We have a hybrid cloud/on-prem infrastructure connected by dark fiber at 10Gbps. Our on-prem stuff is pretty legacy so many of the devices use fixed TCP window sizes. Using the bandwidth delay product, at 36ms ping (typical for our fiber connection), you only get around 15 mbps (calculator here) which is exactly what we are seeing. On prem this is not an issue because ping is <1ms. So on-prem.. we can get up to max speed no problem.. but when it hits the cloud.. our speed drops to 15 mbps for these legacy devices. Way less than the 10gbps we are paying for!
The good news is that for non-legacy devices on prem.. that support dynamic window scaling.. we can get up to 10gbps also! So the link works and all the stuff in between is good.
My question is - is there some kind of device.. router setting.. software.. etc. that can solve this problem for our legacy devices? I tried using an on-prem server with dynamic window scaling turned on, running an SSH SOCKS5 proxy.. and lo-and-behold this worked! But... that is not a sustainable solution :) Wondering if our on-prem cisco routers have some kind of tcp proxy option.. or perhaps the appliance that connects to our 10gbps fiber? or should I just run an on-prem machine and use a TCP proxy like haproxy?
thanks everyone in advance!
