Avaya P333R-LB Application Manual page 5

A p p l i c a t i o n 1 — S e r v e r L o a d
B a l a n c i n g
Server load balancing allows you to provide high
availability and overcome bottlenecks caused by
limitations in the servers.
Configuring mirrored servers via load balancing
means that, if one server is taken off-line for any
reason, users will not experience any delays or
interruptions. You can extend this availability to
a number of levels, from servers to server farms
to load balancing switches.
• Round Robin—This
Load balancing can also overcome bottlenecks
scheme directs traffic
created by "server session overload." Without load
to the next server
balancing, the number of sessions that the server
in the list.
can handle may become insufficient for all the users
who are interacting with your organization.
• Hash—This statistically-
based scheme is
By intelligently dividing the workload among
based on a pre-defined
your servers, you ensure that users can continue
algorithm.
to interact in a bottleneck-free environment.
• Response Time-based—
Installing one or more P333R-LB stackable load
This scheme directs
balancing switches as the server farm switch allows
traffic to the least
you to easily implement Server load balancing.
busy server.
The server farm appears as a single Virtual IP (VIP)
address to the outside world, so you can implement
multiple servers for increased throughput and
redundancy with no disruption.
Load Balancing Schemes: You can choose the load
balancing scheme according to your needs. The
P333R-LB offers Round Robin, Hash and Response
Time-based schemes.
Persistency. Persistency is important since it ensures
that users receive consistent and uninterrupted
responses to their requests.
Persistency is especially important in eCommerce
where persistency of a session with the same
server ensures that transactions between the user
and a specific server will not be lost. Persistency
ensures that the SSL encryption session will not be
cut by moving to a different server. The P333R-LB
offers two persistency schemes: Hash, Source/
destination IP-based (and destination Port), and
activity timer-based.
Health Checking: In order to make the load balancing
more effective in terms of availability, we have
incorporated a health checking mechanism into
the load balancer that ensures that traffic is only
directed to "healthy" servers. This vital component
of load balancing is performed by "pinging" the
servers and checking for a response.
Graceful Shutdown: This allows you to take down
a server to perform maintenance without your users
feeling a thing. Old sessions continue until they are
completed, and new sessions are directed to another
server. Once all old sessions are completed, you can
take down the server.
3
Communication without boundaries