简体   繁体   English

百分位数指标是否遵循Graphite + Grafana中的求和规则?

[英]Does percentile metrics follow the rules of summations in Graphite + Grafana?

We are collecting certain metrics using (Graphite + Grafana) use them as a tool to monitor system health and performance. 我们正在使用(Graphite + Grafana)收集某些指标,并将它们用作监视系统运行状况和性能的工具。

For one of the latency metric, we get the total time as well as the latencies for all the sub-components it is composed of. 对于延迟度量之一,我们可以获得总时间以及它所组成的所有子组件的延迟。

We display 99th percentile for all the values. 我们为所有值显示第99个百分点。 However, if we sum up the 99th percentiles for latencies of sub-components, they do not equate to the 99th percentile of the total time. 但是,如果我们将子组件的延迟汇总到第99个百分位数,则它们不等于总时间的第99个百分位数。

Essentially it comes down if the percentiles can follow summation rules. 本质上,如果百分位数可以遵循求和规则,则下降。 ie

if 
a + b + c + d = s

then,
p99(a) + p99(b) + p99(c) + p99(d) = p99(s) ?

Will this hold? 这会举行吗?

IMHO this would be true only if |a| 恕我直言,仅当| a | = |b| = | b | = |c| = | c | = |d| = | d | If this is not the cause, you should weight your equation by the number of time you pass by each component. 如果不是这个原因,则应通过每个组件经过的时间来加权方程式。

Imagine you have only component 'a' and 'b'. 假设您只有分量“ a”和“ b”。 If for 100 requests passing by component 'a', 'b' is called 900 times then 0.1*p99(a) + 0.9*p99(b) = p99(a+b) 如果对于组件'a'传递的100个请求,'b'被调用900次,则0.1 * p99(a)+ 0.9 * p99(b)= p99(a + b)

PS: you should remove your 'java' tags, and maybe 'graphite' and 'grafana' tags too. PS:您应该删除“ java”标签,也可能删除“ graphite”和“ grafana”标签。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM