Hi there,
I have been hosting traccar on AWS for the last two years. As the number of devices increased we noticed a perfomance degradation of our service. When we started we had RDS on gp2 and the number of iops was 1000. I then noticed RDS logs indicating issues with iops and recommending i switch to provisioned iops where I could have flexible iops independent of capacity. I have switched to that shown in the picture above and I wanted to find out if the resources could hand 500 devices without an issue.
In particular, I want to find out if 3000 iops (input/output operations per second) and 1 Gig ram, 50Gb and storage throughput can handle the workload. If possible, please share how I can calculate/estimate a single device resource consumption on the server so I can have an idea of how to optimise infrastructure as we scale.
N.B: I have implemented cleanup of historical logs and database history to ensure I optimise resource usage.
Thank you.
Why dont you just test it?
A nvme samsung 990pro has 1.5 milion iops
Our general recommendation so far has been to host database on the same server.
Hi there,
I have been hosting traccar on AWS for the last two years. As the number of devices increased we noticed a perfomance degradation of our service. When we started we had RDS on gp2 and the number of iops was 1000. I then noticed RDS logs indicating issues with iops and recommending i switch to provisioned iops where I could have flexible iops independent of capacity. I have switched to that shown in the picture above and I wanted to find out if the resources could hand 500 devices without an issue.
In particular, I want to find out if 3000 iops (input/output operations per second) and 1 Gig ram, 50Gb and storage throughput can handle the workload. If possible, please share how I can calculate/estimate a single device resource consumption on the server so I can have an idea of how to optimise infrastructure as we scale.
N.B: I have implemented cleanup of historical logs and database history to ensure I optimise resource usage.
Thank you.