Yes, it does sound very similar. This is what we had to do to mitigate it (unfortunately nothing fixed it).
- Stop using IE8 with EES (does not apply to you).
- Upgrade OS to Win7 (probably does not apply to you).
- Stop trying to do too many things at once, in particular advise authors and publishers to stop opening up 10 tabs with EES and clicking Save, Save, Save, etc. Literally, have one instance open of EES at a time. Any more than that I find it struggles.
- Check if there's any latency between you and the Matrix instance (we got the proxy.pac updated to go direct, but that was in internal network installation so may not apply to you).
- Don't have long pages with lots of content or divs (affected content tab when saving)
- Don't have too many metadata fields (affecting metadata tab, obviously)
- Avoid over-intensive triggers (which would run when clicking save, in particular when submitting or making live)
Squiz suggested we could have increased memcache timeout to 30 secs (default is 15 secs), but that was rejected by business owners as it has potential to affect live page performance (and they didn't really understand what was happening).
You could add an EES plugin that would take measurements of timings (I've been working on reporting these in GA), then if you found ajax responses working up to 15secs after a human action and then failing after that, then I think you can safely say you've tracked down the culprit. I also used a software proxy/packet sniffer to confirm fails starting after ~15secs. Narrowed it down for us, but in the end all we could was change our behaviours, nothing fixed it.
All this was done in a relatively old version of Matrix, v4.12, so I would have thought it be solved or fixed from the Matrix end by now. The fact it's failing in Matrix v5.1 with modern browsers is a bit of a worry. Hopefully it's a different issue with an easier fix.
Cheers
