SQL Server Failover without requiring network storage or SAN
Failover is the process specifically associated to cluster
systems. Typically it means when the Primary station of a cluster
system dies, the Secondary station will be activated to take over
the active duty of the Primary station. XLink's ClusterReplica
software will monitor the systems in the cluster system and the
pre-defined services to determine if the cluster system is in the
situation that a failover is required.
XLink's ClusterReplica MSSQL edition binds two systems together
directly. For this reason, failover can take place without the
involvement of a network
storage/SAN or any type of independent disk array system like does Microsoft's cluster system.
Failover conditions
ClusterReplica makes decisions on activation of failover on three
types of conditions:
- System failover - This happens when the entire Primary system
is down
- Service level failover - When any one of the monitored
services is not functioning normally
- The network card failure - When the network card on the
Primary station for public connection fails, failover will take
place
Detailed explanations of these conditions are listed below.
- System Level Failover
The first picture (1) below shows a normal working situation
of the cluster server. When the Primary station fails, the
Secondary Station will become the Primary station and work in
STAND ALONE mode as seen in picture (2).
- Service Level Failover
Failover will take place when one of the services fails to
function normally. In this situation, both clustered systems
will be running. However, their roles of ACTIVE and STANDBY will
be switched automatically.
For example: You set MSSQL
server service for failover. One day, it somehow stopped working
on the Primary Station. Then failover takes place - the
Secondary Station is now taking over the ACTIVE role and
becoming the Primary Station. (as in picture (1) and picture
(3)) However, data replication for the SQL database, in this
setting, will not work. So, fix up the SQL server should be your
first priority.
NOTE: when the ACTIVE station is
changed to become the STANDBY Station, you know there is a
service or application on that system failed to function
normally. You should exam the Secondary station and fix the
problem without delay. Resetting the cluster system is also
important.
- When the network card on the Primary station used for public
connection failed, failover could also take place. The Secondary
station will be activated and network clients would be able to
connect to the server through its network card. See
Cluster Setup for configuration details.
Here are some pictures to provide a general description of Failover (1)(2) and Failback (3)(4).
 
 
Back to ClusterReplica Home
page
|
