Welcome to Headwind MDM Q&A, where you can ask questions and receive answers from other members of the community.

Please do not post bug reports, missing feature requests, or demo inquiries. If you have such an inquiry, submit a contact form.

0 votes
Which CPU, RAM, disk space should I use? And how many clients per server?
by

4 Answers

0 votes
As a start-up package, the virtual machine with at least 4GB RAM and 2xCPU is required.

Disk space doesn't matter. 10 GB would be enough to setup the host OS: Ubuntu Linux 18.04. SSD drive is recommended.

This server can run up to 200 online clients with logging, and up to 1000 online clients without logging (you can still turn on logging for test configurations).

For larger amount of clients, you can use a dedicated server with 16Gb RAM and 6xCPU and 250Gb SSD drive. Such server may host up to 2K online clients with logging, and up to 10K online clients without logging.
by (36.8k points)
0 votes
Here are two real examples.

1. The dedicated server Intel Xeon E3-1240 3.4 GHz 4 cores, 16Gb RAM, 220 Gb SSD is carrying about 800 devices (production mode, no logging), and this uses less than 10% of the server resources.

Such server should be able to carry at least 8K mobile devices.

2. The virtual machine 2 x 3.6 GHz CPU, 8 Gb RAM and 30 Gb HDD is carrying about 200 devices in production mode within more than 2 years (large database history), and this uses about 40% of the server resources.

Such virtual machine should be able to carry about 500 mobile devices.
by (36.8k points)
edited by
0 votes

A virtual machine on AWS or DigitalOcean having 16 Gb RAM, 4xCPU, 40Gb SSD, 10Gbit Internet connection - hosts up to 10K online clients (another real example).

Note that for a large number of devices in the database (10K+) the search for devices may be slow. You can finely tune that by setting up the device.fast.search.chars attribute in the XML configuration. This attribute enables fast (indexed) search devices by a few last characters of the device number (by default, 5 characters).

by (36.8k points)
Can you comment on the usage split between Postgres & Server/Tomcat?

We are looking to deploy via Docker for Tomcat server + managed Postgres instance, however the examples above are for everything loaded on a single VM.

Is the split 50/50? would you be looking at a 2xCPU, 8Gb RAM for Postgres/Tomcat in the above scenarios, or is one more heavily used than the other?
Yes, 50/50 would be fine.
0 votes
by (36.8k points)
...