Doubt about horizontal scale

simplementese2 years ago

Horizontal Scale Inquiry

Well I have a doubt.
I can have several instances of traccar , but I must share the same database or I misunderstood the article.
Does it have to be the same database model and everything that reaches instance 1 is replicated to instance 2 and its database?

From already thank you very much

Anton Tananaev2 years ago

You need to have a single database (or a database cluster).

simplementese2 years ago

Excuse the ignorance.

And what would be the use of having 2 different instances, if they share the same db, in having it in only one.

Anton Tananaev2 years ago

There are two main reasons:

  1. Scale resources to support more devices
  2. Improve reliability (if one server goes down, second one handles the requests)

Why are you interested in horizontal scaling if you don't know what it's for?

simplementese2 years ago

There is more information about scaling except what it says " https://www.traccar.org/horizontal-scaling/ " ?

I am interested in reliability, but I thought it worked differently, I thought that an instance with its base replicated in another with its own base, since that way when using it in a failure with the db it would replicate.

Anton Tananaev2 years ago

There's some more information on the forum, but no other documentation.

As for the database, as I mentioned earlier, you can have a database cluster if you need to replicate the database as well.

control taken2 years ago

@simplementese It is very simple, as Anton explained, apparently you do not understand how modularizing services works, but generally the problems that arise in traccar services are not due to the database in most cases, in addition to modularizing It allows to scale the service with a minimum hardware to manage more devices and have backup in case of any event in any instance of traccar
https://www.cyberneticos.com/img/cyberproxy/cyberproxy-02.png
https://galvarado.com.mx/uploads/Untitled-1.png

simplementese2 years ago

I'm sorry, but it seems to me that your message is not constructive and I think that we don't behave like that here.

I've been using traccar for 4 years with around 900 devices and I've never had a problem with the traccar instance, except in this last update with the memory issue.
It was always a very robust instance.
Even to have a backup you can have another instance installed pointing to the same db and with a fast port forwarding you would already have a backup on the spot

simplementese2 years ago

and particularly, pointing to the same DB, the problem with more than 1000 devices... with several months of stored positions, reading when generating a historical report is where there are mostly problems