PTV Group - Logistics Forum

Hello,

We are implementing xLocateServer on a VM scaleset on Microsoft Azure & would be interested in some recommendations so that we can achieve the best performance

- What is a reasonably safe queue length setting and maximum number of query threads for a machine with the spec of our VMs
- Initial deployment is to machine type A2 : 2 CPUs, 3.5 GB RAM, 135 GB HDD 4x500 IOPS) set to scale from 1 to 10
- Targeted deployment is to machine type D2v2 : 2 CPUs, 7GB RAM, 100GB SSD, 4x500 IOPS ) set to scale from 1 to 10

Literature on these machine sizes is available here: [urlhttps://docs.microsoft.com/en-us/azure/virtual-machines/virtual-machines-windows-sizes#dv2-series][/url]

- Are there any other config settings which we would need to modify in order to increase those settings safely without any risk of the server process running out of memory?

Many thanks for your attention

Hello Rac,

I forwarded this question to DEV. I think it's challenging to answer "what is a proper config for a single instance" on a generic base. Maybe the colleagues can give you some info about how we setup our xServer INTERNET servers in Azure®.

The second question is then "how many instances do we need?".

We'll see,

best regards Bernd

Hi,

our measurement on hosting xServer on Azure found, we got the best performance if cores and modules are equal. So I recommend two xLocate modules.
Depending on the usage, one xLocate module can use up to 2GB memory. Together with the Server process and OS usage, I also suggest only two modules for your case.

The queue length is depending on your preferred pattern. A long queue is possible and costs only a bit of memory. A good compromise between response time and rejection is a queue between 50 and 100.

Regards

Marco

Thank you very much Bernd & Marco.

Your (as usual) quick & helpful responses are much appreciated

k