[Bioclusters] LSF vs SGE

Chris Smith bioclusters@bioinformatics.org
12 May 2003 08:50:29 -0700


> > Also, both PBS and SGE follow DRMAA, but Platform
> > pushes NPI -- the LSF API as the "standard".
> > 
> 
> Both DRMAA and NPI seem to be coming from the Global Grid Forum (GGF) 
> these days. I know Platform is involved at some level with GGF.
> 
> Anyone from Platform or GGF who can chime in on what the current 
> situation is? What's up with NPI vs DRMAA?
> 

I'll give a bit of a summary from my point of view. 

NPI (which predates DRMAA by the way ... funny ... Sun didn't join NPI
when invited) is focused on defining an architecture for distributed
computing. It is crafted in terms of the "actors" involved and the
"actions" between them (this is quite clear if you look at some of the
early output of the organization, which include Rational Rose models).
Note that the intention is to stay away from a particular language
binding. Yes ... many of the early interactions described used actions
which reflected Platform's API, but that's only to be expected given our
position as one of the founding members, and from our heavy involvement.
The object model (which is what it is) is no longer constrained this
way, which was always the intention when NPI came together. 

>From my understanding DRMAA is intended to define a C API for DRM
systems. It's also interesting that DRMAA is highlighted as a group "of
interest" to the NPI Architecture Working Group, so perhaps it's not
quite "DRMAA vs NPI". It would be interesting to see if the DRMAA API
would suffice as an instance of the object model. I would surmise that
it would fit quite well this way. 

Platform is also not disinterested in DRMAA ... we watch the activities
closely. We see, though, that the trend in the marketplace these days is
more towards supporting Web Services. With this in mind, most of
Platform's energy at GGF is in the OGSA/OGSI area. I personally believe
that it's more important at this stage to define standard web service
interfaces (language neutral) than to standardize on a particular
language binding. LSF has both a C API (had it since day one) and a Java
interface using a back-end SOAP server .... I'm definitely having to
provide more support for Java and SOAP than for C these days. 

-- Chris