Hi everybody.
SERVER_A: Windows 2003 Server with SQL Server 2005 SP2. This is the
corporative server for all our data. About 30 tables divided in 4
publications. Each publication contains the tables belongs at the same
database. This is the suscriber of one publication I describe below.
SERVER_B: Windows 2003 Server with SQL Server 2005 SP2. This is the
suscriber of the 4 publications and have one publication of 4 tables. This
server is the distributor too.
SERVER_A and SERVER_B are connected to a LAN (100MB) at same switch.
Latency measured by Replication Monitor in the Publications of SERVER_A are
excelent (2 or 3 seconds). In the other side, latency in the publication of
SERVER_B are bad, critical or fatal (30 seconds, 1 minute or more).
I don't know how to measure the changes that we have in the tables belongs
to publications in SERVER_A but I remenber that this are the tables of our
corporative environment and I presume that they are so much.
Changes in publications of SERVER_B are in order of 100.000 rows in 120
seconds (I know it well because this server runs a process created by myself).
How it's possible that both servers sharing the same network the latency are
so different?
How can I decrease this latency?
Thanks a lot.
Paul Ibison - 23 Aug 2008 18:06 GMT
I'm not too sure from the description - are the number of changes made on
the 2 servers the same? Have a look at the size of the msrepl_commands table
on eash server (or run sp_browsereplcmds) to get a feel for the commands
waiting to be replicated.
HTH,
Paul Ibison (www.replicationanswers.com)