I've been working on some new context implementations recently, hosted on a redundant pair of ACE30s running Version A4(2.1) and have been investigating a number of minor issues.
One thing I've noticed upon looking in to the ACE behaviour (and some reading of legacy posts on this site) is that the device clears the Windows Scale and SACK options for incoming connections to the VIP. Obviously, these behaviours are quite easy to alter with a Parameter Map, but I'm just trying to understand the downsides of doing so, and why the ACE30 defaults to that behaviour.
Does anyone have any specific advice for default Parameter Map changes they tend to put in place within their configurations? The majority of clients we have connecting are Windows based, with the majority of links having 100+ ms of latency.