Monitoring Tables

Monitoring Table Alarms in the Control System

See Viewing Active Table Replication Alarms.

Viewing Throughput by Operation Type Using the Control System

Log in to the Control System and do one of the following:

Go to the Summary tab in the table information page.
The Throughput - By Op Type pane displays a graph for the following operations on the table in the last hour:
- Gets
- Puts
- Scans
- Increments
- Appends
- Checks and Puts
- Updates and Gets
Go to the Summary tab in the index details page.
The Throughput - By Op Type pane displays a graph for the following operations on the index in the last hour:
- Puts
- Scans

You can move the cursor over the graph to view the number of operations on the table (across nodes).

Visualizing Table Metrics in the Control System

In the Control System, you can visualize node-level metrics for table operations on a node. You can view charts that show metrics for all tables across all nodes, metrics per table aggregated across nodes, metrics per table per node, and metrics per node across tables. In addition, you can view metrics for activity by indexes and for the table and its secondary indexes.

For the full list of metrics (and associated charts/columns) that you can view in the Control System, see Viewing Table Metrics in the Control System.

Log in to the Control System and do one of the following to view table metrics:

Go to the Metrics tab in the node information page and select Activity by Tables from the drop-down menu.
In this tab, you can view charts/columns that show metrics for all table operations on the node, operations on streams, and other node metrics described here.
Go to the Metrics tab in the Data > Tables page.
In this tab, you can view charts/columns that show metrics per node for all tables.
Go to the Metrics tab in the table information page.
In this tab, you can view charts/columns that show operations on the table and its secondary indexes across nodes.
Go to the Metrics tab in the index details page and select Activity by Nodes from the drop-down menu.
In this tab, you can view charts that show all index operations and index operations per node.

You can select the charts to view by creating a custom chart Board or by modifying an existing Board. See Creating a Custom Board for the Charts or Editing a Chart Board for more information.

The charts on the page show the metrics for the last 24 hours by default. You can select a preset or specify a custom time range.

You can also zoom in (by clicking and dragging the cursor in the pane) for a more granular view. Click Zoom Out to expand time window or click:

to shift time window forwards.
to shift time window backwards.

Note: When you select a granular view, the chart might not show the most accurate data because of the difference between the interval at which the metrics are logged and the downsampling (lowering the sampling rate) for the interval being viewed.

You can switch to a list view, where available, by clicking . In this view, you can select metrics to view by clicking . See Adding and Removing Columns from the List View for more information.

In the list view, you can:

Filter the list by clicking .
Click the node name to go to the node-level metrics page.
Select one or more nodes and switch to charts view by clicking to visualize metrics for the selected nodes only. The legends for the charts reflect the selected nodes.

Note: Certain grouping of metrics, available in the charts view, are not available in the list view. All the metric grouping available in the list view are available in the charts view as well. When you switch from charts to list view and vice versa, your selection of metrics in one view is not carried over to the other view.

The chart and list views allow you to detect and diagnose bottlenecks and performance issues on individual tables and nodes. You can use the charts for measuring the throughput and latency of different RPC operations on a table and for determining which operations on a table are slow or which tables are most frequently accessed.

For example:

Suppose your node is busy and you are noticing intermittent latency spikes on your table. You can compare throughput and latency in the Metrics tab of the table information page and investigate if the latency spike is due to node being very busy or node having high CPU utilization by switching to the list view from where you can navigate to the Metrics tab of the node details page.
Suppose your index queries take minutes instead of seconds to complete. You can compare the get latency percentile with the scan read/response in the Metrics tab of the table information page by zooming in to the area where you see the spike. Switch to Activity by Index to view the index vs primary table scans where you can determine whether excessive scan load went to the primary table.
Suppose you are noticing latency spikes on one of your table as a result of a lot of activity on another table. Observe the get latency percentile spikes in the Metrics tab of the table information page and switch to Activity by Node list view to identify the nodes with high overall aggregate table RPC load and node IOps. Select the saturated nodes in the list view and switch to the chart view. Go back to the list view and click the saturated node to navigate to the Metrics tab of the node details page. Switch to Activity by Tables to determine the most active table.

Viewing Secondary Index Metrics

You can visualize the secondary index metrics in the Control System.

Log in to the Control System and do one of the following:

Go to the Metrics tab in the table information page and select Activity by Indexes from the drop-down menu to view all index-related activity on the table.
Go to the Metrics tab in the secondary index page to view metrics for index-related activity across tables or per node.

By default, the page displays metrics for the last 24 hours. You can select a preset or specify a custom time range.

You can also zoom in (by clicking and dragging the cursor in the pane) for a more granular view. Click Zoom Out to expand time window or click:

to shift time window forwards.
to shift time window backwards.

The page displays charts by default. When viewing activity by nodes, you can switch to a list view by clicking and return to charts view by clicking .

Charts view

You can select the charts to view by creating a custom chart Board or by modifying an existing Board. See Creating a Custom Board for the Charts or Editing a Chart Board for more information. Click the associated with the chart to view information about the graph.

List view

The list view shows the metrics in the columns. You can customize the columns by clicking . See Adding and Removing Columns from the List View for more information. In addition, you can:

Click the column name to sort the table by that column.
Click the node name to go to the metrics page for the node.
Click one or more checkboxes next to the node name and switch to the charts view to visualize metrics for the secondary index activities on those nodes only.

You can use the charts to diagnose and troubleshoot bottlenecks and performance issues. For the complete list of metrics that you can view for secondary indexes, see Viewing Table Metrics in the Control System.

Viewing Region Distribution

Log in to the Control System and go to one of the following pages:

Summary tab in the table information page to view the region distribution for a table.
Summary tab in the index details page to view the region distribution for a secondary index.

The Region distribution pane shows the distribution of the table or secondary index regions across the nodes in the cluster. The shade of the node reflects the sum of the physical size of data on the node with the darker shade indicating increased resource utilization on the node. You can move the cursor over a node to view the following:

Hostname of the node
Number of regions on the node
Total size of data (across regions) on the node

You can click a node to go to the node information page.