||Visit to Caltech by Project Coordinator
to pre-view the SPP2000 and discuss detailed plans with
Messina's Group and the HEP CMS Group.
||PHASE 0 BEGINS
||Attendance at Objectivity/DB training
course by Project Coordinator and member(s) of Caltech
Install Objectivity/DB on the SPP2000. The
availability of Objectivity 4.0.2 for HP/UX 10 is
promised for before this time.
Install the RD45 program suite on the SPP2000, which
includes programs for populating the database, testing
associations, generating random events as objects,
testing access methods, testing replication. These
require the prior installation of Tk/Tcl, CVS and the C++
Allow access by Objectivity engineer(s) to the HPSS
client library at Caltech, for testing purposes.
|Rick Wilkinson attended the Objectivity
course in Mountain View July 7-9th., paid by Caltech/HEP.
Julian Bunn used the tutorial, printed and online manuals
to teach himself.
Objectivity was installed on the Exemplar on 21st.
The RD45 program suite was not available as a portable
package. Instead, a new application was developed by
Julian Bunn and Roy Williams which allowed the
investigation of database population, object
associations, random object member generation, access
methods, selection methods, replication schemes and
various other scaling parameters.
The RD45 collaboration preferred to do this work with
SLAC. The Caltech HPSS system was, however, used for
various tests with the NFS interface. These tests were
asked for by RD45 and the PDP group.
||Present Project status at CMS week in
||A presentation was made to the CMS
Software and Computing Technical board, on 19th. June.
||Begin performance studies work:
of DB (>100 GBytes disk space will be available for
tests). Functionality tests. Comparison with RD45/DEC
benchmark results (already available by end March '97) on
the "impact of using an ODBMS for the physical
organisation of the data". SMP scalability (machine
partitions, replication between partitions, large numbers
of servers and clients working concurrently), large
memory caching, container placement and size, locking.
This will require one or more visits by RD45 experts.
The above performance studies continue throughout
|The performance studies were begun, and
the results so far are fully documented on the GIOD Web
pages. In addition, the results were presented at a
variety of meetings.
In particular, we have measured
scalability of analysis tasks as a function of the number
of simultaneous clients (up to 158 clients on 158 CPUs).
These tests extended some made at CERN with a 6-CPU
machine. We have explored the creation of large databases
using Objectivity and Versant. With Objectivity, we were
able to create a federation of 32000 databases before
hitting an Objectivity software limit that prevented us
from creating more. We noted increasingly longer database
creation times as the size of the federation increased.
We were unable to access federations containing more than
about 20000 databases due to another Objectivity software
limit. With Versant, we succeeded in creating 3000 shared
databases before running out of space on the disk used
for these tests.
We have also installed the complete OO CMS analysis
environment for the H2 and X5 test beam work. A
replica of the H2 test beam data was made at Caltech. We
define a new milestone for installation and analysis of,
a replica of the X5 data. The X5 data is from tests of
the CMS tracker modules: the set-up is more stable than
that in H2. We will develop Java-based GUIs to the
database, and attempt to re-implement in Java the
required C++ methods, and evaluate the usability of the
Java interface for analysing and displaying the tracker
||REVIEW PHASE 0
PHASE 1 BEGINS
|Julian Bunn visited CERN at the end of
September, and presented the status of the project to the
DPG, the RD45/HPSS meeting, and in various other forums.
There were private discussions between Julian Bunn,
Jurgen May, Les Robertson, Jamie Shiers, Tony Osborne,
Fabrizio Gagliardi during this week, to review progress
and help define new goals for the Project.
||Caltech staff to install and test CMSIM
and set up batch queues for CMSIM production. Acquire
disk space for production runs. Start testing CMSIM
production in around 10% of available capacity, and then
full production in the background on the SPP2000.
Production continues throughout subsequent phases.
||The CMSIM program was ported and
optimised for the Exemplar. By December 1997, over 1
million events had been generated on the machine as part
of an ECAL efficiency study. This work was primarily
carried out by Rick Wilkinson and Sergey Shevchenko of
||Present Project status at CMS Week in
||The project was introduced and reviewed
as one of the main sessions during CMS week,
||Install ODBMS on Caltech in-house AIX and
SGI systems, and test it over LANs in the Gbits/sec
||This was done earlier in the year,
although on HP and NT workstations. The results are
documented fully on the GIOD Web pages. The speed of the
LAN, however, was not in the Gbits/sec range.
new milestone is defined that involves using a dedicated
155 Mbits/sec ATM fibre between the Exemplar and an HP
C200 workstation. We will investigate database object
access across ATM as a function of bandwidth, object
sizes, database size, database location, and so on.
Emulate a possible future WAN connection by simulating
various traffic patterns on the ATM link.
||REVIEW PHASE 1
Present Project status
to RD45 and CMS at CERN (8/12 - 12/12) (CMS Week)
PHASE 2 BEGINS
|The project status was presented during
the RD45 workshop. There will also be a presentation
during the CMS week.
||With Objectivity engineer, PDP expert and
assistance from CACR, integrate the Objectivity/DB as a
client for the HPSS. This assumes that the basic
Objectivity HPSS support is available.
||We expect to obtain shortly the
AMS interface to HPSS from Objectivity and SLAC. This
will be installed on the CACR SP2 (a 24 node machine). We
will thus be participating actively in the AMS/HPSS field
test. We intend to check the HPSS cache hit rate, and
define policies for object placement in the whole system.
||Install and test LHC++ and CMS
OO environment, including IRIS Explorer. Install GEANT4,
preferably with CMS geometry, if available, and make
tests using Objectivity/DB as the event store. Ensure
LHC++ environment functions correctly with the ODBMS.
Make comparison tests with PAW and nTuples?
||We intend to install LHC++ on the
Exemplar, and to assist with porting it to Windows/NT
(our primary desktop OS).
A new milestone is
defined related to our identification of a strong need to
have available the CMS geometry defined in GEANT4. This
will be worked on at Caltech, in collaboration with the
people responsible in CMS for the current CMSIM. In
particular, the geometry of CMS will be converted from
GEANT3 format to GEANT4 using a standard tool. Once this
work is complete, we intend to populate an Objectivity
database with simulated events, and explore the
development of simple reconstruction algorithms which
work with a minimal set of space point information. This
database will be populated with a large number of events,
and scaling behaviour explored.
We have no current plans to compare our results with
nTuples, although this decision could be revised if
||Report on results of ODBMS performance
studies started in Jul '97 and contrast with results
||To be done.
||Install, test and re-evaluate all
components of the environment after the SPP upgrade to
384-512 nodes (if available). This would require visits
by PDP and RD45 experts. Initiate new performance and
||Presently (December 1997) there is a
second Exemplar (of 128 CPUs) installed at CACR,
alongside the Exemplar we have been using. It is possible
that the two machines will be merged, yielding a machine
of 384 CPUs. If this occurs, then this work item will be
We define a new milestone of
extending our tests with multiple database clients from
the present level (158 clients, see above), with the
intent of reaching a target of 500 clients, if possible.
||Implement networked ODBMS over OC12
WAN with the peer machine at SDSC and make
scalability and head-to-head comparison tests.
||For this work item, we in fact prefer to
concentrate in this time frame on our tests with 155
Mbits/sec ATM (see above). We do, however, still intend
to use the peer machine at SDSC as described.
area, in collaboration with RD45, we have made Object
database replication tests between CERN and Caltech using
Objectivity/DB. We will extend these tests to
include SDSC peer machine. RD45 have also asked us to
explore the behaviour of client applications that use the
replicas, when the replicas contain references to
external objects. In this context, RD45 have also
requested that we work with the BaBar experiment to
investigate and test replication schemes for BaBar data
by using the Objectivity FTO and DRO options.
||REVIEW PHASE 2
||PHASE 3 BEGINS
between Caltech and CERN and other large LHC centres.
Replication tests. Caching tests.
|These trials are still planned.
||Test full model implementation targetting
CMS as client. The detailed work plan defined in the
Phase 2 review begins.
||Install ODBMS on SPP3000. Test
performance according to criteria established earlier in
the Project. Compare the performance on this smaller,
high performance, next generation system, with the
SPP2000. Evaluate the system in-house and as a networked
centre with SDSC and/or other sites with multi-TFLOPS
||REVIEW PHASE 3
PHASES AS APPROPRIATE
END '97 - '99 PROJECT