Focus Effort Title: HLA Integration for HPC Applications applied to CMS.

Thematic Area(s): Scalable Computing Migration

PI Name: David Bernholdt

PI EMail Address: bernholdt@npac.syr.edu

PI Telephone: 315 443 3857        PI Fax: 315 443 1973

Project Description:

This proposal is in response to two related goals. Firstly the need
for message passing based coarse grain parallelism for CMS and
secondly integration of classic HPC applications into FMS
simulations. The latter is essentially equivalent to HLA compliant
versions of these applications. Both goals can be elegantly addressed
by using a computing environment built around distributed (or meta)
computing built on top of HLA. As the associated runtime RTI supports
time management, this builds in a powerful scheduling model. We
describe below, a prototype of these ideas built around the CMS
application.  The basic infrastructure can be applied to applications
built under our "web interface" tasks to integrate them into HLA
compliant simulations.

From the parallel architectures point of view, this project will
develop distributed memory version of CMS which can be then easily
ported to distributed memory MPPs such as SP-2.  From the
metacomputing management point of view, this task develops and tests
tools for runtime management of distributed resources, initially
prototyped in a controlled commodity cluster envioronment, and in the
next stage to be expanded to a multi-MSRC metacomputing
platform. Finally, from the parallel hardware availability point of
view, this task will effectively bring NPAC commodity cluster,
including a mix of Linux and NT PCs, as one of the HPC resources
available for our Metacomputing CMS experiments.  By combining the
metacomputing aspect and the parallel computing aspect of WebHLA, it
will be possible to consider large-scale CMS simulations involving
instances of the code running on multiple HPC systems at multiple
sites, with each instance being parallel either in a distributed
memory fashion via WebHLA or a shared memory fashion via the separate
parallel CMS project and past work.

The WebHLA system will operate in a cluster (distributed computing)
environment as follows. Each node of such cluster will run JWORB with
OWRTI service. Hence, each node can represent itself as a system level
federate, participating in the Cluster Federation (CF). All user level
federates on a given node join Node Federation (NF), managed by local
JWORB server on this node. Finally, all federate modules forming a
particular distributed application join Application Federation (AF). A
set of heartbeat monitors, running across these three federations
assures high availability of the whole cluster, failover support for
all nodes and fault tolerance for all distributed applications.

Such a WebHLA based Cluster Management service will be then ported to
CEWES and ARL MSRCs to support Multi-MSRC Metacomputing CMS
application running continuously or on-demand in a robust,
fault-tolerant, highly-available self-sustained mode. The proposed
WebHLA based Meta-Cluster infrastructure is expected to be also useful
for other HPC applications that need to federate distributed
multi-MSRC resources.

Benefits:

The DoD has mandated use of HLA for modeling and simulation
applications in a short time frame (by 2004).  The increasing use of
high-fidelty physics-based simulations in conjunction with what might
be considered more traditional FMS applications means that a growing
number of DoD HPC applications will have to become HLA compliant in
the coming years.  This project will result in tools that will make it
easier for HPC applications and HPC systems to take part in HLA-based
modeling and simulation exercises in accord with the DoD mandate.

Require Resources:

Deliverables:

o WebHLA based cluster management including CF, NF and AF Object
  Models (Jun'99) 
o Distributed CMS running on NPAC commodity cluster (Sep'99)
o NPAC commodity cluster included into Metacomputing CMS experiments (Jan'00)
o Expanding the cluster mgmnt tools  to metacomputing environment in 
  support of multi-MSRC parallel CMS (Mar'00)
o Technical Report (Mar'00)

Notes:

To install and maintain a WebHLA installation as described in order to
support on-demand use will require a modest amount of assistance from
on-site staff of the CEWES MSRC.  More specifically,
a) A CEWES system administrator or other appropriate individual to
   assist with the installation of WebHLA, ModSAF, and CMS (expected
   to require no more than a week)
b) Routine on-site support for JWORB acting as WebHLA CEWES Amassador
   (similar to maintaining a web server with small, fixed number of
   static pages)