White Papers

20 Dell Storage PS Series Architecture: Load Balancers | TR1070
11 Scenarios in which APLB action is limited
There are several cases when the APLB will not take action. First, if all of the members of the storage pool
are exhibiting low latencies, then the resources do not need to be balanced. In this instance, the workload is
receiving an appropriate level of performance (low latency); it is possible that in a low-resource utilization
scenario such as this, a member with lower performance characteristics could host more active data than a
member with higher performance characteristics. In this case, the APLB merely monitors the latency of the
pool members, but since there is nothing in need of optimization, it takes no further action. This is shown in
Figure 4.
Fair workload distribution
Second, if several members are all supporting a workload and all are experiencing high but similar latency,
then no amount of data exchange will improve the situation where more is being asked of the resources than
they can provide. The APLB will not attempt to optimize arrays when there is no configuration that would
provide improved performance. SAN HQ can be used to observe the situation, and if not a result of network or
server misconfiguration, the likely resolution is to add additional members to the pool so there are more
resources to handle the workload. For more information on SAN HQ, consult the document, Analyzing PS
Series Storage with SAN HeadQuarters.
The third case in which the APLB will not take action is if a member is changing latency very rapidly, swinging
from high latency to low latency and then back again. The APLB will detect this thrashing and will not attempt
to alter the distribution of active data since the application is not behaving in a predictable manner and is not
able to be optimized.