Disaster Tolerance and Recovery in a Serviceguard Cluster

Understanding Types of Disaster Tolerant Clusters

Table 1-1

Comparison of Disaster Tolerant Cluster Solutions (Continued)

 

 

 

 

Attributes

Extended Distance

CLX

Continentalclusters

Cluster

(HP-UX only)

 

 

 

 

 

 

Key

No ability to check the

Specialized storage

No automatic failover

Limitation

state of the data before

required. Currently, XP

between clusters.

 

starting up the

with Continuous Access,

 

 

application. If the

and EVA with

 

 

volume group (vg) can

Continuous Access are

 

 

be activated, the

supported.

 

 

application will be

 

 

 

started. If mirrors are

 

 

 

split or multiple paths

 

 

 

to storage are down, as

 

 

 

long as the vg can be

 

 

 

activated, the

 

 

 

application will be

 

 

 

started.

 

 

 

Data resynchronization

 

 

 

does not have a big

 

 

 

impact on system

 

 

 

performance. However,

 

 

 

the performance varies

 

 

 

depending on the

 

 

 

number of times data

 

 

 

resynchronization

 

 

 

occurs. In the case of

 

 

 

MD, data

 

 

 

resynchronization is

 

 

 

done one disk at a time,

 

 

 

using about 10% of the

 

 

 

available CPU time

 

 

 

and taking longer to

 

 

 

resynchronize multiple

 

 

 

LUNs. The amount of

 

 

 

CPU time used is a

 

 

 

configurable MD

 

 

 

parameter.

 

 

 

 

 

 

32

Chapter 1