Opinions on hardware for a workstation running InfluxDB, Telegraf, Grafana

InfluxDB 2
, , ,
1

Setting up a new Influxdb at work. This will be my first installation of the software and first venture into databases. The machine will be running influxdb, grafana and node red. All of these will be deployed using Docker. Also will be used for some excel work, mainly for organizing data into reports. Everything will be hosted on one machine and only that machine will be using the data. Below is probably as demanding as the database will get.

For writes

40 buckets

200 Field values/bucket

Updating once every 10 seconds.

Data stored for 2 years

Cardinality should be pretty slim as this will all be done through the modbus plugin.

For Querying I am probably looking at around 20-25/s max.

Looking at the hardware sizing It looks like I am thinking a 6-8 core xeon (W-1270p or similiar). I am thinking tis would be a little overkill. Probably will start with 16gb of ram as upgrading to 32 later on is easy enough. For storage, I am still trying to determine the exact space we need but was thinking something along of the lines 3x1tb in raid 5 would be a good starting point.

Is this a good starting point? This will be my first venture into databases and would prefer not to cause my self headaches by accidently under sizing the workstation.

2

I hope you will find helpful the following article:

it applies also to InflxDB2 as it has the same storage engine.

3

I have taken a look at that. I guess the part that I am getting hung up on is the processor portion. Core counts are easy enough but that’s only one part. Frequency and caches also play a a large part in processor performance. So a better question would be when working with influxdb2 how important is the frequency of the cpu and cache. I assume its minimal and I am reading way to much into things.

4

To be honest I really appreciate this kind of deep dive into details… But I can’t really help as I’m no hardware expert and I don’t know that much about the InfluxDB engine.
@Anaisdg are there some additional resources or someone who can answer to this kind of detail?

5

I think it only lists core because concurrency is a bigger factor than clock speed. Any modern CPU should have enough speed and cache to handle almost any workload you throw at InfluxDB, so it becomes purely a question of how many concurrent workloads do you need to run at any given time.