With EVE Online's single-shard universe, the server hardware and game code have to be continually updated to cope with the game's constantly expanding playerbase. Each time the server is upgraded, node stability and lag in large fleet battles improves significantly. Unfortunately, this trend of periodic improvements took something of a wrong turn when the Dominion expansion came along. Before Dominion, fleet battles took place with over a thousand ships on each side and no crippling lag or node deaths in sight. Since the expansion's release, however, fleets of only a few hundred have regularly experienced unplayable conditions.
Frustrated by CCP's lack of visible progress in tackling lag, members of the EVE community have been making their outrage known. A recent appeal to the gaming media by a few players resulted in the issue getting very high exposure, which could be bad for CCP's public image. CCP issued a response explaining that lag is a high priority, and that there is even an entire development team dedicated to tackling it. Earlier this week, CCP Zulu expanded on CCP's lag-busting efforts in the first of a new EVE devblog series dedicated entirely to lag.
CCP Zulu began the blog with an honest admission that lag has been getting worse since Dominion was released. He went on to give a brief introduction to four upcoming devblogs, each on a different aspect of CCP's lag-fighting strategy. CCP Tanis will be reporting on the results from player-assisted mass testing events on Singularity, the EVE test server. CCP Warlock will explain the challenges associated with MMO scaling, and how exactly CCP spreads server load during a large fleet battle.
CCP Atropos will follow this up with an article on automated testing, and how CCP's new "thin client" tool allows for testing with hundreds of simulated players. Finally, CCP Unifex will talk about CCP's organisational structure and the approach that CCP's developers take to making games. If you're interested in what's going on behind the scenes at CCP HQ, the four upcoming EVE devblogs on lag are sure to be an informative read.
Reader Comments (4)
Posted: Aug 16th 2010 9:06PM (Unverified) said
''Before Dominion, fleet battles took place with over a thousand ships on each side''
is a tad wrong. Largest fight I have ever seen had around 1200 people involved before the system crashed. If you meant regular fights involving 1000 people in a system pre-dominion then that would be correct.
is a tad wrong. Largest fight I have ever seen had around 1200 people involved before the system crashed. If you meant regular fights involving 1000 people in a system pre-dominion then that would be correct.
Posted: Aug 17th 2010 4:14AM (Unverified) said
I was in a pre-dominion fight with 1000 on our side. I guess there would have been similar numbers on the other side. It was magnificent. There was lag, but it was playable.
Posted: Aug 17th 2010 4:42AM anduz said
Pre-dominion you had large fights, whether you'd call it playable or not is probably a matter of personal opinion. I mean, having to take your guns and modules off auto-cycling and press them manually every 5 minutes because they'd lock up for the rest of the fight due to lag if you didn't isn't exactly what I'd consider awesome gameplay.
However you didn't have those mass disconnects and server lockups on jumping that you do now, which actually render the game unplayable because your ship will be floating around helplessly without a pilot while you stare at the login server load screen.
I don't see any reason to kiss CCP for this though, it's damage control and until they actually fix the game they can sod off with their devblogs.
However you didn't have those mass disconnects and server lockups on jumping that you do now, which actually render the game unplayable because your ship will be floating around helplessly without a pilot while you stare at the login server load screen.
I don't see any reason to kiss CCP for this though, it's damage control and until they actually fix the game they can sod off with their devblogs.
Posted: Aug 17th 2010 8:44AM johnnliu said
Honestly, the devblogs list looks like a bunch of fluff that doesn't address anything. Designed to make lots of noise without any actual plan.
> Topic: Singularity Mass Testing Report, CCP Tanis
Summary: your tests are helping us, really! we got some ideas, may be. but look at all our logs! it will take us 18months to get to the bottom of this trust us.
> Topic: The Long Lag and MMO Scaling, CCP Warlock
Summary: This is how we did it before and honest it worked last time... We even got this presentation prepared where we showed how awesome we are. Not sure what broke now but.
> Topic: "Thin Clients" and Automated Testing from CCP Atropos
Summary: We do automated testing just like any other modern software projects. But we aren't simulating mass player tests or player behaviour. Just the simple stuff. So that's why... oh did you see how much logs we had from the test server!
> Topic: Carbon, CCP's Core Technology , CCP Unifex
Summary: It is extremely important for us that the core group are assigned on the next biggest thing. But don't worry we have got all the interns (but we call them "special developers" ) doing maintenance work. That's why they don't know how to read the logs, or see what they are supposed to see.
I may be vastly simplifying things, but:
1. You simulate real situations and look at the performance logs
2. You nail the features that are causing the most performance degradation. "auto-cycling" causing problems? Why not turn off a few suspect features on the Singularity and ask the players to show up for a big test event. Get some real numbers to validate your thoughts and then narrow it down again.
3. You want to tell your players what's going on? POST the results after this big test: In our studies if we switch off auto-cycling we noticed we can now handle 800 ships without anyone disconnecting.
If it works before but not now... the "possible culprit" can't be endless.
> Topic: Singularity Mass Testing Report, CCP Tanis
Summary: your tests are helping us, really! we got some ideas, may be. but look at all our logs! it will take us 18months to get to the bottom of this trust us.
> Topic: The Long Lag and MMO Scaling, CCP Warlock
Summary: This is how we did it before and honest it worked last time... We even got this presentation prepared where we showed how awesome we are. Not sure what broke now but.
> Topic: "Thin Clients" and Automated Testing from CCP Atropos
Summary: We do automated testing just like any other modern software projects. But we aren't simulating mass player tests or player behaviour. Just the simple stuff. So that's why... oh did you see how much logs we had from the test server!
> Topic: Carbon, CCP's Core Technology , CCP Unifex
Summary: It is extremely important for us that the core group are assigned on the next biggest thing. But don't worry we have got all the interns (but we call them "special developers" ) doing maintenance work. That's why they don't know how to read the logs, or see what they are supposed to see.
I may be vastly simplifying things, but:
1. You simulate real situations and look at the performance logs
2. You nail the features that are causing the most performance degradation. "auto-cycling" causing problems? Why not turn off a few suspect features on the Singularity and ask the players to show up for a big test event. Get some real numbers to validate your thoughts and then narrow it down again.
3. You want to tell your players what's going on? POST the results after this big test: In our studies if we switch off auto-cycling we noticed we can now handle 800 ships without anyone disconnecting.
If it works before but not now... the "possible culprit" can't be endless.








