We've had a UDP version of the InfluxDB Line Protocol (ILP) reader in QuestDB for quite some time, but we've had customers ask for a TCP version of it, so we delivered!
Using it, and configuring it, are relatively simple so don't expect this to be a long post but I'll walk you through the basics of how to set it up and use it. For an added bonus I'll show you how to migrate from using InfluxDB to using QuestDB with a less than a line of configuration.
Configuring TCP InfluxDB Line Protocol listener
Here's the best part, at least for a basic implementation that you don't need to performance tune at all: It's already set up.
That's right, as soon as you start QuestDB both the UDP and TCP ILP listeners start automatically on port 9009. Yes, TCP and UDP both use the same port. No, that's not a problem since one is UDP and one is TCP.
There are a bunch of configuration options you can tune in your
conf/server.conf file if you're interested. I won't go through them here, but
you can read all about them in our docs. I
hope they are relatively self-explanatory.
InfluxDB Line Protocol (ILP) refresher
If you have used ILP before, this should all be review. If you're new to ILP, this will tell you how you should write your data to QuestDB.
Pretty basic. So let's dive into what each element actually is, and how to structure a line of ILP for writing.
The first element is the
table_name portion, which tells the ILP writer which
database table to write values into.
Next comes the set of tags you want to use. These are standard
pairs, and you can add as many of them as you want or need. Just separate them
There should only ever be 2 spaces in your line protocol. No more. The first
space separates your
tags from the
values you want to associate with those
tagss. The second space separates the values from the timestamp for those tags
The values are also
key=value pairs, and again you can send as many as you
want in a line.
Finally comes your
timestamp value, typically in µSeconds.
Let's use an example of writing some environmental data to QuestDB. I have a sensor that reads temperature, atmospheric pressure, humidity, and the altitude.
And I want to use the following
|Tag Name||Tag Value|
Now I have all the basic elements I need to construct my ILP, which will look like this:
And yes, I rounded those values. But you'll notice that I did not add a
timestamp value. In this case, it's because I am sending the values from a
small, embedded sensor device that really doesn't have a great sense of time. By
sending the ILP without a
timestamp I'm telling the database itself to add one
for me, using the arrival time as the
One of the cool features of using the ILP reader (well, QuestDB in general really) is the ability to do 'Schema on Write'.
What that means is that if an ILP message arrives, QuestDB will automatically
create tables and columns to fit the incoming ILP. So if you need to add a
later, you can add it to the new device's tagset and start writing. The new tag
will get added to the schema.
If you leave a tag value off, and it exists in the database, it will get filled
When I start writing the above ILP to QuestDB, I'll get a table that looks like this:
This is what that table looks like in the QuestDB Web Console:
But how did you write that?
Oh, so how did I write that ILP to QuestDB? Well, my sensor is an Arduino, with
a Bosch BME280 sensor attached. It is WiFi connected, so a
WiFiClient can do
the TCP write for me:
Will connect the client to the QuestDB Server defined by
Quest_Server on port
If I then have a line of ILP like this:
buffer I can call
espClient.printf(buffer); and that line of data will
be written to QuestDB.
Can I do batch writes?
Of course you can! Just put each line of ILP on a separate 'line', separated by
\n and then when you have all your batch built up, write the whole
thing to QuestDB.
Of course, if you're relying on QuestDB to add
timestamps for you, then just
be aware that the entire batch will be given sequential timestamps based on when
they are read/written to the database.
InfluxDB Line Protocol (ILP) is a simple, well-known, and relatively compact data format for sending Time Series data to a database. That's why we decided to support it.
As I told you in the beginning, I'm now going to give you a simple, less-than-one-line configuration change to migrate from using InfluxDB to using QuestDB. If you're using Telegraf as a data collector, that is.
/etc/telegraf.conf file (it may be in different places, depending on
your operating system) and change the line:
That's it. That's the migration. Now all data that was previously being written to InfluxDB will now be written to QuestDB.