Avaya P333R-LB Manuel D’Utilisation
Chapter 14
Load Balancing in the P333R-LB
52
Avaya
P333R-LB User’s Guide
Load Balancing Metrics
There are different methods, or metrics, that a P333R-LB can use to distribute traffic
among multiple cache servers. These metrics tell the router which cache server
should receive the next session.
AR uses the following metrics:
•
among multiple cache servers. These metrics tell the router which cache server
should receive the next session.
AR uses the following metrics:
•
Round Robin.
•
Hash.
•
MinMiss Hash.
The default metric for AR is MinMiss Hash on the destination IP.
Each load balancing metric can be performed on one of the following metric
parameters:
•
Each load balancing metric can be performed on one of the following metric
parameters:
•
SRC IP (source IP address).
•
DST IP (destination IP address).
•
SRC and DST IP (a combination of source and destination IP addresses).
In addition, each of the metrics may be weighted. For full information on weighted
Real Servers, see Weighted Real Servers on page 54.
Real Servers, see Weighted Real Servers on page 54.
Round Robin
Using the Round Robin metric, new sessions are issued to each server in turn, based
on the Real Server weight. The first Real Server in the group receives the first ‘n’
sessions, where ‘n’ is the Real Server weight. The second Real Server receives the
next ‘n’ sessions, and so on. When all the servers receive at least one session, the
issuing process starts over with the first Real Server.
on the Real Server weight. The first Real Server in the group receives the first ‘n’
sessions, where ‘n’ is the Real Server weight. The second Real Server receives the
next ‘n’ sessions, and so on. When all the servers receive at least one session, the
issuing process starts over with the first Real Server.
Hash
Using the Hash metric, sessions are distributed to cache servers using a predefined
mathematical hash function. This hash function is created by a hash value,
performed on a specific parameter. The source and destination IP addresses are
used as the hash function input.
P333R-LB creates a list of all the currently available cache servers. The result of the
hash function is used to select a cache server from the list. Specific addresses always
produce the same hash function, providing natural persistency.
If a cache server is added or removed to the group, the persistency is broken. This
occurs, since the order of the cache servers in the list changes, but the hash still
points to the same list entries. Figure 14.16 illustrates how a loss of persistency
occurs when a cache server becomes non-operational.
mathematical hash function. This hash function is created by a hash value,
performed on a specific parameter. The source and destination IP addresses are
used as the hash function input.
P333R-LB creates a list of all the currently available cache servers. The result of the
hash function is used to select a cache server from the list. Specific addresses always
produce the same hash function, providing natural persistency.
If a cache server is added or removed to the group, the persistency is broken. This
occurs, since the order of the cache servers in the list changes, but the hash still
points to the same list entries. Figure 14.16 illustrates how a loss of persistency
occurs when a cache server becomes non-operational.