all groups > sql server replication > february 2007 >
You're in the

sql server replication

group:

Peer to Peer question.


Peer to Peer question. ChrisR
2/6/2007 1:05:01 PM
sql server replication:
BOL link:

ms-help://MS.SQLCC.v9/MS.SQLSVR.v9.en/rpldata9/html/23e7e8c1-002f-4e69-8c99-d63e4100de64.htm

Howdy all. I was just reading up on this topic and have a question. It says
on this page (Topology with two participating databases) that reads will be
spread out between the two servers, thereby giving better performance. But it
doesnt say how that is accomplished. Does the app have to be designed a
specific way, like it would be if updates were going to go between the two
boxes in the same scenario?

TIA, ChrisR
Re: Peer to Peer question. Hilary Cotter
2/6/2007 6:10:50 PM
It's scale out read and write performance. Instead of having 100 users
banging away against one box you have two boxes and 50 of the 100 users
banging away against box 1 and the other 50 banging away against box 2. You
get improved performance because of reduced io contention, lock contention,
and improved concurrency.

--
Hilary Cotter

Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com



[quoted text, click to view]

Re: Peer to Peer question. Paul Ibison
2/6/2007 10:21:58 PM
Hi Chris - I would guess that they are talking about a load-balancing
middle-tier solution, but I can't see the exact link here. Can you post up
the topic header and I'll check out the page on the web.
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com

Re: Peer to Peer question. ChrisR
2/7/2007 5:29:00 AM
I understand why you would want it to be spread out, but I don't understand
how it gets spread put. Does the app need to be written a certain way?



[quoted text, click to view]
Re: Peer to Peer question. ChrisR
2/7/2007 5:31:00 AM
replication [SQL Server], peer-to-peer replication/ Peer to Peer
Transactional Replication.



[quoted text, click to view]
RE: Peer to Peer question. ChrisR
2/7/2007 8:03:03 AM
Thanks guys.

[quoted text, click to view]
Re: Peer to Peer question. Hilary Cotter
2/7/2007 10:23:18 AM
DML which occurs on one server is sent to each node in the peer-to-peer
topology. Each command has an originating server tag on it so it does not
loop back.

The recommendations for writing a client app which uses the peer-to-peer
topology is that all updates are directed at one server to prevent
conflicts, the tables are perfectly partitioned to prevent conflicts and all
systems are quiesced when you make schema changes.

--
Hilary Cotter

Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com



[quoted text, click to view]

Re: Peer to Peer question. Paul Ibison
2/7/2007 2:22:47 PM
Yeah - the article just says "an application can be programmed to route
traffic" without giving details. I suppose this is fair enough really
because it is outside the scope of SQL Server documentation. There are other
sited which deal with the nity-gritty of load-balancing eg
http://www.codeproject.com/useritems/dbloadbalancerservice.asp
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com



AddThis Social Bookmark Button