Support Questions
Find answers, ask questions, and share your expertise

understanding HBase latency metrics

understanding HBase latency metrics

Explorer

Hi,

The HBase doc mentions latency metrics https://hbase.apache.org/book.html#hbase_metrics but without giving any unit ;-(

hbase.regionserver.<op>_<measure>

Operation latencies, where <op> is one of Append, Delete, Mutate, Get, Replay, Increment; and where <measure> is one of min, max, mean, median, 75th_percentile, 95th_percentile, 99th_percentile

QUESTION_1 : is "millis" the unit ? micros or nanos ?

QUESTION_2 : how is measured avg metrics for example ? is it calculated from t0 (startup time), or from a given period of time (and reset at the begining of every period of time) ?

Many thanks.

3 REPLIES 3
Highlighted

Re: understanding HBase latency metrics

1: Yes, millis.

2: From the start of the process.

Highlighted

Re: understanding HBase latency metrics

Explorer

Thanks Josh.

Due to (2), how to measure latency for, let say, the last 5 mn ?

If one has only a metric for measuring median latency from t0 (startup time), it's quite useless, for example, for detecting a surge in latency.

Any hint ?

Thanks.

Highlighted

Re: understanding HBase latency metrics

Actually, it looks like I might have missed something the first time. I think the metrics are rolled every 60 seconds (based off of https://github.com/apache/hbase/blob/branch-1.1/hbase-hadoop2-compat/src/main/java/org/apache/hadoop... and https://github.com/apache/hbase/blob/branch-1.1/hbase-hadoop2-compat/src/main/java/org/apache/hadoop....