Statsd: Which metric to use to measure data throughput

Question

I'm fairly new to statsd, I want to measure the aggregated data volume/second that my process uses, but cannot find the correct statsd metric for it:

• increment would be good, as it gives rate, but cannot increment with a value
• gauge doesn't aggregate, just sets a value
• timers also don't feel right

So which one should I choose, and how should I use it?

Answer 1

Use a counter, you can increment it with a value. For example to increment foo by 3

echo "foo:3|c" | nc -u -w0 127.0.0.1 8125

From the data types docs:

You can increment a counter by more than one by passing a second parameter:

statsd.incr('some.other.event', 10)