The TPC-DS subcommittee continued
its work on developing TPC-DS in the areas of queries, data
generator
and schema. The current data generator was advanced to generate
data in parallel. Bugs found by members were addressed and
the subcommittee decided to move to weekly drops to accelerate
development time.
A proposal to reduce the schema to speed up benchmark
development time was made. However, the subcommittee felt
that the rich
schema is needed to assure that auxiliary data structures
are accounted for appropriately.
Reviewing the existing query
set the subcommittee increased the number of ubertemplates
by 31 to a total of 53. In preparation
to generate prototype data the subcommittee expanded 31
ubertemplates to 93 queries. This brings the total number
of queries to
159. Contingent on solving compilation issues with dbgen,
the following
companies promised to deliver prototype data for all 159
queries by the next meeting in June 2004: HP, Sun, Unisys,
Microsoft,
Oracle and IBM. The prototype data includes building a
100GB database, investigating query results and presenting
normalized
elapsed times for all queries.
Going forward the group
is evaluating the existing queries to finalize the query
set
and reviewing
the data maintenance functions to be able to perform
full benchmark runs. This is necessary to make a final decision
on the metric.
|