SQL Graphs

Connections

The graph plots the number of database connections to each Transaction Engine (TE) stacked to arrive at the total number of connections at the top line.

Connections.
Figure 1. Connections

Use Case

The number of connections must be the same for all TEs. Increasing or decreasing the number of connections must be investigated. Typical applications would use connection pooling to manage the number and use of connections. Other methods may add variance to the number of connections per TE.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Connections: for raw ClientCncts, sum.

Active Connections

This graph plots the number of connections that have active transactions.

Active Connections.
Figure 2. Active Connections

Use Case

Active connections will increase during periods of workload. Check the history of workload to understand peaks.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Connections: for the value of SqlListenerSqlProcTime, sum

Commits

This graph plots the number of commits per second per TE. Each TE is represented by a different color. The graph is stacked, so the top line is the total number of commits per second.

Commits.
Figure 3. Commits

Use Case

The number of commits per second should rise and fall with the workload. Review workload history to understand peaks and troughs. A single line should not be out of step with others for too long. Asymmetrical load balancing may affect the number of commits to each TE.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Commits: for the rate of Commits, stack

Rollbacks

This graph plots the number of SQL rollbacks per second per TE.

Rollbacks are transactions that are deliberately rolled back and not committed. All rollbacks logged are not user-generated; some are database generated.
Rollbacks.
Figure 4. Rollbacks

Use Case

Rollbacks are workload-dependent. Review workload history to understand peaks and troughs. A single line should not be out of step with other lines for too long.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Rollbacks: for the rate of Rollbacks, stack.

Rows Inserted

This graph plots the number of rows inserted per second per TE.

Rows Inserted.
Figure 5. Rows Inserted

Use Case

The number of rows inserted is workload-dependent. Review workload history to understand peaks and troughs in rows inserted. A single line should not deviate for too long.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Rows Inserted: for the rate of Inserts, stack

Rows Updated

This graph plots the number of rows updated per second per TE.

Rows Updated.
Figure 6. Rows Updated

Use Case

The number of rows updated is workload-dependent. Review workload history to understand peaks and troughs in the updated rows. A single line should not deviate for too long.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Rows Inserted: for the rate of Updates, stack

Rows Deleted

This graph plots the number of rows deleted per second per TE.

Rows Deleted.
Figure 7. Rows Deleted

Use Case

The number of rows deleted is workload-dependent. Review workload history to understand peaks and troughs in rows deleted. A single line should not deviate for too long.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • Rows Inserted: for the rate of Deletes, stack

Number Inserts Pending

This graph plots the time spent waiting for inserts to be completed.

Number Inserts pending.
Figure 8. Number Inserts Pending

Use Case

Typically, pending inserts depend on workload. A heavier workload indicates more time spent waiting for inserts to be completed. Longer or higher periods where inserts are pending may need investigation. These typically correlate to times when the SQL is blocked or locked.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • PendingInsertWaitTime: for the value of PendingInsertWaitTime, average

Number Updates Pending

This graph plots the time spent waiting for updates to be completed.

Number Updates Pending.
Figure 9. Number Updates Pending

Use Case

Typically, pending updates depend on workload. A heavier workload indicates more time spent waiting for updates to be completed. Longer or higher periods where updates are pending may need investigation. These typically correlate to times when the SQL is blocked or locked.

Graph Creation Detail

To create a graph using other tools, use the following detail:

  • PendingUpdateWaitTime: for the value of PendingUpdateWaitTime, average