Generating a visualization of a metric at one or multiple levels of execution of a database workload
Described herein are techniques for generating a visualization relating to execution of a workload. Multiple measurements of a plurality of metrics relating to execution of the workload may be aggregated at multiple levels of execution. A visualization may be generated that comprises a representation of the measurements of a metric at one of the levels of execution.
1. A method performed by a processing system comprising:
receiving multiple measurements of a plurality of metrics relating to physical operators, in a physical execution tree, executing a database query;
determining logical operators in a query execution plan for executing the database query, wherein the query execution plan is determined from the physical execution tree, and each logical operator is associated with at least one of the physical operators;
aggregating the multiple measurements of the plurality of metrics at multiple levels associated with the logical operators;
generating a visualization representing execution of the database query, the visualization comprising a representation of the measurements of at least one of the metrics at one of the levels; and
updating the visualization periodically to include current measurements, wherein the updating of the visualization includes:
suppressing unhelpful information by one of hiding a non-executing portion of the database query, hiding an insignificant metric value of the database query, or using a logarithmic scale for a displayed chart;
receiving a first instruction to drill down on the visualization from a node level to a path level, wherein the first instruction includes a transition from displaying a first metric of a first logical operator at the node level to displaying the first metric of the first logical operator at the path level; and
pivoting, upon receiving a second instruction, to display the visualization along a different dimension, wherein the pivoting includes displaying the first metric for all of the logical operators at the path level.
2. The method of claim 1 , further comprising:
providing a user interface to allow selection of one or more of the plurality of metrics at one of the multiple levels;
receiving a selection of one or more of the plurality of metrics at one of the multiple levels; and
generating a new visualization representing execution of the database query, the new visualization comprising a representation of the measurements of the selected one or more metrics at the one of the multiple levels.
3. The method of claim 2 , wherein the user interface allows selection of a time period, the generated new visualization comprising a representation of the measurements of the selected one or more metrics at the one of the multiple levels over the selected time period.
4. The method of claim 1 ,
wherein the multiple levels comprise at least two levels from a query level, a query phase level, the node level, the path level, and an operator level.
5. The method of claim 4 , wherein:
the visualization generated at the query level comprises a representation of execution of the entire database query,
the visualization generated at the query phase level comprises a representation of execution of the database query at a current phase of the database query,
the visualization generated at the node level comprises a representation of execution of the database query on one or more nodes of a query engine executing the database query,
the visualization generated at the path level comprises a representation of execution of logical operations of the database query, and
the visualization generated at the operator level comprises a representation of execution of physical operations of the database query.
6. The method of claim 5 , further comprising modifying the visualization to show measurements of at least one of the plurality of metrics at a different level than the one of the levels in response to a user input.
7. The method of claim 1 , wherein the plurality of metrics comprise at least two metrics from elapsed time, execution time, memory allocated, memory reserved, rows processed, processor utilization, and network activity.
8. The method of claim 1 , wherein the visualization further comprises a representation of estimates of at least one of the plurality of metrics at the one of the levels.
9. A system comprising:
a database to store measurements of metrics relating to physical operators, in a physical execution tree, executing a database query, wherein the metrics are collected by a query execution engine during execution of the database query;
an aggregator, executed by a processor, to aggregate the measurements of the metrics at multiple levels of execution of the database query, wherein the multiple levels are associated with logical operators, and each logical operator corresponds to at least one of the physical operators;
a web server to provide a user interface to allow selection of one or more of the metrics and one of the levels; and
a visualization generator, executed by a processor, to generate a visualization representing execution of the database query, the visualization comprising a representation of the measurements of a selected metric at a selected level of execution,
wherein the user interface is further to allow updating of the visualization periodically to include current measurements, and further wherein to allow the updating of the visualization the user interface is to:
receive a first instruction to drill down on the visualization from a node level to a path level, wherein the first instruction includes a transition from displaying a first metric of a first logical operator at the node level to displaying the first metric of the first logical operator at the path level; and
pivot, upon receiving a second instruction, to display the visualization along a different dimension, wherein the pivoting includes displaying the first metric for all of the logical operators at the path level.
10. The system of claim 9 , wherein the multiple levels comprise at least two levels from a query level, a query phase level, the node level, the path level, and an operator level.
11. A non-transitory computer-readable storage medium storing instructions for execution by a computer, the instructions when executed causing the computer to:
receive multiple measurements of a plurality of metrics relating to physical operators, in a physical execution tree, executing a database workload;
determine logical operators in an execution plan for executing the database workload, wherein the execution plan is determined from the physical execution tree, and each logical operator is associated with at least one of the physical operators;
aggregate the multiple measurements of the plurality of metrics at multiple levels associated with the logical operators;
generate a visualization representing execution of the workload, the visualization comprising a representation of the multiple measurements of at least one of the metrics at one of the levels; and
update the visualization periodically to include current measurements, wherein to update the visualization, the instructions when executed cause the computer further to:
suppress unhelpful information by one of hiding a non-executing portion of a database query, hiding an insignificant metric value of the database query, or using a logarithmic scale for a displayed chart;
receive a first instruction to drill down on the visualization from a node level to a path level, wherein the first instruction includes a transition from displaying a first metric of a first logical operator at the node level to displaying the first metric of the first logical operator at the path level; and
pivot, upon receiving a second instruction, to display the visualization along a different dimension, wherein the pivoting includes displaying the first metric for all of the logical operators at the path level.
12. The medium of claim 11 , wherein the workload comprises a query or a workflow.
13. The medium of claim 12 , wherein if the workload comprises a query, the multiple levels comprise at least two levels from a query level, a query phase level, the node level, the path level, and an operator level.
14. The medium of claim 12 , wherein if the workload comprises a workflow, the multiple levels comprise at least two levels from a workflow level, the node level, a logical operator level, and a physical operator level.