HP StorageWorks Enterprise File Services WAN Accelerator 2.1.5 deployment guide (June 2006)

108 10 - SERIAL CLUSTER AND CASCADE DEPLOYMENTS
Serial clustering operates in a spill-over mode where TCP connections beyond the
capacity limit of one of the HP EFS WAN Accelerators in the cluster are automatically
handled by the next HP EFS WAN Accelerator in a cluster. If one HP EFS WAN
Accelerator fails, the next HP EFS WAN Accelerator automatically take over.
Figure 10-1. Serial Cluster
In this example, HP EFS WAN Accelerator1, HP EFS WAN Accelerator2, and HP EFS
WAN Accelerator3 are configured so that they do not answer probes from each other
and do not intercept inner connections from each other. Similarly, HP EFS WAN
Accelerator4, HP EFS WAN Accelerator5, and HP EFS WAN Accelerator6 are
configured so that they do not answer probes from each other and do not intercept inner
connections from each other. The HP EFS WAN Accelerators are configured to
perform auto-discovery so that they can find a peer HP EFS WAN Accelerator that is
available and that is on the other side of the Wide Area Network (WAN).
You specify peering rules that when the HP EFS WAN Accelerator receives an auto-
discovery probe it is passed through, accepted, or processed normally based on the
sender of the probe, the client IP address and subnet, the server IP address and subnet,
or the server port.
In a serial cluster when an HP EFS WAN Accelerator reaches its capacity limit, it stops
intercepting new connections and passes them through to the next HP EFS WAN
Accelerator that will intercept them. This process continues until there are no more HP
EFS WAN Accelerators available in the cluster.
In serial cluster deployments:
The peering rules table is a ordered list of rules and the first rule that matches the
rule is applied.
To avoid interceptions on inner connections created by other HP EFS WAN
Accelerators in the same cluster, in-path rules are specified to pass-through
connections originating from those HP EFS WAN Accelerators.