Intergraph WebScale User Manual page 55

WebScale™ User's Guide
During convergence, the hosts continue to handle incoming network traffic as usual,
except that traffic for a failed host does not receive service. Client requests to working
hosts are unaffected. At the completion of convergence, the traffic for a failed host is
redistributed to the remaining hosts. Load balanced traffic is repartitioned among the
remaining hosts to achieve the best possible new load balance for specific TCP or UDP
ports. If a host is added to the cluster, convergence allows this host to take over
handling ports for which it has the highest priority and to receive its share of the load
balanced traffic. Expansion of the cluster does not affect ongoing cluster operations and
is achieved transparently to both Internet clients and to server applications. However, it
may affect client sessions that span multiple TCP connections when client affinity is
selected, since clients may be remapped to different cluster hosts between connections.
WebScale assumes that a host is functioning properly within the cluster as long as it
participates in the normal message exchange among the cluster hosts. If other hosts do
not hear from any member for several periods of message exchange, they initiate
convergence to redistribute the load previously handled by the failed host. Both the
message exchange period and the number of missed messages required to initiate
convergence can be controlled by the system administrator. Their default values are
respectively set to 1,000 milliseconds (1 second) and 5 missed periods. Since these
parameters are not usually modified, they are not included in normal WebScale Setup
dialog. They should be adjusted manually in the registry as necessary.
Adjusting Convergence Parameters
To adjust WebScale's convergence parameters, start the Windows NT registry editor
and select the key
HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\WebScale\Parameters.
The AliveMsgPeriod value holds the message exchange period in milliseconds, and the
AliveMsgTolerance value specifies how many exchanged messages from a host can be
missed before the cluster initiates convergence. You should pick these numbers based
on your fail-over requirements. A longer message exchange period will reduce the
networking overhead needed to maintain fault tolerance, but it will increase the fail-over
delay. Likewise, increasing the number of message exchanges prior to convergence will
reduce the number of unnecessary convergence initiations due to transient network
congestion, but it will also increase the fail-over delay. Using the default values, 5
seconds are needed to discover a missing host, and another 5 seconds are needed for the
cluster to redistribute the load. A total of 10 seconds to complete fail-over should be
acceptable for most TCP/IP applications, and this configuration incurs very low
networking overhead.
Scaling Remote File / Printer Access and Adjusting NBT Support
Parameters
When using the Microsoft network commands (such as net use ...) or when mapping
in remote shares, you can access a WebScale cluster as a whole using a single NetBIOS
machine name. This lets you use WebScale to scale read-only file services or print
51