Re: Db2, Oracle, SQL Server

"Valentin Minzatu" <v.a.l.e.n.t.i.n.m.i.n.z.a.t.u_at_y_a_h_o_o.c_o_m> wrote in message news:tKWdnTrD5KFGxpbfRVn-sg_at_rogers.com...
> I still cannot recollect which IBM document I've seen where it was stated
> that DB2's HA feature does not replicate DDLs or some of them, but I will
> get back once I find it. Consider this one closed for now.
>

There are several flavors of HA in DB2, none of them the same as RAC.

• HACMP cluster - shared disk
• HADR - all SQL replicated via logs
• Q Replication - only DML replicated via MQ Seies
• Data Partitioning Feature (DPF) - Not applicable

HADR does do all SQL (including DDL) on both nodes, anything that hits the logs.

> Per my understanding: DB2 partitions the data at the cluster level (must
you
> have a cluster, parallel server?), therefore each node has its own data
and
> when the query is submitted, one node acts as a dispatcher while the
others
> execute it and then the results are merged in memory. Oracle allows for
this
> as well, but it does not impose it as a limit. One could easily execute
the
> same query in a serial manner (i.e. one single server process retrieves
all
> the data querying each partition). The advantage is that if one of the
nodes
> go down you do not need to repartition the table or move data around at
all.
> It is already available as it is shared by all nodes. - please correct me
if
> i am wrong - I would also like to know if DB2 has the ability to exchange
> partitions without taking the table/partition offline.
>

This would be for DPF. HACMP uses shared disk. Q-Replicaiton and HADR also can use shared disk (but only for the logs or the Q Replciation que). Received on Thu Feb 10 2005 - 09:06:16 CST