Hello Karen -
The question on the usage of replication as a part of disaster recover
plan is quite common. As you correctly pointed out, MS would not recommend
replication but it very well recommends log shipping which you have already
explored. The manual intervention deemed necessary even in the case of
replication. If your publisher site is down, how would you make the clients
connect to the subscriber automatically ? If you can automate this, then I
believe you can automate log shipping as well.
Regarding failover distributor, you can have the option of running the
distributor database in a clustered environment which by default will have
the failover capabilitites.
Regards,
Chandra
--------------------
||From: karensundq@yahoo.com (Karen Sundquist)
||Newsgroups:
microsoft.public.sqlserver.replication,microsoft.public.sqlserver.clustering
||Subject: Using merge replication for high availability (in addition to a
cluster)
||Date: 22 Feb 2004 08:11:58 -0800
||Organization:
http://groups.google.com ||Lines: 34
||Message-ID: <9999c907.0402220811.48e06b0@posting.google.com>
||NNTP-Posting-Host: 194.165.171.187
||Content-Type: text/plain; charset=ISO-8859-1
||Content-Transfer-Encoding: 8bit
||X-Trace: posting.google.com 1077466318 19117 127.0.0.1 (22 Feb 2004
16:11:58 GMT)
||X-Complaints-To: groups-abuse@google.com
||NNTP-Posting-Date: Sun, 22 Feb 2004 16:11:58 +0000 (UTC)
||Path:
cpmsftngxa06.phx.gbl!cpmsftngxa10.phx.gbl!TK2MSFTNGXA05.phx.gbl!TK2MSFTNGP08
.phx.gbl!newsfeed00.sul.t-online.de!t-online.de!newsfeed.gamma.ru!Gamma.RU!n
ewsfeed.icl.net!newsfeed.fjserv.net!c03.atl99!sjc70.webusenet.com!news.usene
tserver.com!sn-xit-02!sn-xit-01!sn-xit-08!supernews.com!postnews1.google.com
!not-for-mail
||Xref: cpmsftngxa06.phx.gbl microsoft.public.sqlserver.clustering:12966
microsoft.public.sqlserver.replication:47622
||X-Tomcat-NG: microsoft.public.sqlserver.clustering
||
||Hi there
||Are any of you using merge replication to increase the availability of
||your sql server? I have a clustered sql server 2000 but I need to
||cover for the eventuality of loss of network connection to the
||datacenter or other event that would put the cluster offline.
||
||At first I set up log shipping but the change over process is manual
||(I have to get out of bed if it fails during the night). Also, it is
||possible to lose some data depending on the frequency of log copies.
||
||I know replication is not recommended by MS as a failover solution but
||it does seem to offer a warmer backup than log shipping and the
||possibility of using both servers in a load balanced arrangement. My
||application is OLTP but the transactions are not that strict - if we
||lose a little data or end up with a half completed transaction
||somewhere it's not the end of the world.
||
||The merge replicated server would sit across a WAN. All hardware
||involved is high powered for the load.
||
||If anyone is successfully running this I'd love to hear about it.
||
||Otherwise, I guess we could get some 3rd part stretch cluster
||solution. Any of you using something like this which allows for
||automatic failover across a WAN and costs not more than $100K?
||
||Final question: the distributor seems to be a single point of failure
||in a replicated failover solution; is there a way to set up a failover
||distributor?
||
||Thanks!
||
||
||Karen
||
