Exchange Geographical cluster

Today most of the organizations invest millions of dollars for IT to enhance productivity and business continuity. To get the return of investment definitely those IT systems has to play major roll in their business operations. Once those systems integrated to organization business process of failure of those systems heavily affect the business. Business continuity is very important factor for any organization that is one of the key areas for IT vendors to focus.

 

In this article discuss about business continuity of one of the major IT application that is Microsoft Exchange server.

 

Exchange server plays major roll in present IT market as dominant mail server. There are couple of ways to make exchange server available all the time.

 

Option one is Exchange cluster running on single site, it provide high availability within the site but in case of site failure whole system will goes offline. 

 

Option two make Exchange server geographically available.

 

There are few third party Exchange replication tools, which enable to replicate data among geographical sites and automate the failover in case of site failure. There are some doubts about third-party replication software’s like data integrity issues, time taken to release of new version updates and patch updates for new Microsoft patch update and version update.  

 

Microsoft Exchange geographical cluster can be implemented without using third party products.

 

Microsoft always recommends going for majority node clustering, to implement MNS cluster three sites required because in MNS cluster always majority number of nodes should be online, if cluster doesn’t have majority it will go offline. 

 

In two site MNS cluster only one site can have majority number of nodes, if that site fails cluster will go offline because other site doesn’t have majority.

 

In three site MNS cluster always two sites can keep majority, single site failure cluster does not affected the cluster, but in case of two sites failure cluster will go offline. Failing two sites simultaneously will be a rare situation.

 

In MNS cluster no need to have Exchange server and storage in third site; third site will act as only a witness. There should be SAN level replication between two Exchange server sites to replicate information and always recommend having 500ms maximum latency between three sites.

 

In two site cluster scenario there is high chance of split-brain, because in case of WAN link failure two sites are unaware about other sites but in MNS cluster this can be avoided.

Published 06-22-2006 6:18 PM by Malinda Pererea
Filed under: