As part of our ongoing performance optimizations, we have been spending a lot of time analyzing and addressing problems in Cyrodiil. I'm going to give a summary of what has happened over the years, what we have been doing to alleviate the problems, and our future plans.
When talking about ESO server performance, it is important to understand that all abilities in the game (with some exceptions) have a "soft" limit on the number of times they can be executed. An example of a "hard" limit on an ability is an ability that can be executed one time every two seconds, enforced by the server. ESO doesn't have many of those. Instead, we designed the game to have soft limits, which mean you can execute as many abilities as you want as long as you have the resources - Stamina or Magicka - to cast them. There is a global "cool down" timer on all abilities, which is set to 1000 milliseconds.
The design goal of soft limits in any game is to allow players to create builds that let them execute abilities more often depending on build choices made, and not to have hard ceilings on damage or healing per second. This gives a lot of control to the player, which makes for a really fun and interesting system, but it can lead to situations where players cast too many abilities too quickly (and continuously) if strict limits on resources are not enforced.
Over the years, player power has grown considerably. With the addition of the Champion System, various armor/weapons sets, and changes to abilities, we have reached the stage where players - with the right build - can cast near-infinite numbers of abilities. If you add in a properly managed group, with some focused on damage and some focused on healing and regen boosts, you have a perpetually running never-ending stream of abilities.
With that in mind, consider how Area of Effect (AOE) abilities work: when cast, they look at a specific target area - almost always the area directly around the caster - find targets, and perform the ability. Each of those steps requires server calculations.
At launch, Cyrodiil’s processes were able to keep up with the number of AOEs cast, because most players couldn’t cast that many of them: they ran out of Magicka or Stamina, so they had to use AOEs judiciously. Over time, as player knowledge grew and regen builds grew in power, more players could cast more and more AOEs before running out of resources.
During these years, we found and fixed many performance issues in Cyrodiil, and we continue to do so. We've fixed significant issues, uncovered more, and continue to find and fix with every Update.
However, one foundational issue remains. At some point, we crossed a threshold where most players in PvP were able to cast endless AOE abilities, without ever running out of resources. This is done through player knowledge, builds and group mechanics – resulting in a constant stream of AOEs with many players never using any other type of ability.
This is not what we intended, but part of the fun of Elder Scrolls games is designing a build that has unexpected and powerful results, and we allowed it. However, as this behavior grew more prevalent, we reached a point where casting so many continuous AOE abilities in such a small area started to overwhelm the server process for that area, leading to situations where the "lag meter" spikes and the server becomes unresponsive for a period of time.
Our initial response to this problem – starting with Update 22 – was to find (and fix) more than a few problems with AOE ability calculations and make them more performant - and we stayed (mostly) ahead of the problem. But as more players reached maximum Champion Level and more players started utilizing this particular method of "AOE spamming", we have reached the point where we cannot fix this issue around the edges: we need to address the core problem, which we will be starting with Update 27.
To do this, we will need to first do some analysis, and we can only do this on the live servers. As much as we try to avoid running tests on live servers, they are the only place where the combination of player behavior, specific builds, and mass battles happen. So, starting the week of August 24, we will be running a series of tests on the live PC servers – both NA & EU, only in Cyrodiil. Please note that we will not be running these tests on any Console servers. Each test should take about a week, but if needed, we will extend the testing time.
The first round of tests we are planning will focus on Area of Effect (AOE) abilities in Cyrodiil and will make it more difficult for AOE abilities to be the only abilities used, adjusting cooldown, cost and regen values of all AOEs (damage and
Specific details on the tests we will be running in Cyrodiil are as follows:
- Test 1 – Shared global AOE cooldown - 3 second timer. This test adds a global 3 second shared cooldown to any AOE ability. This means that when you cast an AOE, you will not be able to cast another for 3 seconds. For example, as a Templar, if I cast Ritual of Retribution, I wouldn’t be able to cast Puncturing Sweep for 3 seconds.
- Test 2 – Individual AOE cooldown - 3 second timer. This test adds an individual AOE cooldown to each AOE ability. This means that when you cast an AOE, you will not be able to cast that same AOE ability for 3 seconds. For example – as a Templar, if I cast Puncturing Sweep, I wouldn’t be able to cast Puncturing Sweep again for 3 seconds.
- Test 3 – No cooldown, global ramping AOE cost. This test adds a global ramping AOE cost for each AOE cast. Similar to how streak or roll dodge works, where when you cast an AOE you receive a debuff for 5 seconds, each stack of the debuff increases the cost of any AOE cast.
- Test 4 – Individual AOE cooldown – 3 second timer, global ramping AOE cost. This test adds an individual AOE cooldown to each ability as in test 2, but also combines that with a global ramping cost from test 3.
During the times that any of these tests are active, we will be awarding double Alliance Points for anyone active in Cyrodiil.
After we complete the above tests, we may try other combinations of cooldown, cost, and regen values on AOE abilities. However, we need to run these tests first and then assess the data. We will then let everyone know what we found and how we will move forward. We will be very upfront, but please be aware that if these tests confirm our hypothesis, then chain-casting AOE abilities will no longer form the core of the ESO PvP experience in the way it has for the last few years. We would then go through each class and ensure that there are viable builds for each and make adjustments as necessary.
The code for these changes will be going into today’s PTS patch, and we will be running some basic tests throughout the day on Tuesday in Cyrodiil to ensure we’re able to make the above changes without requiring any maintenance or downtime. Once we launch Update 27 for PC on August 24, we will announce when one of the tests are beginning through an in-game announcement and will have a forum thread detailing which test is currently being performed. Remember, these tests will be limited to Cyrodiil, so AOE abilities in other PVP spaces (Imperial City and Battlegrounds) and will remain unchanged for now. We will evaluate how these tests go, and let everyone know next steps.
Thanks for being patient and understanding with these tests. They are extremely important for the team and will help us work towards improving Cyrodiil performance.