Sizing considerations for the BMC TM ART Adapter
The maximum best practice recommended capacity for a single BMC TM ART Central server is 350,000 measurements per hour, as identified on the Administration > System Health page in the BMC TM ART user interface. This capacity also assumes that:
- Both raw and 15-minute interval aggregate information is kept for less than 30 days.
- Data deletions are executed at least 2 or 3 times a week, or even daily.
- Online (that is, non-blocking) reindexing of indices for the SV_TimeSeriesData table is executed periodically as measure write time requires.
- Periodic reorganization of the SV_TimeSeriesData table is performed if this table becomes overly fragmented due to infrequent data delete executions under high volume.
For further information, see "Repository Maintenance" in the BMC Transaction Management Application Response Time Administration Guide.
In addition, the BMC ProactiveNet Server is limited in the number of monitors and attributes that it can process per time unit, regardless of their source (whether from the BMC TM ART Adapter or from any other adapter). You can use the following table and formula to estimate the number of monitors and attributes received from the BMC TM ART Adapter:
Monitor instance type (in hierarchical order) | Number of attributes per instance |
|---|---|
BMC TM ART Device | 0 |
Project Monitor (multiple discovered) | 5 |
Location Monitor (multiple discovered) | 2 |
Transaction Monitor (multiple discovered) | Up to 126:
|
Total number of attributes per polling cycle = 5 x number of projects + 2 x number of locations + (6 + number of custom timers x 2 + number of custom counters) x number of transactions