Home | Contact Us | FAQ | Search & Site Map | Link to Us
Sign In | Join | Other 45 Sites in Network
Home
Discussion Groups
DB Engine
SQL ServerMSDESQL Server CE
Services
Analysis (Data Mining)Analysis (OLAP)DTSIntegration ServicesNotification ServicesReporting Services
Programming
CLRConnectivitySQLXML
Other Technologies
ClusteringEnglish QueryFull-Text SearchReplicationService Broker
General
Data WarehousingPerformanceSecuritySetupSQL Server ToolsOther SQL Server Topics
DirectoryUser Groups
Related Topics
MS AccessOther DB ProductsMS Server Products.NET DevelopmentVB DevelopmentJava DevelopmentMore Topics ...

SQL Server Forum / Other Technologies / Replication / February 2007

Tip: Looking for answers? Try searching our database.

multi-master 'merge' replication

Thread view: 
Enable EMail Alerts  Start New Thread
Thread rating: 
Timi - 26 Feb 2007 11:04 GMT
SQL Live-Live – what are the options for the DB clusters to be in an
ACTIVE/ACTIVE configuration with multi-master “merge” replication? What is
the network requirements, what about data integrity, latency, performance?

Basically I am trying to design a new SQL Database infrastructure to support
our webfarm which spread across two geographically dispersed sites, and
trying to make sure that the data is load balanced across the two sites.
Paul Ibison - 26 Feb 2007 11:37 GMT
Hi Timi,
I'm not too sure what you mean by "multi-master" in your description - merge
operates with a publisher-subscriber setup (although alternative sync
partners cloud the issue slightly) and presumably the active/active cluster
represents the publisher with the subscriber being at a remote site. I
haven't set up the configuration you mention, but I see some issues. The
load balancing will need to be done by a middle-tier (ie not out of the box)
and this is fine, but what I would be concerned about is the conflict
resolution. Do you intend to partition the data accross the 2 merge
replicated databases and have the load-balancer determine which partition is
accessed? If not there will be a strong possibility of conflicting updates.
If this is allowed in the business model then it sounds fine.
Latency/performance in your setup will depend on many factors and I would
recommend testing empirically. Apart from that I'd set up large ranges for
the identity columns as per usual.
Cheers,
            Paul Ibison SQL Server MVP, www.replicationanswers.com
Timi - 26 Feb 2007 14:37 GMT
Hi Paul,
Thank you for your prompt reply,

I at setting up a multi master bidirectional replication scenario where both
servers can work autonomously in case of a DR situation, or planned downtime

> Hi Timi,
> I'm not too sure what you mean by "multi-master" in your description - merge
[quoted text clipped - 13 lines]
> Cheers,
>              Paul Ibison SQL Server MVP, www.replicationanswers.com 
Paul Ibison - 26 Feb 2007 15:09 GMT
OK Timi - I think I understand your case but BTW terminologywise
multi-master is not normally a phrase applied to replication, and
bidirectional normally refers to transactional. Also you now mention DR but
originally mention load-balancing, so if it is just DR then my comment about
conflicts being an issue can be disregarded.
For DR, as long as the identity ranges are large and the publication's
retention period and history retention period are both sufficient, then
offline access could continue.
Cheers,
            Paul Ibison SQL Server MVP, www.replicationanswers.com
Timi - 26 Feb 2007 16:51 GMT
Hi Paul
THANKS AGAIN FOR YOUR REPLY

THE SENARIO HAS CHANGE SEE BELOW

SQL Server database load-balancing

a way to load-balance across the SQL Server groups in each data centre to
make failover automatic if a SQL server stops responding

Solution must provide a way for Operations to manually take SQL servers
in/out of the load-balanced group without requiring any action by another
group.

Solutions mustn’t impact performance by >5%

> SQL Live-Live – what are the options for the DB clusters to be in an
> ACTIVE/ACTIVE configuration with multi-master “merge” replication? What is
[quoted text clipped - 3 lines]
> our webfarm which spread across two geographically dispersed sites, and
> trying to make sure that the data is load balanced across the two sites.
Paul Ibison - 26 Feb 2007 16:58 GMT
OK - then the data isn't partitioned in any way and everyone can change
everything. In this case I'd expect the main problem to be one of coping
with conflicts. The load-balancing component could possibly mitigate this by
determining which users to send to which server eg UK users will edit uk
data so they go to the UK merge replicated database, similarly for US users.
When the load-balancing component detects a failed heartbeat, all users go
to the same server.
Cheers,
            Paul Ibison SQL Server MVP, www.replicationanswers.com
 
Sign In
Join
My Latest Posts
My Monitored Threads
My Blog
My Photo Gallery
My Profile
My Homepage

Start New Thread
Enable EMail Alerts
Rate this Thread



©2009 Advenet LLC   Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.