Concepts and Terminology
On this page
Before querying these workload profiling views, it’s important to understand the concepts behind the terminology used in each view’s schema.
SingleStoreDB Cloud is instrumented to divide its execution time into tasks.
A single query’s execution is instrumented with a task per partition, as well as a single task on the relevant aggregator.
A single garbage collection pass is instrumented with a single task.
A single replication pass against a database partition is instrumented with a single task.
Tasks may or may not be
For example, a query task which commits a transaction is successful, but a query task which encounters rollback is not.
An activity is a set of identical tasks.
All tasks associated with the same node ID, database, partition, and name comprise the same activity.
In practice, tasks share an activity when they instrument multiple repetitions of the same recurring work.
Every activity has a type of either Query, Database, or System.
Every activity has a name and an aggregator activity name.
If an activity is not high-level, then its aggregator activity name is the name of the high-level activity to which it contributes.
In practice, we report aggregator activity name in order to link per-partition query activities with the unique aggregator activity associated with their query.
All tasks instrumenting the same query against the same partition share an activity, which has Query type.
Tasks instrumenting the query on different partitions have different activities.
All garbage collection pass tasks share an activity, which has System type.
All replication tasks against the same database partition share an activity, which has Database type.
All activities associated with a query have the same aggregator activity name, which is the name of the unique high-level activity associated with the query on the aggregator node.
Last modified: September 30, 2022