There is clearly a difference between the "usual" lag and the one which, as it has been said several times, appeared at the implementation of VU 17.1 (on March 3rd if I remember correctly)
It's true that the usual lag should be dealt with as well, the servers clearly have a lot of trouble when there's an influx of many players, especially at the weekend.
Or when a lot of players are in the same place, for example during an event, then the server gets a lot of time behind on the calculations and everything is done with a constant delay.
But the peaks of lag followed by 3 Failed, are something different IMO, which is created by something specific that causes problems. (in the coding or in the infrastructure)
This weekend was particularly terrible since we had both kinds of lag at the same time..
I think it would be a good idea to roll back before the VU 17.1.
First of all it would confirm or deny that these new lag spikes come from there.
It would only take 24 hours to figure out if it was created by that VU.
If it is the case, it would then be necessary to reimplement the VU 17.1 part by part to see which part is the cause of the problem.
If it is not the case... well good luck to find out where the problem comes from.
Apart from that there is one thing I noticed about this "new" lag.
It seems to me that most of times, it's the exact moment we log-in that defines whether or not we'll have these lags during the session.
Most of the time, either I have these lags for the whole session, or I don't have them at all.
For example, it almost never happens that the lag spikes appear after 1 hour ingame.
Either they don't appear at all, or they appear at the moment of connection and stay until disconnection.
I don't know what that means, but to me it sounds like it's technically useful information.
Sometimes I get these famous lag spikes every 30 seconds, I relog and they're gone.
Sometimes I relog 3/4 times in a row and I can't "catch a session" without lag.
And this strangely reminds me of something, I don't remember when at all, but I'm sure it's already happened in the past.