[
https://issues.apache.org/jira/browse/CASSANDRA-20250?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17924068#comment-17924068
]
Benedict Elliott Smith edited comment on CASSANDRA-20250 at 2/5/25 1:32 PM:
----------------------------------------------------------------------------
I mean the implementation. It's bad, in many ways. {{LongAdder}} is the wrong
choice for most of these metrics - the additional memory latency alone from
indirection likely more than offsets any contention benefit. But, most of these
metrics could be maintained in a single-threaded way anyway. And we have way
too many counters, as we count the same thing dozens of times. Most metrics
have a denominator of the same value, but even within each metric, we count the
same things repeatedly - in each codahale Meter by default we count every tick
_four times_ (one for the count, and three for each moving average), each one
with its own {{LongAdder}}. Ignoring the fact that most of these moving
averages are entirely unqueried, they don't need to be updated independently.
This mistake is then multiplied over every single metric, of which there are
many.
It's honestly a shambles.
was (Author: benedict):
I mean the implementation. It's bad, in many ways. {{LongAdder}} is the wrong
choice for most of these metrics - the additional memory latency alone from
indirection likely more than offsets any contention benefit. But, most of these
metrics could be maintained in a single-threaded way anyway. And we have way
too many counters, as we count the same thing dozens of times. Most metrics
have a denominator of the same value, but wven within each metric, we count the
same things repeatedly - in each codahale Meter by default we count every tick
_four times_ (one for the count, and three for each moving average), each one
with its own {{LongAdder}}. Ignoring the fact that most of these moving
averages are entirely unqueried, they don't need to be updated independently.
This mistake is then multiplied over every single metric, of which there are
many.
It's honestly a shambles.
> Provide the ability to disable specific metrics collection
> ----------------------------------------------------------
>
> Key: CASSANDRA-20250
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20250
> Project: Apache Cassandra
> Issue Type: New Feature
> Components: Observability/Metrics
> Reporter: Dmitry Konstantinov
> Priority: Normal
> Attachments: cpu_profile_insert.html
>
>
> Cassandra has a lot of metrics collected, many of them are collected per
> table, so their instance number is multiplied by number of tables. From one
> side it gives a better observability, from another side metrics are not for
> free, there is an overhead associated with them:
> 1) CPU overhead: in case of simple CPU bound load: I already see like 5.5% of
> total CPU spent for metrics in cpu framegraphs for read load and 11% for
> write load.
> Example: [^cpu_profile_insert.html] (search by "codahale" pattern)
> 2) memory overhead: we spend memory for entities used to aggregate metrics
> such as LongAdders and reservoirs + for MBeans (String concatenation within
> object names is a major cause of it, for each table+metric name combination a
> new String is created)
>
> The idea of this ticket is to allow an operator to configure a list of
> disabled metrics in cassandra.yaml, like:
> {code:java}
> disabled_metrics:
> - metric_a
> - metric_b
> {code}
> From implementation point of view I see two possible approaches (which can be
> combined):
> # Generic: when a metric is registering if it is listed in disabled_metrics
> we do not publish it via JMX and provide a noop implementation of metric
> object (such as histogram) for it.
> Logging analogy: log level check within log method
> # Specialized: for some metrics the process of value calculation is not for
> free and introduces an overhead as well, in such cases it would be useful to
> check within specific logic using an API (like: isMetricEnabled) do we need
> to do it. Example of such metric:
> ClientRequestSizeMetrics.recordRowAndColumnCountMetrics
> Logging analogy: an explicit 'if (isDebugEnabled())' condition used when a
> message parameter is expensive.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
