The 'Health Dashboard' is designed to give a quick high-level overview of database cluster health and activity. There are 24-hour sparkline graphs of each of the five six main cluster attributes: Transactions per second, CPU utilization, Rebalancer status, In-memory table utilization, Storage, and Read and Write latencies. These graphs are shown in green, yellow, or red depending on how they compare to threshold levels. Visual Alerts are displayed for attributes which are 'too high'. outside normal ranges. All data is polled and updated every 5 seconds. Note: the 'big graph' showing 24 hours of data is displayed in 5-minute intervals. The 'Health' Icon will pulse red if any of the 5 main cluster attributes has exceeded the Alert thresholds.
Health Dashboard Tabs
The TPS section displays the number of database transactions - per - second across all nodes, including sections for:
- Current transaction execution times
- Percentage shows the percentage of transactions occurring within a threshold of Execution timesCount shows how many transactions occurred within a threshold of Execution times, broken down by execution time
- Current transactions per second by node
- Transactions per second in the last 24 hours (5 minute intervals)
You can select whether to display transactions or queries per second.
The CPU section: displays Average CPU utilization across all the database nodes in the cluster including:
- CPU usage breakdown
- Current CPU usage by node, which can be used to identify imbalances
- CPU usage over the last 24 hours
The memory section highlights provides information on how memory is being utilized used, including for In-Memory tables.
- The current memory allocation allocations for In-Memory Tables and other Memory Usage are listed at the top of the screen with
In the middle is the current percent utilization per node of allocated In-memory table memory, Buffer Manager, Working Memory, Reserve, OS + Other Apps, and Free
- % of Total indicating how much of the total memory is used for each item
- Current memory utilization per node
- The graph at the bottom shows the percent utilization of the cluster's allocated memory for In-memory tables over a memory by type over 24 hour period, in 5 minute intervals
The Storage section displays how storage is being consumed across the active nodes in the cluster, including the following:
- User: Amount of User space
- Undo: Amount of Undo log
- ReplicationBinlog: Amount of storage used by replication binlog(s)
- System: Amount of storage used by system processes
- Free: Amount of free space in the cluster
This section also includes a breakdown of "Storage by node" and displays the "Storage allocation in the last 24 hours." These are useful for identifying uneven or "lumpy" distributions in data or recent changes to storage.
- "Latency over the last 24 hours": Displays the query read and write latencies (two-color chart) over the last 24 hours in 5 minute intervals
- "Read Latency" The current average read latency of the cluster
- "Write Latency" The current average write latency of the cluster
- Exec Count: The number of times the query was executed
- Avg Exec Time: The average amount of time each iteration of the query took to execute
- Query Statements: Displays the query statement, up to two lines. Click 'Show More' to display the full Query.
- Avg Rows Read: The average number of rows read by each iteration of the associated query
- Avg Rows Written: The average number of rows written by each iteration of the associated query
- Avg Rows Output: The average number of rows output by each iteration of the associated query
- Database: The database on which the query executed
Recent Queries can be filtered via the Query Filters in the left pane or by entering text into the 'Filter by text' field.
The 'Graphs' tab displays important database attributes over a user-specified interval. There are predefined collections of graphs for Diagnostic (ClustrixDB Specifics), Health, Locks, Performance, Queries and Rows, Replication, Sessions and Transactions, and Storage. You can also define custom collections.
The default interval is 24 hours, but can be range from 15 minutes to 7 days. Choose an interval width on the data selector, or the top ruler (using the left/right cursors) , then drag by dragging that selected interval across the date range. Refresh Refresh the screen to go back to the default time frame.There are multiple Categories of graphs, including: Performance, Diagnostic, Queries and Rows, Locks, Sessions and Transactions, Replication, and Disk. Click on the desired category to expand graphs. The graph below views TPS, Read/Write Latency, CPU Utilitzation, and Network Latency.
- Moving the cursor back and forth over a graph creates a vertical 'needle' which displays the attributes (on the left) for each visible graph at that time-stamp. All the graphs are aligned by timestamp, to aid in point-in-time analysis.
- Single-click to place a vertical 'needle' for comparison purposes with the (moving) needle under the mouse
- There are 2 ways to zoom in on Graphs
- Zoom into the x-axis (time) of all of the visible graphs by clicking inside a graph and dragging horizontally across the graph, you will see a vertical column highlighting the area (time frame) you wish to zoom in on.
- Zoom into the y-axis of a particular graph by clicking in a graph and dragging vertically on a graph. You will see a horizontal bar across the screen highlighting the area of the y-axis that will be zoomed in on.
- Double-click the graph to reset the zoom level to normal, you may need to double click several times if you have done many zooms. You may also go to a different tab and return to the Graphs tab.