Neoview Workload Management Services Guide (R2.3)

Consider the system resource usage when queries associated with those services request to be
executed on the system:
Maximum Memory
Usage
Memory UsageMaximum CPU
Busy
CPU BusyTotal QueriesService Name
85%60%100%50%80SERVICE1
85%60%100%50%20SERVICE2
85%60%100%50%25SERVICE3
Because the system resources are within the limits (100% for CPU busy and 85% for memory
usage), WMS allows all incoming queries to start executing. If ten additional queries request to
be executed in the low-priority service, SERVICE3, and the system resources reach their limits
when the fifth query starts executing, WMS puts the five remaining queries into the waiting
queue:
Memory Usage
(Limit = 85%)
CPU Busy (Limit =
100%)
Waiting QueriesExecuting QueriesTotal QueriesService Name
86%90%08080SERVICE1
86%90%02020SERVICE2
86%90%0+5=525+5=3025+10=35SERVICE3
Of the incoming queries, if 20 queries associated with the high-priority SERVICE1, 10 queries
associated with the medium-priority SERVICE2, and five queries associated with the low-priority
SERVICE3 request to be executed, WMS puts all those queries into the waiting queue:
Memory Usage
(Limit = 85%)
CPU Busy (Limit =
100%)
Waiting QueriesExecuting QueriesTotal QueriesService Name
86%90%0+20=208080+20=100SERVICE1
86%90%0+10=102020+10=30SERVICE2
86%90%5+5=103035+5=40SERVICE3
As queries in each of the services finish executing, system resources free up, and WMS dynamically
balances the waiting and executing queries based on the percentage of executing queries for each
service and how the percentage compares with the expected percentage, which is based on the
service's priority weight. For example, suppose that 15 queries finish executing in SERVICE1:
Waiting QueriesExpected
Percentage Based
on Priority Weight
Percentage of All
Executing Queries
Executing QueriesTotal QueriesService Name
2056%65/115 = 56%80–15=65100–15=85SERVICE1
1033%20/115 = 17%2030SERVICE2
1011%30/115 = 26%3040SERVICE3
For each service, WMS calculates the percentage of all executing queries and compares it with
the expected percentage for the service. If the percentage of executing queries is equal to or lower
than the expected percentage, WMS allows the queries in the waiting queue to start executing,
starting with queries in a higher priority service and ending with queries in a lower priority
service. For example, SERVICE1 queries have a higher priority than SERVICE3 queries and start
How Services Work 27