Round Robin data storage for non-regular interval time series data

Question

I am looking for the best mechanism to store a bunch of environmental data and like the concept of a round-robin database (rrd) to cap the size of the database. This will be used for a Java Web-app so I am more keen on using Java based technology. The tool I have been looking at is JRobin and RRD4J

However, it appears RRD is designed to store data at fixed interval. Is this correct?

The data I am looking to store are temperature, humidity, event driven data such as door open/closed, dry contact open/closed.

Temperature and Humidity data can be polled, but the hardware also send updates to us which is based on when the hardware detects a change (so not regular-interval). I would also like to store this as well.

Event-driven data will not be polled and will only be stored when the sensor send us the data.

To summarize, my question will be 2 folds:

1. Can RRD tools be used to store non-regular interval data?
2. If RRD tools cannot be used to store non-regular interval data, any suggestion on how to store non-regular interval data in a way which the storage mechanism automatically caps the size of the data store?

Answer 1

You might find some relevant info in the reasons why the Graphite project wrote their own RRD clone (called whisper): http://graphite.wikidot.com/whisper It seems to be for the reasons you mention.

One thing I might suggest here is to use StatsD https://github.com/etsy/statsd which while isn't written in Java, does have a Java interface and is used successfully for thousands of different stats at Etsy. It plugs into Graphite (http://graphite.wikidot.com/) which then stores the data for you and generates graphs. Since the graphs are all accessible via URL it might be the case that you don't need a pure Java solution.