Features and Benefits of HSG80 ACS 8.6P Data Replication Manager

Features and Benefits of HSG80 ACS 8.6P Data Replication Manager 15

As shown in Figure 8, there are two servers at each site. These site-specific pairs of servers could

be clustered during normal operations to provide highly available applications. When the failure

of one of the sites occurs, the application that was running at the failed site is moved to the

backup member of the surviving cluster and started using the surviving storage and the other half

of the surviving cluster.

With additional servers and storage, the bi-directional DRM can scale to 20 switches in each

fabric supporting, for example, 96 servers and 8 arrays at each site. All intersite technologies such

as DRM over ATM, direct fiber, IP, and WDM support bi-directional use of DRM.

Host A

Host B

Host W Host X

Switch A

Switch Y

Switch B

Switch Z

Storage

Controller A

Storage

Controller B

Storage

Controller Y

Storage

Controller Z

Initiator

Target

Initiator

Figure 8: Bi-directional DRM configuration

Stretch Cluster Support

DRM supports stretched Microsoft Cluster Servers (MSCSs) running Windows 2000 or

Windows NT. This means that half the cluster is at the primary site and the other half is at the

alternate site. Should the primary server fail, MSCS fails over the application to the surviving

server at the alternate site and resumes operations using the primary site storage.

Applications running in a stretched cluster in failover mode incur a performance penalty because

of the time it takes to read or write data across the intersite link. This performance penalty is

directly proportional to the distance between the two sites. The greater the distance between the

sites, the greater the penalty. Tests have shown that almost no impact was observed with

separation distances up to 100 km (62 miles).

For more information on stretch cluster support, see the Compaq ProLiant HA/F500 website at

http://www.compaq.com/solutions/enterprise/ha-f500.html