trafficserver-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From James Peach <>
Subject Re: traffic server usage / stats (clarification)
Date Tue, 30 Sep 2014 20:51:34 GMT

On Sep 30, 2014, at 12:21 PM, Kevin Kruzich <> wrote:

> I'm resending with clarifications: Trying to calculate bandwidth usage from ATS (GET
requests to internet clients). The counters and
appear to be the correct choice

I believe that these metrics count all bytes sent and received over the network. You might
want to measure 


> but 1) what is difference between, net.write

"read" is bytes read (received) from the network, "write" is bytes written (sent) to the network

> and 2) Assuming this count is since start of proc, how might I slice up the data to increments
such as day/week/month?

Typically, people sample the metrics into a monitoring system, like Graphite or Circonus.
Those kinds of systems will aggregate and rate convert metrics to give you throughput and
utilization numbers.

> Thank you,
> --kkruzich
> ---------- Forwarded message ----------
> From: Kevin Kruzich <>
> Date: Mon, Sep 29, 2014 at 7:17 PM
> Subject: traffic server usage / stats
> To:
> Hello --I'm trying to determine how much data has been served via a given ATS server,
specifically GET requests to internet clients. There are many stats available on the /stats
page (same or similar results are offered via 'traffic_line -r'), yet of all of these, the
one which seems the closest I'm looking for is and .net.write_bytes.
But two questions about this:
> 1) What is meant by vs. net.write?
> 2) This is a count since last start of the process. How might I calculate bytes served
in a day/week/month?
> http://<ats-server>:8080/stat/

View raw message