Latency monitoring · Valkey

Latency monitoring

Description

Valkey is often used for demanding use cases, where it serves a large number of queries per second per instance, but also has strict latency requirements for the average response time and the worst-case latency.

While Valkey is an in-memory system, it deals with the operating system in different ways, for example, in the context of persisting to disk. Moreover, Valkey implements a rich set of commands. Certain commands are fast and run in constant or logarithmic time. Other commands have O(N) complexity and can cause latency spikes.

Finally, Valkey is mostly single-threaded. This is usually an advantage from the point of view of the amount of work it can perform per core, and in the latency figures it is able to provide. However, it poses a challenge for latency, since the single thread must be able to perform certain tasks incrementally, for example key expiration, in a way that does not impact the other clients that are served. Some other operations, such as reading, parsing, polling, and writing to the socket, can be offloaded to a configurable pool of I/O threads, but command execution itself still happens on the main thread.

For all these reasons, there is a feature called Latency Monitoring, that helps the user to check and troubleshoot possible latency problems. Latency monitoring is composed of the following conceptual parts:

The rest of this document covers the latency monitoring subsystem details. For more information about the general topic of Valkey and latency, see Valkey latency problems troubleshooting.

Events and time series

Different monitored code paths have different names and are called events. For example, command is an event that measures latency spikes of possibly slow command executions, while fast-command is the event name for the monitoring of the O(1) and O(log N) commands. Other events are less generic and monitor specific operations performed by Valkey. For example, the fork event only monitors the time taken by Valkey to execute the fork(2) system call.

A latency spike is an event that takes more time to run than the configured latency threshold. There is a separate time series associated with every monitored event. This is how the time series work:

The framework monitors and logs latency spikes in the execution time of these events:

How to enable latency monitoring

The acceptable latency depends on the application. Some applications require all queries to complete within 1 millisecond. Other applications may tolerate occasional latency of 2 seconds for a small number of clients.

Set a latency threshold in milliseconds to enable latency monitoring. The default threshold is 0, which disables latency monitoring.

Valkey logs events that exceed the configured threshold as latency spikes. Set the threshold according to the application’s latency requirements.

For example, if the application requires a maximum latency of 100 milliseconds, set the threshold to 100:

CONFIG SET latency-monitor-threshold 100

Latency monitoring is disabled by default. Latency monitoring requires very little memory. However, enabling it increases the baseline memory usage of a Valkey instance.

Report information with the LATENCY command

The user interface to the latency monitoring subsystem is the LATENCY command. Like many other Valkey commands, LATENCY accepts subcommands that modify its behavior. These subcommands are:

Refer to each subcommand’s documentation page for further information.