Hi. I am trying to minimize the overhead of using a proxy. I have created proxies in the same data center as my application server (Vultr Tokyo) and am comparing unproxied requests with proxied requests. There is a 4ms-5ms overhead (e.g., 20ms with no proxy, 25ms with proxy). The ping response between application server and proxy server is about 0.5ms. So let's say the net overhead of the proxy server is 4ms.
For my application, even 1ms matters, so I'm wondering if there are areas I should research for how to reduce latency. Thanks!
Awesome! This question has already 3 answers..
Hi,
This is probably dependent on the distance between the servers in the data centre, and the number of connectors between them.
So, you can either try a few server combinations and pick the combination with the least latency, or contact Vultr and request servers that are located in the same rack.
Thanks, Viren
Thanks. I *think* the issue is actually related to the use of sessions. My application pulls from the same exact url and I'm using a requests.session connection pool to keep the cxn warm which speeds things up. But I'm rate-limited and therefore want to use some proxies.
Is there a way to configure the proxy servers to maintain a similar session? Otherwise, I have the normal http cxn overhead on every request, even though I'm always accessing the same endpoint.