Database Reference
In-Depth Information
7.3
Performance monitoring dashboard
You can build a dashboard to make monitoring your HANA database much easier. Before
you learn how to build a performance monitoring dashboard in Design Studio, you need to
understand what the key performance indicators (KPIs) are for HANA.
7.3.1
HANA key performance indicators
Some of the KPIs for HANA are easy to monitor from HANA Studio. The memory usage,
CPU usage, and alerts are all readily available from the
O
VERVIEW
tab of the administration
console in HANA Studio, as shown in Figure 7.5.
You can even take a more detailed look at the memory usage in HANA Studio by selecting
the
M
EMORY
O
VERVIEW
in the context menu for any HANA system that is connected.
There are, however, limitations to this memory overview. An overview of memory in use
across all hosts is not part of the memory overview, and any memory used by temporary
computations is also not visible.
In addition to the memory usage, CPU usage, and alerts that are mentioned above, you
should monitor two more KPIs to have a more complete understanding of the health of your
HANA system. You could be forgiven for making the mistake of assuming that I/O through-
put has limited importance in an in-memory database like SAP HANA; however, the HANA
appliance still has to occasionally read and write data to disk during the transaction com-
mit process. These transactions are usually processed asynchronously, but may require syn-
chronous I/O in some circumstances. Both synchronous and asynchronous I/O can lock im-
portant data structures and cause the system to become unresponsive.
The speed of I/O throughput in your network hardware can also limit the speed of your
HANA system. During one large DMO project, after the development environment was
migrated, we encountered unanticipated slowdowns in the operation of the HANA system
which were ultimately traced to a failing network card which was limiting the I/O through-
put. For these reasons, it is important to carefully monitor the I/O throughput in your HANA
system.
You should also monitor the response time for your HANA system — this depends on the
percentage of database read time to the overall response time for your system. The new stat-
istics server functionality in the latest release of HANA records the execution time for SQL
