InfluxDB 0.9.0 in approach and compatible Raspberry Pi
I had spoken to you of InfluxDB there is this some time for my future home automation project but also just for those who wish to install it on the Raspberry. Today a small announcement about the future version of InfluxDB which should soon be out.
What I like in this database is its optimization for among other very quick access to data based on time. It is thus very easy to write a CPU usage tracking, disk space, temperature, etc. The data read is quick with a language based on the SQL with functions of collection and handling to simply retrieve large volumes of data.
The new version will bring a lot of new features with in summary :
- Clustering. The part where Raspberry Pi is may be the least concerned. A new management of clusters with term a treatment capacity of 1 from 2 million records per second. If you are at this level for home automation, you make me fear ! 🙂
- The addition of tags. It will be possible to add metadata about the columns of type string. These tags will be indexed, facilitating research as for example : What are the sensors from the 1st floor ? You can in a simple query give the evolution of temperature in a room without having to handle tables and more. Just with a tag on the data received. Each tag becoming a series, the basis that can store millions of series, the work is greatly simplified.
- API. In addition to cleaning, the principle of continuous queries is reviewed. Reminder, continuous queries allow a consolidation of several series as their arrival data. A gain in terms of performance since it is driving him even that manages it and not the user with multiple queries. Example : you want to query several series of sensors, which give a temperature, outside the home and have their evolution over 24 hours. continuous queries will save the data in a dedicated series and in a query you will have all the data to date.
- Storage engine. There also a simplification. If you had seen my article on compiling InfluxDB on the Raspberry, you saw that it was a little hard because we had to install several libraries in function of 4 possible engines. This time, a single engine to make everything turn : BoltDB. I give you link analysis different engines and why this choice (number of open files, write speed, resilience in the event of judgment, backup, etc.)
- 100% GB. A discovery for me this language. More than C or C . To Go. And the thing pretty amazing is the speed of compilation of the package ! As a Go compiler exist on the Raspberry so you can have InfluxDB.
Reminder, the web server integrated with InfluxDB is always present. After compilation you launch influxd and you connect on port 8083 your server. You can then manage your base and play with.
You also of a package to create dashboards with Grafana
So if you want to test this database, You can watch on this page where is the instructions. It's really simple and apart from the installation of the Go compiler, everything happens relatively fast (a little less than 20 total minutes). So why not use it ? 🙂