(DI-2111) Complete List of Default KPIs

The Average CPU Load (1, 5, and 15 minute averages) is the average number of processes that are ready for execution, but which must wait for the CPU.  In SAP, you can see this indicator with OS07. If the average number of processes waiting is higher than the number of available CPUs, this indicates the CPU is reducing the overall performance of the system. "CPU load %" is the CPU load divided by the number of CPUs available on the hardware. Values are captured for the database and for all application servers. The chart displays only the CPU of the database. There are many AS and CPU of AS is usually not an issue.
But a bad exec plan can increase dramatically CPU load of the DB.

any server

This KPI is collected using HANA System View M_UNDO_CLEANUP_FILES.
It represents the count of "files" in state CLEANUP, which ideally should be as small as possible. With this KPI we can be alerted early when the Garbage Collection is impacted or isn’t working. Also, if the number is high, a database restart will be slow as it has to delete all temp files and possibly transactions rollback.

Each undo or cleanup file in the system is represented by one row in this view.
Undo files contain information needed for transaction rollback. These files are removed on the transaction end. If data is deleted but must still be accessible because of MVCC isolation, the corresponding information is written to cleanup files. On the transaction end cleanup files are passed to history management. Garbage collection uses the cleanup files to finally remove data. Undo files and cleanup files may be cached and reused because of performance issues. Cached files get assigned the type "FREE". Undo files for a row store get the type "EXTERNALUNDO".

HDB_SQL_EXP_STMT_NUM

Number of expensive statements exceeding the memory limit

Expensive statements are individual SQL statements whose peak memory size exceeds a configured threshold during their execution. The threshold is configured as a collector's parameter MEMORY_SIZE_LIMIT_IN_MB with the default value 500MB. The KPI represents a number of expensive SQL statements which are recognized from the last monitoring snapshot. The SQL statement is stored in the Detail table "HANA expensive statements" with additional information as are - which application or user executed it, how long it was running, and maximum memory consumed by the execution.

HANA

SYS

any server

Remote system

/DVD/MON_HDB_CL_COL_SQL_STMT

HDB_5MIN

5 MI

KPI HDB_THR_MC_CPUT is a maximum ratio between the delta CPU time of the thread and the number of seconds between 2 snapshots on each host. It’s a number of parallel threads that are processed by MemoryCompactor (also called incorrectly Garbage Collector).

When “defragmentation” / thread MemoryCompactor is running, ERP is slow and can be extremely slow as defragmentation requires an IPMM lock (it can create locks on NRIV or other critical tables).

HDB_THR_MC_CPUTH is a ratio between the delta CPU time of the thread and the number of seconds between 2 snapshots on each host. It’s a number of parallel threads that are processed by MemoryCompactor (also called incorrectly Garbage Collector).

When “defragmentation” / thread MemoryCompactor is running, ERP is slow and can be extremely slow as defragmentation requires an IPMM lock (it can create locks on NRIV or other critical tables).