[Beowulf] : Q&A

[Mark Hahn]

> 1. Cluster by ip , e.g a cluster with pc placed in different locations and
> each location does a different task based on its ip (the task has not been
> specified )

sure, I've certainly used IP to provide a useful mnemonic for the "role"
of hosts within a cluster.  but I'm worried about your use of "location"
here: HPC clustering is normally done in a single location.  if you're 
into clustering across geographically significant distances, then your 
latency becomes high and therefore you're probably doing grid stuff.

> 2. Mobile clusters : any one by any small chance looked into clusters over
> mobile application such iphone or andriod ?

wifi, let alone 3g, would introduce dramatic issues of latency and
bandwidth.  I don't think I'd that clustering.

> things could be headed in that direction , you never know. although it could

eh?  nothing much has changed in HPC clustering - incremental improvements
to networks is all.  you might be more interested in ad-hoc mesh-oriented 
networking, which is very, very different (in context, purpose, etc).
if you pursue this, I expect you'll find that there's a huge cultural gap
between telco-oriented stuff like SMS, and "proper" networking (IP/TCP/etc).

regards, mark hahn.