[Beowulf] Why need of a scheduler??

[Mark Hahn]

> 1) Do I really need a cluster scheduler installed on the cluster? Should I
> use scheduler?

even for a dedicated cluster, there is some appeal of a queueing system,
since it lets you run concurrent jobs more easily, and to queue a set of 
jobs which wait until resources are available.

> 2) Is there any effect/benefit on the running of a parallel code with or
> without cluster job scheduler?

no, unless the scheduler (or monitoring tools) are inefficient.

> 3) How you differentiate between cluster scheduler and cluster resource
> manager?

theoretically it would be possible to create a pure cluster-controller 
(resource manager: starts, suspends, stops jobs; maintains node status)
separated entirely from any mechanism to choose which job to run, where, when
(scheduler).  I'm not convinced there would be any real point to this.

> 4) If there is any significant difference between a scheduler and manager
> then plaese tell me that which of the fall in which category:
>
> OpenPBS, PBS Professional, SGE, Maui, Moab, Torque, Scyld, LSF, SLURM etc.

I think of scyld as more of a cluster platform/infrastructure - 
I don't know whether it actually provides a queueing system.
all forms of PBS/SGE/LSF are functionally equivalent; slurm is 
perhaps a little more stripped down - closer to a res manager.

> 5) What is maent by  " PBS/SGE/LSF supports integration with the Maui
> scheduler?

it means that the people responsible for product X have added enough
goop to attach Maui to their system.  PBS/SGE/LSF are all large systems
which contain their own schedulers, but if you want even more complexity,
you can glue on another ;)

> Precise, easy and brief reply requested.

IMO, it's bad taste to be picky when asking for help.