Managing Replication and High Availability Scenarios › How Arcserve RHA Supports Load Balanced SharePoint WFE

How Arcserve RHA Supports Load Balanced SharePoint WFE

SharePoint WFE (Web Front End) servers can be deployed in a load balanced system to improve WFE performance. In many-to-many configuration mode, each WFE server is protected in a scenario with a dedicated standby for failover. In many-to-one configuration mode all WFE servers are consolidated into single standby server and the scenarios are in a single group to ensure group failover. Since the whole group of load balanced servers is protected as one, only when the last active server is down, the standby is going to be active.

Arcserve RHA provides high availability for these servers in many-to-many and many-to-one configuration methods. Considering that you can choose different kinds of load balance systems (hardware solution or software based solution such as DNS round robin or Microsoft NLB), Arcserve RHA will not handle load balance related logic during switchover such as workload distribution. The load balance system must ensure that your requests can be sent to a new server node and no further service requests are distributed to a removed node.

If you have used Microsoft NLB as WFE load balanced solution, it is easier to redirect the network flow to standby after failover. Consider the following scenarios:

• If you have configured a scenario with redirection-DNS or switchover-hostname property, Arcserve RHA lets you dynamically start or stop NLB based on the scenario status. You can do this by specifying start-db and stop-db scripts.
• If you have configured a scenario with move-IP as network resource, the IP address you specify can be the private IP address of the server or the IP of the NLB cluster (virtual IP).
• For many-to-many configurations, if you want to move the virtual IP from the active server to a standby, you must first make all standby servers part of NLB, and then dynamically start or stop NLB in addition to moving virtual IP to standby. The reason for this is to avoid IP confliction when you are sharing the same IP among several non-NLB standbys.
• For many-to-one configurations, you must move the virtual IP from active to standby and then the standby will host the virtual IP address to respond to user requests.