Use multiples InfluxDB backends is recomended?

[pablo.chamorro]

Hello everyone,

We are testing diferents ways to implementate our influxDB infraestructure on WinCCOA 3.17P12.

In our case, we have a complete production circle, and it means, really diferent data collections, not at all interconnected.

So, our question here is, is better store all the data in the same data base (backend), with diferents measurements (archive groups), or directly, create a diferent database (backends) for each production environment?

In terms of robustness, we prefer to separate each production environments data, in diferents databases (backends). But, in terms of performance, we are not sure if this is the correct way.

Testing to use a database for each production environment, we saw that when you active 25+ backends, the NGA manager crashes. (InfluxDB in localhost with each backend in 'in-proc' execution type)

Does it means that, use multiple backends is a bad practices? Is posible that, for each plugin process that winccoa creates, the performance is afected?

InfluxDB is completly new for us, and we are not sure how we have to use it in winccoa, to have the best performance and data robustness.

Thanks in advance!

[kilianvp]

You run multiple production environments (WinCC OA Projects) on the same host?

[pablo.chamorro]

We have only one WinCC OA project running.

When I talk about differents production environments, I'm talking about our company have a multiple plants, and each plant does different process and services, not at all connected to each other.

That's why we want to separate the data of each production environment (plant) in archive groups or directly in a differents backends.

But, as we post, we have doubts about how WinCC OA can be affected in terms of performance, having multiples InfluxDB backends.

[adaneau]

Hi,

Why would you run several DB to increase robustness? The logic is not good, you are just having several points of failure. If you want robustness, just go for a standalone influxdb and put in on HA cluster. Then you can put all your data there.

BR
Alex

[leoknipp]

I am not sure if using several Influx database on the same host for the same projects is technically possible.
If the NGA manager is started and the Influx database is not running the NGA manager will start it referrring to the influxdb.conf located in the config directory of the project.
As there can be only one influxdb.conf possibly the "wrong" database is started.

It probably will work if every Influx database is started beforehand but as far as I know we have never tested such a solution.

Best Regards
Leopold Knipp
Senior Support Specialist