Technical data

Brocade ServerIron ADX Advanced Server Load Balancing Guide 97
53-1003441-01
Traffic distribution based on cache server capacity
2
NOTE
This command generally disrupts the existing cache persistence.
Traffic distribution based on cache server capacity
In order to determine the health of a cache server, the Brocade ServerIron ADX performs basic
health monitoring functions such as check IP connectivity and check reachability of application
ports. Basic monitoring of these metrics however, may not provide an aggregated view of the health
of cache servers. ServerIron ADX utilizes an advanced capacity check mechanism for Blue Coat
CacheFlow servers by polling the state of Blue Coat servers through SNMP.
The capacity check feature is available with Layer-4 switching of TCS traffic. Starting release
12.4.00a release, it can also be enabled in environments where Brocade ServerIron ADX is
configured for cache selection through application-specific Layer-7 rules.
The following are some guidelines for using this feature:
• This feature can only be used for HTTP traffic over port 80.
• By default, the cache server state is set to OFFLINE. If an out-of-range SNMP MIB value is
received, the cache server state is set to the default state (OFFLINE).
• Only SNMPv2 is currently supported
• Layer 7 SNMP functionality (as shown in Table 11) only applies at the time a new server is
selected. It does not affect existing connections.
This feature operates by performing queries to cache server states that are held in the SNMP agent
on the cache server. The ServerIron ADX then uses the cache server state obtained from the query
to determine the load balancing action to take, as described in Table 9.
TABLE 9 Load balancing action as determined by cache server state
Cache server states ServerIron ADX load balancing action
UNDERUSED (1):
Cache is operating normally with low resource
utilization.
For UNDERUSED (1) or NORMAL (2) or BURDENED (3), the
action is the same:
The ServerIron ADX will allow new connections to these
cache servers and will also forward connections that were
originally hashed to other stressed cache servers to these
cache servers.
NORMAL (2):
Cache is operating normally with no resource
constraints.
BURDENED (3):
Cache is operating normally but nearing
resource constraints.
STRESSED (4):
Cache has reached its effective capacity limit
and cannot stably handle additional load.
The ServerIron ADX will not allow new connections to
these cache servers if other under used or normal cache
servers are available. If no under used or normal cache
servers are available, ServerIron ADX will continue to use
these cache servers.
OVERTAXED (5):
Cache load has exceeded its maximum
sustainable capacity and should be decreased.
For OVERTAXED (5) or OVERLOADED (6), the action is the
same:
The ServerIron ADX will not allow new connections to
these cache servers but will continue servicing existing
connections. Any new connections will instead be
forwarded to the under used and normal cache servers.
OVERLOADED (6):
Cache load has reached an unsustainable level
and must be decreased immediately.