Devon IT HC12 Manual do Utilizador descarregar pdf (Página 56)

ThinManage 3.1 Administration Manual

Failure Scenarios

Now that we have outlined all the high availability features of the ThinManage Enterprise solution, we will present some

possible failure scenarios that could occur and examine how ThinManage eliminates each of these points of failures.

Figure 3-C: Possible points of failure in the ThinManage environment

Failure Scenario A : If a failure occurs somewhere on the load balancing tier, then requests will not be distributed to

appropriate ThinManage servers. Obviously, a failure on this level would cause a breakdown of services in the system

immediately. Using a load balancer equipped with a hot standby failover mechanism eliminates this possible SPOF. The

Cisco LocalDirector appliance is an excellent example of a load balancing component possessing this type of failover

feature.

Failure Scenario B : A second component that is susceptible to failure is the ThinManage server. Generally speaking,

there are two possible reasons a ThinManage server may go down:

 Planned Events : An administrator may intentionally shutdown the server for maintenance purposes.

 Unplanned Events : The ThinManage server may cease to function due to unexpected reasons, such as

software glitch or accidental powering off of the host server (ie. VMware Server).

Whether the reason is planned or unplanned, the fact that other servers are online and waiting to handle new requests

eliminates any possible disruption of services. The load balancer will recognize that a particular server is unavailable, and

re-route the request to one of the other available servers.

Failure Scenario C: Lastly, the database itself is a potential failure point. The ThinManage servers depend on this

database for retrieving scheduling and other shared information. Thus, losing connectivity to this central repository would

cause serious disruption. Database clustering is the key to eliminating this situation. If one of the server nodes fails, its

responsibilities are taken over by another server in the cluster.

1 2 ... 51 52 53 54 55 56 57 58 59 60

Sem comentários

Devon IT HC12 Manual do Utilizador Página 56