin, asymmetric, content aware, ...).

The machines behind the LB are scalable if needed.

If high availability is wanted then a cluster + load-balancing would be the way to go.

Red Hat can provide either of the above (assuming that you want/need support), but I see you already found your way to Conga and the lot :)

Setting up and configuring a cluster can be tricky, I would advise to have a good look at the documentation and manuals that Red Hat provide. They can help in choosing the correct cluster/load-balancing type and are more then sufficient to actually set it up.

Any specific problems that arise can probably be found on-line, or you can ask here at LQ.

BTW: Clusters always contain an odd number of nodes.

Hope this helps.

Thank you everybody,

As I am more on kernel programming work, I don't know much about this cluster work :confused:. Our application is just a Desktop application (its not a network application). I don't know which programming language has been used to build this one (we have purchased this application from outside). But developer of this application has told me that this application is build to run on cluster also.


I simply want to run this application on the cluster. So do we need to configure our application during the steps of cluster configuration? I went through Red Hat Documents. These are very helpful. But I failed to get the actual solution of my problem.


An application running on a cluster. This thing I want to achieve.

thanks & regards
Manoj Kumar

Hi Drunna,


My application is CLC genomic workbench

Hi,

What is it you actually want to achieve? I.e.: What is the problem, 'cause it seems you already decided on a certain solution that might (or might not) be the way to go.

- Do you want/need high availability,
- Do you want/need failover,
- Do you want/need load balancing.

Looking at the CLC genomic workbench (PDF) document it seems that this, the CLC genomic workbench, is only a small part of the whole CLC Software suite. Which parts can/cannot be clustered? Does it need special configuration options?

Without knowing the specifics of the CLC suite it is hard to come up with a "simple" clustering solution.

I would suggest talking to the CLC support team and ask what solutions there are for your problem. You are definitely not the first one that runs into this problem.

Hope this helps.

Thanks Druuna,

I will first confirm with CLC support Team.