Graphite

All-in-one solution for storing and visualizing time-series data in an efficient manner

Whisper

Fixed-size time-series numeric database

carbon

Metric receiver, aggregation and disk flushing

Graphite web

Django app for rendering graphs and pulling arbitrary data

basic config

[min]
pattern = \.lower$
aggregationMethod = min

storage-schemas.conf

storage-aggregation.conf

[my_schema]
pattern = .*
retentions = 10s:6h,1min:6d,10min:1800d

basic usage - easy!

http://graphite.host/render?target=metric.name

Send metric

<any.metric.name> <number> <timestamp>
stats.coffee.left.cups 5 1434305622

get graph

Dozens of yummy transformation functions

• Averages, max/min, percentiles...
• Aggregate by time window (ex. hits per minute)
• Derivative, integral, power, scaling, differences
• Time warping
• Check the docs!

Also JSON and CSV

Feed your own app!

statsd

Network daemon that listens for statistics and send aggregates to pluggable backend services

Uses UDP: fast, fire-and-forget

Client libraries in all popular languages

Main metric types: counter, timer, gauge

<metricname>:<value>|<type>
echo "foo:1|c" | nc -u -w0 127.0.0.1 8125

pystatsd

Python client for StatsD

>>> import statsd
>>> c = statsd.StatsClient('localhost', 8125)
>>> c.incr('foo')  # Increment the 'foo' counter.
>>> c.timing('stats.timed', 320)  # Record a 320ms 'stats.timed'.

pip install statsd

django-statsd

Convenience wrapper to use pystatsd in Django

>>> from django_statsd.clients import statsd
>>> statsd.incr('python.meetup')

pip install django-statsd-mozilla

awesome goodies!

• StatsD config in settings.py
• Requests, database, cache and frontend timings
• Models create, update, delete signals count
• Celery signals

final notes

how do i install all this???

graphite graphics are fugly!

graphite components are too slow!