World of Warcraft

1 . 2 . 3 . 4 . 5 . 6 . 7 . 8 . 9 . 10 . 11 . 12 . 13 . 14 . 15 . 16 . 17 . 18 . 19 . 20 . 21 . 22 . 23 . 24 . 25 . 26
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 0. Pre 3.1 Exhaustive Tank comparisons   03/08/2009 02:55:10 PM PDT
limit-reached limit-reached
BEFORE YOU READ: Understand that this is not the work of Tankspot, Xav, Ciderhelm, or goldfish. This is from an individual on the EU realms, Parri. http://eu.wowarmory.com/character-sheet.xml?r=Dentarg&n=Parri

This is his original post:
http://forums.wow-europe.com/thread.html?topicId=8246768118&sid=1


He asked me to mirror it here because he doesn't have a US account to post it on.
------------

What follows is - I think - the most in depth comparison between the tanking classes that has ever taken place. Before I go into the methodology though, I would like to point out some ground rules for the discussion that will follow this thread:

1.) Sartharion + 3 Drakes is a banned topic. Seriously. Stay away from this landmine. This is not a Sartharion topic. Sartharion+3 in both 10 and 25 man versions is a gimmick fight that even Druids have to use cooldowns for, and they're one of the better tanks on this fight. Is it easier for some tanks? Yes. Is this okay? YES. Because it has been done many, many times with a Warrior or Paladin tank. It's not impossible without a Druid/DK, just harder. Now you can make the same arguments about our dear Shears and Fears, but the point is, it is perfectly acceptable for some bosses to be harder for some classes than others. Even if that boss is currently the hardest in game. Especially if that's the case. This is not "yet another Sarth+3" QQ topic. Do not turn it into one.

2.) Do not muddy the water. By this I mean, do not try to justify deficiencies in one class for an entirely different deficiency your class has. Tanking has many different aspects which are important to it. Some are admittedly more important than others. Just as you cannot consider them in a complete vacuum, you have to accept that there is some kind of balance between all tanks in each of the areas. Being slightly above or below is fine. But justifying absurdly high threat because your relative survivability is abysmal is not - and vice versa. Because some tanks are better at picking up and holding aoe threat does not mean they deserve less survivability. There is enough variability within the whole survivability field (because it's seriously massive) that each tank has plenty of room for distinction and that dreaded "flavour" within it.

3.) Leave your baggage at the door. And I'm not just talking about butthurt hybrids here that like to see other classes suffer for perceived injustices in previous expansions. It's helpful to no-one. I also mean any well meaning individual who perpetuates various class myths that are either out of date, or were flawed to begin with. One I can name offhand is: the idea the block may scale too well with endgame gear. It doesn't, it won't, it will never. Whichever developer got the idea that it would was quite clearly mistaken. They are mistaken a lot.

4.) This is not about Polar gear. It's not about the amount of health or avoidance a tank can reach. It's not about the armour difference. It's not about any one single tank issue by itself. You cannot properly examine the differences in tank classes when you take each property of the tanks stats by themselves and disregard how they interact with each other. The point of this thread is to get a measure of the overall strength of tanks. Not to point out individual strengths and call foul when not considering what drawbacks are made at the expense of those strengths.


With that out of the way, lets take a look at the important factors that sum up tanking, and specifically, the ones we'll be looking at in this thread. Boiled down to its very roots, tanking is about 3 things which are variably important depending on the encounter.

1.) Staying alive
2.) Reducing the damage you taken
3.) Holding aggro

To anyone who hasn't tanked before, 1 and 2 look remarkably similar. They're not. They have a relationship - to be sure. But it's more like an internet romance than anything serious. Gearing for survival and gearing to avoid damage are entirely different schools of thought, often headed by belligerent posters on either side insisting they're right. It's a pretty shaky internet romance.

80
View All Posts by This User Toggle Ignore / Unignore This User
  • 1. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 02:56:13 PM PDT
limit-reached limit-reached
Surviving is about constants only. Armour mitigation, health. The Warrior Ciderhelm summed it up really well by calling it not a question of how much damage you take, but how much you can take. If you assume a worst case scenario, how quickly will you die. Every class will die eventually if they don't get a heal, but often you find that healers can have other things to do. They may get silenced, stunned, punted away, some may die or may be forced to heal the raid through heavy incoming damage. This leads to a potential period of time where the tank is taking no healing. If you get unlucky and don't avoid those incoming attacks, how long will you live for a heal to reach you? That's survival. Similarly, you could get a debuff, or the boss gains a buff that allows him to heavily burst your HP. At some point, the burst can be too much that it's impossible to expect a healer to react to or anticipate. The time you survive is important here because - assuming you don't, or cannot avoid the incoming damage, you must take it. The longer you can survive it without dying means the longer a healer has to react. Unhealable can become healable with enough health and enough healers. Always.

Now, that's not to go saying that survival is everything. There is a caveat, which is part 2. At some points within the game, we find that tanks are tested in a different way. It is true that theoretically, if you throw enough health and enough healers at something, it will eventually become healable. Realistically, we often do not have this option. The encounter can have harsh DPS checks that limit you to a certain amount of healers to kill it in time. We have limits to just how much health we can get. Sometimes, the health threshold required to remain reliably healable over a period of burst damage is impossible to get within the gear we have available to us. Many argue that this was the case with Brutallus. You would take Stomp Damage, then simultaneous MH+OH hits. The amount of health required to survive Stomp + MH+OH+MH+OH while at 50% armour was not reachable. Stomp hit often enough that that it could not be negated by cooldowns alone. Our only option then is to trust to luck. Instead of dying reliably, you avoid stamina stacking and chose to live semi-reliably through avoiding that damage. Now of course, sometimes a boss like Brutallus would just decide: you're going to die. And you did. No fault to the healers, and nothing more you could have geared for. There is the potential even with high avoidance to take strings of attacks together and still be unhealable. In fact, even more unhealable than you would have been without it, because all that avoidance you spent your allowance on was looking the other way while you got blown up. The point is, in a roundabout way, avoidance is what you aim for when either::

You are in no danger of dying.
You are in such grave danger of dying that it's impossible for you to reach the health plateau that ensures your survival. You can avoid death from a bullet by wearing a bulletproof jacket, but if you can't afford the bomb-shelter, then don't try to catch the missile.

Holding aggro: Make sure the boss hits you and not someone else, Fairly easy now, and relatively balanced between most tanks apart from Paladins, who are 10% or so ahead. This isn't a topic we'll be going into. And like discussed in rule 3, not any reason to balance survival or damage intake against.

Methodology

So, how are we going to do it? How are we going to compare all tanks in a way that makes sure – as close as possible – that each tank is fairly represented? The answer is: by not relying on any one single example. Get as many as possible from different sides of the spectrum, against different types of bosses, and compare that to different setups of other tanks against the same hypothetical boss. The importance here is that we fully represent the variability of boss design by making sure that the most typical mechanics used in-game are represented in the tests. The theory is that if each tank is suited to different types of tanking jobs then this will show through in our testing. One tank may be inferior against one type of boss, but it should for the sake of fairness be equally superior to a similar number of different bosses. The point is not to exactly replicate bosses of the past and test tanks against old content – but we would be foolish in the extreme not to recognise boss archetypes that Blizzard have used repeatedly in Vanilla and an expansion pack's worth of raiding encounters. We should recognise these archetypes as what they are and expect to see variations on the theme in future encounters.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 2. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 02:57:45 PM PDT
limit-reached limit-reached
So, we know what we're going to do. We're going to make various gear profiles of each of the tanks and compare these profiles against each other and with the profiles of the other tanks. We then put each of these profiles against a wide range of different bosses with different types of mechanics that should in theory promote different types of tanks at different times. That's the plan. But what if it doesn't happen?

Point of failure

Before we get into the nitty gritty of the numbers and graphs, we have to state a reasonable consensus whereby it becomes obvious that one or more of the tanks has failed the test. These are, I hope, common sense and I'm wasting my time including them, but there are still some bigots out there who believe things as stupid as Warriors deserve to be last because they've had it so good for so long, to things like Warriors deserve to be top because they've been the tank for the longest and don't have the options that Paladins/Druids have to sit back and heal. I'm sure depending on your perspective, either or both of those opinions are distasteful.

At what point is it clear that one tank is failing the test? It'd be easier to start by saying what doesn't cause a tank to fail. Taking more damage doesn't imply failure. Lower EH doesn't imply failure. Less time to live during a burst, does not imply failure. Less avoidance does not imply failure. Being worse at a boss does not imply failure.

If none of these things imply failure, then what does? When there is a consistent pattern of a tank under-performing compared to others. If a tank takes more damage than any other, more burst than any other with less HP than any other, there is a problem. If it's just against one boss, then maybe it's just a weakness against that type of encounter. It does not necessarily mean there's a large flaw with a class – perhaps just a small tweak required to balance the level of encounter difficulty when tanked by one class compared to another. If however this drawback remains consistent – and this part is the most important – with no similar uprise in performance in a similar number of realistic encounters, then it is likely that there is a more systemic problem with the class that is not merely a weakness magnified by encounter design. Failure of the test is not just down to bad performance though. Failure is measured by not adhering to an overall average. A tank that performs far better than any other on most fights has also failed.

At the time of writing this, I haven't run the full tests myself, so I don't know for sure how it will go. I have my beliefs, which are no doubt well known if you've seen me post before. I could however be entirely wrong. I hope I am, because it would mean my concern would be nothing more than needless worry. Regardless of the results, I will post them because the worst they can do is inform other people who may be similar to myself in being sure that there is some deficiencies with the class they play. And importantly, because the tool I have developed for this purpose I have found extremely useful on a personal basis to plan for gearing against particular bosses. It goes a level beyond traditional modelling programs and gives real numbers, which I can appreciate more than an arbitrary rating system given by the modelling programs available today. The tool, called ThinkTank, will be available later in the post to download, view the probably messy source code to make sure I haven't skewed the results and test things for yourselves.

The setup

Wait! Rules first. Boring I know, but they maintain a certain level of sanity over our results that makes sure that they are all fair, relevant, and importantly: realistic. We must always remember that although what we're working with has the potential to be incredibly accurate, that bad input leads to bad output.

Bosses:

No Gimmicks. No Fears. No Shears. No one-shotting breath from a dragon on steroids. These are gimmick encounters and should be treated as such. Balance on these is tenuous and not at all representative of the actual tanking strengths of the class. They test the toolbox rather than the strength of the class. We know by Blizzard's own admission that they do not design bosses specifically with the intent in mind that they're suited for one particular tank. Therefore, neither should we.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 3. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 02:58:48 PM PDT
limit-reached limit-reached
No boss will have a swing timer of anything under 1.0 seconds. We've seen DW bosses and fast hitting bosses, but to my experience, nothing that goes beyond 1.0 after the TC debuff is up. If it's never been seen, then not only is it never likely to be seen, but it's also likely that if it ever is, it'll be as a gimmick encounter and thereby falling under rule 1. Similarly, going the other way to prove a point is equally out of the question. Upper limit for boss swing speed is 2.5. A non-DW boss will therefore swing a maximum of 10 every 10 seconds and a minimum of 4 times every 10 seconds. A DW boss will have a max of 20 swings every 10 seconds and a minimum of 8 swings every 10.

Every boss will be hard. I will test every boss I simulate against a sort of simulation Combat Dummy. The Combat Dummy will have 0% dodge, 0% miss and 0% parry. It will have 0 armour. If the damage output of the boss is not above 20k dtps on the dummy, then the boss is either retuned until it is, or scrapped if it's determined that it's nearly unworkable to have a boss that functions in that way and still does outputs a challenging level of damage. This is a very important requirement because there are numerous bosses in the game that have almost no chance of killing your tank unless every single one of your healers falls asleep. Think Noth or Heigan. Any of the tanks being “good” against trivial bosses is not a result that anyone cares about, so it is a test that we do not care about.

(That's not to say that bosses that aren't at risk of killing the tank aren't or can't be a challenging encounter for us, but the point is, the challenge we're looking to show balance between is that of survival; of which a large discrepancy cannot be justified from outside sources.)

L2read

Now we must learn to read. Not literally of course, you're reading this now! What we have to learn to read is the data that ThinkTank gives us, and learn to interpret these statistics so they provide us with valuable information. Lacking the proper interpretation, the output is meaningless data. It is our understanding that makes it information. So, a quick run through of the things that ThinkTank will be tracking – the important stats that individually pinpoint weakness, and considered as a whole, show the strength of the class in tanking.

DTP/S: Stands for Damage Taken Per/Second. Fairly obvious, this is the average figure of total damage taken divided by the total time the fight lasted. This is one of the most important tank statistics. It's not the primary focus, but taking less damage is always something that makes one tank easier to heal than another – assuming equality elsewhere.

Lowest TTL: This is the most useful and accurate TTL figure that ThinkTank tracks. Every time the tank takes damage, it is recorded as a potential starting point of a string of burst damage that can kill a tank. As the simulation progresses, it keeps tracking these periods of damage and when it finds a newer, faster tank death, it records the time taken in MS. This is the figure of reference that should be used for every tank, being more accurate and more applicable than any of the other TTL variations that ThinkTank tracks. This is the figure that matters, the other TTL's are merely references and framing around this one.


Armory lasts a day, Rawr profiles are forever

There will be no chardev or armoury profiles of these characters. Chardev has too many bugs to be considered reliable, and an armoury profile can dramatically change within a few hours, leaving no record behind. Each of the profiles will be made on Rawr and available to download. In addition, no character profile will use PVP gear at all. We'll start now:
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 4. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 02:59:50 PM PDT
limit-reached limit-reached
Druid

We're going to take 3 Druid profiles. The spec will be the same each time, because there really isn't much variability in Druid tanking specs.

The first profile, Druid#1 is an avoidance nutjob. He'll use double avoidance trinkets and will favour a higher agility item over a higher stamina item if the quality of them is comparable (similar ilevel). He'll gem for agi in every single red slot, and when there's a a blue slot with a nice bonus, he'll put agi/stam in. He's not a complete moron though, he recognises that HP is useful as is armour, so he will not pass up a large amount of these stats to get a small gain in agility.

Druid#2 is the second profile. He prefers a balance. He liked avoidance up until the point where he started to see very little gain on his character sheet for larger and larger investments. Now he rolls with one stam and one avoidance trinket. Most gems including red ones are stam/agi.

Druid#3's favourite show is Spongebob Square pants. Spongebob reminds him of himself – and it's not because of the square pants, because he can't wear pants in bear form. This Druid drops almost any agility he can within reason to pickup more HP. Every blue gem is 24 stam. Every red socket with a bonus he doesn't care about is 24 stam. Red sockets with a nice bonus, he puts a stam/agi gem in and curses his luck. He uses a pair of stam/armour trinkets to give him the most EH. He drops better ilevel enchants on several items to pickup more stamina. The other tanks in his guild cry themselves to sleep when they think about his HP.

Links:

Druid#1: http://www.kamuf.org/parri/profiles/Druid_High_Avoidance.xml
Druid#2: http://www.kamuf.org/parri/profiles/Druid_Balanced.xml
Druid#3: http://www.kamuf.org/parri/profiles/Druid_High_Stam.xml

Summary:

Druid#1: 38561 armour, 44481 health, 50.33% dodge, 5.25% miss
Druid#2: 38354 armour, 48541 health, 47.62% dodge, 5.21% miss
Druid#3: 38735 armour, 50481 health, 43.09% dodge, 5.21% miss

Now I'm sure some will feel compelled at this point to say: You could have gained 1% here or 3% there. This isn't the point. The point is to create realistic profiles at different sides of the HP/Avoidance spectrum. It's not min-maxing to see the height of what's possible, we just need a large enough difference between all 3 to get a good idea of what's going on. For a long time, any other tank that suggested on the forum that Druids found it easy to reach 50k HP were bombarded with complaints from very bad Druids complaining that any gear with 50k HP was worthless and only idiots would wear that. 50K HP was met with the sentiment within the Druid community that it was only possible by stacking stamina at the expense of everything else and even going so far as to wear Polar gear in several slots. For every one of the above examples, this is not the case. Normal gear is used in every slot. What differs are the occasional jewellery swaps and gem/enchant changes. There is no evidence whatsoever to suggest that Druids can only get to 50k HP by sacrificing an absurd amount of stats for it. Is there a sacrifice? Yes. 6K HP for 7% avoidance between the upper and lower bound. If you don't want to sit on an extreme, you can easily pick the more balanced option, and trade a mere 2.5% avoidance for 4k hp. Even the Druid with over 50k HP however manages to get over 48% combined avoidance, which is not far from a Warrior's level of avoidance at all. Getting 50k hp does not require Polar gear, it does not require double stam trinkets and it does not require min-maxing professions like JC. The 50k hp profile here has none of those things.

I have news for Druids that think these levels of health aren't realistic: they are. Missing 2.5% avoidance does not turn you into a mana sponge. So I really do not want to see any of the traditional garbage spouted by bad Druids that these profiles are unrealistic. They're as realistic as they come.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 5. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:01:05 PM PDT
limit-reached limit-reached
Warrior

The Warrior in this setup is talented for the most common mitigation talents a Warrior will pick. A lot of Warriors nowadays end up dropping mitigation talents to pick up threat talents that they feel are necessary to remain on par threat/dps wise. I view this spec however as an aberration. Warriors cannot use the excuse that they have to drop mitigation to remain competitive in threat because all that does is shift the problem away from its real area. If Warriors have a threat issue, that should be dealt with separately from balancing the tanking classes when damage intake/survival is the issue. So none of the Warrior specs/profiles will be spec'ing for Impale/Deep Wounds. Similar to the Druid, we're going to take 3 separate Warrior profiles with a varying focus primarily on avoidance vs stamina in each of them.

Warrior#1: http://www.kamuf.org/parri/profiles/Warrior_High_Avoidance.xml
Warrior#2: http://www.kamuf.org/parri/profiles/Warrior_Balanced.xml
Warrior#3: http://www.kamuf.org/parri/profiles/Warrior_High_EH.xml

Warrior#1: 27166 armour, 36783 health, 30.29% dodge, 19.75% parry, 21.90% block, 1450 block value, 9.06% miss
Warrior#2: 26855 armour, 40008 health, 26.72% dodge, 18.37% parry, 27.11% block, 1500 block value, 8.84% miss
Warrior#3: 27052 armour, 41668 health, 25.04% dodge, 19.01% parry, 23.07% block, 1530 block value, 8.86% miss

Paladin

Paladin tanking talents are pretty easy, there are very few variances in how the talent spec is picked. Gearing however can be considered potentially different from a Warrior. Due to Holy Shield, even at the first tier of gear, a Paladin can guarantee that as long as he isn't stunned or silenced, he can always block. Now, what's the difference between a Paladin at a 5% chance to be hit and one at 0% chance? It depends on how you look at it. The Paladin with the 5% chance to be hit will not be taking a massive amount more damage. In fact he could be taking less, depending on how much real avoidance the other Paladin traded for 5% block. But, as tanks we know that a minor difference in the overall damage taken in the fight means almost nothing. The real issue on hard bosses is reducing potential burst damage. That Paladin with the 5% chance to be hit can be hit twice in a row without entering the realm of absurd probability. It's a general rule within the Paladin community that you go for the 102.4% number. So each of the Paladin gearsets being tested will adhere to this rule.

Paladin#1: http://www.kamuf.org/parri/profiles/Paladin_High_Avoidance.xml
Paladin#2: http://www.kamuf.org/parri/profiles/Paladin_Balanced.xml
Paladin#3: http://www.kamuf.org/parri/profiles/Paladin_High_Stam.xml

Paladin#1: 27309 armour, 37096 health, 27.56% dodge, 18.42% parry, 14.45% block, 1677 block value, 9.57% miss
Paladin#2: 27269 armour, 40101 health, 24.99% dodge, 18.25% parry, 17.43% block, 1689 block value, 9.33% miss
Paladin#3: 27269 armour, 41611 health, 23.59% dodge, 18.25% parry, 18.83% block, 1741 block value, 9.33% miss
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 6. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:02:20 PM PDT
limit-reached limit-reached
Death Knight

Death Knights will show up on each of our graphs 3 times, once for each spec. For simplicity and fairness sake for cross-spec comparison, each DK will use the same gear. Unlike the previous chardev profiles, these are realistic specs which aren't tuned for a gimmick encounter. It is correct to assume that high effective health and high burst effective health are useful, but creating strawmen out of a spec that provides just these things at a huge cost to a normal threat output doesn't prove anything. Will we see encounters that require high burst EH again? Probably. Almost certainly. But will those encounters have no threat requirement for the tank to maintain? That's doubtful. And a note in advance, while I do understand quite well the DK tanking mechanics, the threat details are something I'm not at all sure about and don't claim to know. If the specs are a little “off” because of this, then this is why.

So first of all, the avoidance setups:

Death Knight (Blood): http://www.kamuf.org/parri/profiles/DK_Blood_High_Avoidance.xml
Death Knight (Frost): http://www.kamuf.org/parri/profiles/DK_Frost_High_Avoidance.xml
Death Knight (Unholy): http://www.kamuf.org/parri/profiles/DK_Unholy_High_Avoidance.xml

Death Knight (Blood):30692 armour, 38928 health, 29.46% dodge, 32.90% parry, 10.13% miss
Death Knight (Frost): 30692 armour, 37528 health, 29.46% dodge, 32.72% parry, 13.13% miss
Death Knight (Unholy): 30692 armour, 37988 health, 29.46% dodge, 32.79% parry, 10.13% miss

Balanced setup:

Death Knight (Blood): http://www.kamuf.org/parri/profiles/DK_Blood_Balanced.xml
Death Knight (Frost): http://www.kamuf.org/parri/profiles/DK_Frost_Balanced.xml
Death Knight (Unholy): http://www.kamuf.org/parri/profiles/DK_Unholy_Balanced.xml

Death Knight (Blood): 30692 armour, 40048 health, 28.80% dodge, 32.78% parry, 9.94% miss
Death Knight (Frost): 30692 armour, 38578 health, 28.80% dodge, 32.60% parry, 12.94% miss
Death Knight (Unholy): 30692 armour, 39068 health, 28.80% dodge, 32.67% parry, 9.94% miss

And the stam heavy setup:

Death Knight (Blood): http://www.kamuf.org/parri/profiles/DK_Blood.xml
Death Knight (Frost): http://www.kamuf.org/parri/profiles/DK_Frost.xml
Death Knight (Unholy): http://www.kamuf.org/parri/profiles/DK_Unholy.xml

Death Knight (Blood):30692 armour, 42888 health, 27.24% dodge, 29.51% parry, 10.51% miss
Death Knight (Frost): 30692 armour, 41258 health, 27.24% dodge, 29.33% parry, 13.51% miss
Death Knight (Unholy): 30692 armour, 41798 health, 27.24% dodge, 29.40% parry, 10.51% miss

Without further ado, lets meet our bosses!
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 8. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:03:22 PM PDT
limit-reached limit-reached
We'll be starting simple and ramping up slightly in complexity. Simple is as follows: We have a boss with a melee attack. He has nothing else but straight melee damage to the tank. It's highly important that we understand the melee damage breakdowns that tanks take, because the vast majority of boss encounters have the majority of your tank damage coming from the bosses auto-attack. Even bosses with powerful special attacks and various dots, aoe damage, and debuffs rely a great deal on their melee damage prowess. So we're going to say the boss does 22000 damage when swinging every 1.0 seconds, and scale that up to 55k damage at 2.5 seconds. Overall DTPS is the same, but distributed differently.

The damage output of this boss should remain static throughout these against the combat dummy, and it does. I won't bore you or waste time with the results because this is beyond obvious. The question obviously being posed here is, at higher and lower attack speeds, do some tanks perform better than others due to their differing mechanics? It's an oft said line that Paladins and Warriors will significantly outperform other tanks at a high enough speed, due to the linear nature of block value vs damage incoming. This makes sense and it's probably true. But the question is, by how much? At what point is this reached? And how far behind are the Warrior/Paladin behind at the other side of the scale? Lets see:

http://www.kamuf.org/parri/TestOne.html

I know, it's a lot of numbers. And no, I don't expect you to pour over them, but they are there in case you are interested. The graphs of the important figures on these fights:

DTPS

Avoidance setup: http://www.kamuf.org/parri/graphs/TestOneDTPSAvoidance.png
Balanced setup: http://www.kamuf.org/parri/graphs/TestOneDTPSBalanced.png
Stam setup: http://www.kamuf.org/parri/graphs/TestOneDTPSStamina.png


Okay, now we see something meaningful. To be clear, the graphs show the average Damage Taken Per Second of each of the 16 fights for each of the 12 profiles. The higher the lines get means the more damage the tank is taking. And we do immediately see a large difference between classes. As expected, the Druid figures are pretty constant throughout. The Warrior figures however are a fair bit lower when around 1.0-1.3 attack speed, but shoot up fairly rapidly beyond that to taking significantly more damage than the Druid. Paladins are in pretty much the same boat. They take a bit more damage on average due to the lesser avoidance, but block more often. Right away, we go to dispelling our first myth. It's a myth so ingrained into tanking communities that even a lot of Warriors and Paladins believe it themselves. There was a reasonably well done article too on WowInsider that fell victim to it, and I'll quote the relevant part here:


Q u o t e:
Warriors and Paladins are also designed as avoidance tanks. Avoidance is vastly superior to health in TTL (Time To Live) calculations for nearly every situation, barring that of the burst magic damage described above. Over the course of most encounters, Warriors and Paladins take less damage than Druids and Death Knights
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 9. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:04:25 PM PDT
limit-reached limit-reached
As we already saw from the profile creation, the avoidance difference between shield tanks and Druids is almost negligible. If you wish to make such a big deal out of 1-4% avoidance then perhaps the same should be done about the 2% advantage on Protector of the Pack. Furthermore, this test is an example to an encounter that doesn't hit for particularly hard amounts at the low end of the scale, yet Druids are still within close reach of the DTPS levels. As soon as the boss starts to hit a little harder, and a little slower, the Druid quickly overtakes. On any normal, average encounter, a Druid will be taking considerably less damage than a shield tank. At its very worst, you could call this level of DTPS to be balanced when viewed across the board. Lets imagine for a second that it is. But it begs our first question: Even if DTPS is the same even assuming that Warriors and Paladins have more avoidance, then why do Druids have more HP? I'll remind you of the HP delta on each of those setups: #1: 8k, #2: 8.5k, #3: 9k. That's a massive difference, and it has to be justified somehow. Perhaps in how the potential burst damage can affect each of the classes. We'll see about that shortly. Regardless, the myth that Paladins and Warriors take less damage than a Druid/Deathknight is clearly not holding true for this example.

What about the Gorilla? Yeah...I'm getting there. Just look at DKs. It's ridiculous. In every single setup they take orders of magnitude less damage than every other tank. Blood is, as far as DK only comparisons go, the weakest spec. But even Blood considerably outperforms every other tank at every other setting. Blood also carries with it other advantages that are not represented in a DTPS format. The DK stam heavy set with over 42k hp takes less damage than the Warrior/Paladin sets at 6k lower HP. I doubt that the itemisation even exists to allow either of the shield tanks to catch up on avoidance to a stam DK. A Paladin, Warrior or Druid simply cannot gear themselves to the point that they take less damage than a DK. Even if they get to the point of 6k less HP, it's just not possible.

TTL

DTPS isn't the only important stat however. It's a big one, but not the biggest. For progression viability, a tank has to prove his worth at standing up to both predictable and unpredictable (RNG) burst. A tank that takes less damage could theoretically be benched on hard progression bosses because he lacks the ability to take a significant amount of damage when it does eek through. Thankfully, ThinkTank tracks strings of damage that would kill a tank, and records the fastest death recorded within a 4 day time period. At this duration, we do see significant occurrences of deaths that are just incredibly unlikely. So ThinkTank also tracks the number of times a death at this speed happened, and in addition, the highest amount of overkill on that death. These figures are all in the previously linked test results table. Now we'll see in a more readable format.

Avoidance: http://www.kamuf.org/parri/graphs/TestOneLowestTTLAvoidance.png
Balanced: http://www.kamuf.org/parri/graphs/TestOneLowestTTLBalanced.png
Stamina: http://www.kamuf.org/parri/graphs/TestOneLowestTTLStamina.png

Surprise surprise. I wish I could say I was shocked to learn that the tank with the most health that took equal/lesser damage survived burst easier. I'm not though. What's very interesting here is the DK representation though. On the DTPS front, they were a country mile ahead of everyone. Here however, they're about equal. We see Blood doing a lot better here thanks to the effect of Will of the Necropolis on attacks that would otherwise kill. It does not always proc, but the results quite obviously show that often, that a Will of the Necropolis proc is enough to survive another hit. The disadvantage Blood showed earlier is compensated in comparison to the other DK specs by a substantially higher time to live through burst damage. Notice also the absurdly high results obtained by Paladins when stacking stamina and taking a lot of smaller hits. The effect of Ardent Defender here is really noticeable. It's a talent largely shunned by Paladins because it can be leapfrogged, but we can see even at the slower end of the attack speed that it can be the difference between dying and living another attack.

Apart from the Druid results, I'd feel fairly comfortable in calling those results pretty even. Some are slightly above/higher than others, but not hugely so. An important thing to notice also was that although Warriors had an advantage in DTPS over Paladins, they did fare weaker in survival. Also, Warriors outperformed Druids on DTPS on the fast hitting side of the scale, but this was not an advantage that translated into TTL. The Druid's huge HP and reliability of their armour allowed for the Warrior to fall behind in a worse case scenario of not blocking some attacks in a row.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 10. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:05:45 PM PDT
limit-reached limit-reached
Now I can just imagine the DK complaints. You need the avoidance because your mitigation is inferior. I just don't believe that. The numbers don't either:

Average Damage Per Hit: http://www.kamuf.org/parri/graphs/TestOneAverageDamageHit.png

The only DK spec to take more damage per hit than the Warrior and Paladin is Blood, which gets away with it by having an unrepresented large amount of self healing, a last stand on a minute cooldown and Will of the Necropolis adding a significant amount to the DK hp. As things stand before 3.1, a DK will not only avoid more hits, but they'll hit him for less too. On top of an unquestionable superiority on magical damage.

Test Two

If only one test proved everything. If it did, I'd be finished already. However, 1 test, even though it offered 16 iterations through 12 profiles is not enough. The DTPS from our tanks in that last test was a little weak. Not challenging enough I think. So lets see what happens when the boss turns it up a notch. The next series of tests is similar to the last, but the boss will do 25% more damage. How will this affect the Paladin/Warrior mitigation via block?

Results: http://www.kamuf.org/parri/TestTwo.html

DTPS:

Avoidance: http://www.kamuf.org/parri/graphs/TestTwoDTPSAvoidance.png
Balanced: http://www.kamuf.org/parri/graphs/TestTwoDTPSBalanced.png
Stamina: http://www.kamuf.org/parri/graphs/TestTwoDTPSStamina.png

And now we see what really happens to Block. Block's power on our first test at the faster end of the attack speed actually has nothing to do with the weapon speed. It's all about how hard you're getting hit. There can be bosses where fast attacking also means weaker hits, but this is largely not often the case. Any boss which is designed to potentially kill a tank fast needs to do more than just hit fast. He needs to hit fast and pretty hard. So not 5k per hit. More like 10k per hit. At this point, the worth of Block plummets. The point is, it is not enough to merely design a fast attack speed boss and think it's going to favour shielded tanks. Because if it has the potential of killing a tank, it will do more damage and is more likely to kill a Warrior than a Druid. Regardless of block.

Time to Live

Avoidance: http://www.kamuf.org/parri/graphs/TestTwoLowestTTLAvoidance.png
Balanced: http://www.kamuf.org/parri/graphs/TestTwoLowestTTLBalanced.png
Stamina: http://www.kamuf.org/parri/graphs/TestTwoLowestTTLStamina.png

Once again, we see a different story. The previously hideously overpowered Death Knights are right down along the bottom with Warriors. Paladins are jumping all over the place, and this is solely due to the effect that Ardent Defender has, occasionally allowing them to survive an extra hit. If you look up, you can perhaps see Druids far off in the distance with a time to live almost twice that of any other tank. How? Druids didn't take significantly less damage than Warriors, and they did take a lot more than Dks, so how do they survive for so much longer? The Druid is a constant. That's how. A DK can be hit without his cooldowns up. He can be hit 2-3 times in a row like that. When this happens, he is just as squishy as any other tank and dies just as fast. Now, if you take it a step further and look into the figures behind the TTL graph, you'll see that the frequency of Dks dying at that speed is very low. Merely a fraction of the times that Paladins and Warriors died. This is down to having a much higher level of avoidance, and a much higher uptime on cooldowns that trivialise incoming damage. A DK may be only vulnerable to damage spikes for 10 out of every 60 seconds, and for those 10, he's still 12-15% more likely to avoid them. The Druid however, cannot be hit without his armour up. It's always there, like a constant % modifier block or cooldown. His worst case scenario is merely a melee hit. Couple this with a health pool that is often the difference between surviving another hit or two and you have your answer.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 13. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:07:14 PM PDT
limit-reached limit-reached
Average Damage Per Hit: http://www.kamuf.org/parri/graphs/TestTwoAverageDamageHit.png

Similar to the last test really, Druids in the lead again, everyone else lagging behind. Probably balanced when you consider Druids as slightly more likely to not avoid an attack.

Overall, this test simply proved that attack speed has no direct effect on making block a worthwhile mitigation stat. Only so far as a fast attack speed would impose a lower amount of damage on the swing.

Test Three

But this last test was also fairly weak. There is more than one way to increase a boss auto-attack damage beyond simply adding 0's to the end. I know, shocking isn't it? It's a way that Blizzard's encounter designers seem to neglect. Dual Wield. For our next test, the Test Two boss will pick up another weapon and start swinging them both at once. Our Dual-wielding behemoth will swing OH and MH at the same swing speed. Like most challenging DW bosses, he does not suffer from the 25% miss penalty. The OH damage is based on his MH damage, but at half the power.

Results: http://www.kamuf.org/parri/TestThree.html

DTPS

Avoidance: http://www.kamuf.org/parri/graphs/TestThreeDTPSAvoidance.png
Balanced: http://www.kamuf.org/parri/graphs/TestThreeDTPSBalanced.png
Stamina: http://www.kamuf.org/parri/graphs/TestThreeDTPSStamina.png

So, a DW boss should theoretically favour shield tanks, right? Wrong apparently. It is only on the very extreme end of the attack speed scale that Warriors and Paladins take less damage. Once again, the mythic advantage of a shielded tank is just that: a myth. It holds no actual truth.

TTL

Avoidance: http://www.kamuf.org/parri/graphs/TestThreeLowestTTLAvoidance.png
Balanced: http://www.kamuf.org/parri/graphs/TestThreeLowestTTLBalanced.png
Stamina: http://www.kamuf.org/parri/graphs/TestThreeLowestTTLStamina.png

And once again, the Warrior sits across the bottom with the DKs, the only thing sparing the Paladin from this fate is Ardent Defender. Even the minor Warrior/Paladin advantage at the very fast end of the attack speed range does not translate into any kind of advantage in terms of time to live.

Average Damage per hit: http://www.kamuf.org/parri/graphs/TestThreeAverageDamageHit.png
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 14. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:08:50 PM PDT
limit-reached limit-reached
Conclusion

As tanking stands, before any of the changes on the PTR, both Death Knights and Druids are operating at a significant advantage to Warriors and Paladins. However this is happening at different points. A Death Knight's time to live is not at all imbalanced compared to a Warrior. They are capable of dying just as fast. However, the likelyhood of this fast death is a lot less than a Warrior due to cooldown uptime and higher avoidance. Most importantly for Dks, their avoidance leads them to take a hugely significant amount less damage; as little as 50-60% of an equivalently geared tank of a different class. The average damage per hit for each of the classes, excluding Druid, is similar, so there is no reason for such a large avoidance gap. Already DKs in tier 7 are above the levels of avoidance of tanks in t6 entering Sunwell – a level that Blizzard felt was overpowered because it cornered them into an approach of 2 hits = a tank death. This is a problem that will only continue to grow as we get more mudflation. The DK avoidance advantage is primarily down to talents and weapon enchants offering avoidance above that of other classes that is not on the DR curve. Any good tank will appreciate that as our avoidance increases, 1% avoidance is worth an increasing amount. It's the reason that avoidance was put on DR in the first place. Death Knights will always be 10-15% avoidance ahead, but depending on the content level, it will mean different results. 10-15% avoidance when you start at 0% is 10-15% less damage taken. 10-15% avoidance when you're at 55% avoidance however is 20-30% less damage taken. As we get more avoidance from gear, the gap will only continue to grow. DK avoidance has to be reined in.

Druids are on a different side of the overpowered spectrum. They don't take significantly less damage than other tanks. But a Druid does have 1.5-2 times higher time to live against worst case scenarios. This is due to the combination of 2 effects. Druids cannot get an “unblocked” hit. Of course, Druids cannot block, but as long as their mitigation equivalent is represented solely as armour, it can be impossible to bypass. The Warrior worst case scenario is significantly more dangerous than the Druid one, because he can get hit with 11k less armour and without the block to cover it. In addition, the Druid has a significant HP buffer on the Warrior. Combined, these two factors allow Druids to survive unpredictable burst that would kill another tank. Often a tank is measured not on his ability to avoid damage, but to survive it. This is especially true in progression content. As long as Druids are operating far above the baseline set by other tanks, they will be prioritised in this regard.

The problem of Block

The tests also show a significant problem that exists for shielded tanks. The discrepancy between damage taken for tanks is not solely due to an unfair advantage held by Dks and Druids. It's also partially due to the inability for Block to keep up with the incoming damage. This is a common observation from all classes and certainly does not require such a large amount of tests to prove. However. The suggested “fixes” for this imbalance are little more than turning block into another percentage mitigation. Basically reforming block into a renamed armour. As much as I dislike how the term “homogenisation” is used to fearmonger the community into avoiding tank balance issues, this is one where the fear is true. Block cannot be converted into some kind of percentage damage mitigation or you entirely lose the uniqueness of Paladin and Warrior tanks. More than ever, we are defined by block. And I do not believe it is any coincidence that encounters that highlight shortfalls in Blocking also highlight deficiencies of Paladins and Warriors.

The issue is that Block must scale better. But that leaves us with a dilemma. If Block were to ever scale to reliably mitigate a large portion of a 20-30k post mitigation hit, then it would entirely trivialise lower end content to an extent that would be completely unfair to Druids and Dks. You cannot approach this problem from a single angle and hope to solve it. You need a multiple pronged approach that solves other pressing tanking issues at the same time. The following steps would, with some number tweaking possibly, allow Block to scale to a less penalised extent when dealing with more lethal damage intake. It would also slightly lower the effectiveness of Block tanks against waves of trash mobs the kind of which seen in Felmyst, Sartharion, etc.

1.)Remove the base 5% chance to be missed. Possibly remove/lower the 5% dodge talents on every tanking class.

[ Post edited by Xav ]

80
View All Posts by This User Toggle Ignore / Unignore This User
  • 15. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:10:11 PM PDT
limit-reached limit-reached
This is very important. Avoidance is already far too high. We're in tier 1 of WotLK and reaching avoidance comparable to the third tier of gear in TBC. But how does this effect block based tanks? You have to understand the importance of the hit table when you consider the advantages a block based tank gives you. Paladins are so superb on trash because they can already reach 102.4% combined avoidance+block. Basically, they will never take an unblocked hit. This is a problem waiting to happen. It's a bit similar to Druids capping out on armour in t4. From this tier onwards, the Block rating stat is completely worthless to a Paladin – which realistically will not ever be reflected in itemisation. By lowering avoidance as a whole, you put this cap 5-10% higher than it is now. It's harder to reach, so allows room for further scaling. It also means that Warriors and Paladins are more likely to take those unblocked hits. Trash and trash tanking becomes a bit more equalised.

2.)Remove the Block Value stat. Rework Strength to give a 1:1 STR:Block Value ratio.

Two stats that very maginally increase the effectiveness of one ability is an inelegant and kludgy solution. It has always traditionally led to confusion for new players of the class. It also creates a gulf of potential block value between items that are highly itemised for it and those that are not, which is a balance problem later. Rein in the potential variability of Block Value for each tier so we are able to apply multiplier effects to it later without reaching extreme case problems.

3.)Slightly rework the Block Rating stat. It should provide a lower than current % chance to block, but also increase your Blocked value by a %.

The avoidance change earlier is not enough. Block rating is so cheap for each % removal of the hit table that it would only take to Ulduar BiS gear to run into the same problem again. Putting Block rating on DR however would be far too harsh, because it is not avoidance and provides nothing close to the same benefit. Blocking should be something a Warrior/Paladin can do very reliably. It should happen most of the time. It should however come with a small chance to not happen, and when it doesn't happen, the Warrior/Paladin should take more damage that hit than other tanks would have taken.

4.)Rework Holy Shield/Redoubt slightly.

These abilities/passive talents bring both classes too close to the Block cap. Possibly reduce Holy Shield to 20% bonus chance to block, but in return allow it to provide a 20% bonus to the amount of damage blocked. Remove Redoubt effect entirely, as stacking between these two leads to the same problems again. Similarly with Shield Specialisation, slightly reduce the chance to block, but increase the effectiveness when a Block does occur.

5.)Give Critical Block to Paladins + Warriors baseline. Base it off a stat that will only realistically increase in a fully raid buffed setting.

IE, Crit %. Blocking 1500 of a 29k hit is negligable. So would blocking 2.2 or whatever with minor % modifiers. But if you could sometimes block for 3k-4.4k, that's a lot more significant. It's still a linear reduction in boss damage, but the return is a lot closer to the amount armour would have reduced by. Basically, each tanking class should have the chance to critically block damage based on their crit %. The current Critical Block talent for a Warrior would further increase this chance by another 30%. This furthers the current distinction between Warriors and Paladins: Warriors will more often block for a higher amount, but Paladins have the greater chance to block in the first place.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 16. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:11:12 PM PDT
limit-reached limit-reached
So I hope it's clear that these changes would allow Block to have a more significant impact when the damage range of the boss reaches over 20k post mitigation. But it is perhaps not as clear for trash/add-wave scenarios. Consider the following:

Mob hits Paladin for 0 (1.5k blocked)
Mob hits Paladin for 0 (1.5k blocked)
Mob hits Paladin for 0 (1.5k blocked)
Mob hits Paladin for 0 (1.5k blocked)
Mob hits Paladin for 0 (1.5k blocked)
Mob hits Paladin for 0 (1.5k blocked)

That's what we have now. Trash is already trivialised by blocking. The damage a wave of mobs can do is balanced assuming a tank that cannot reduce each hit to 0 damage. So it will never be significantly higher than the Paladin's block value. What the changes I proposed would do is change that previous combat log into something like this:

Mob hits Paladin for 0 (1500 blocked) (500 overblock)
Mob hits Paladin for 0 (1500 blocked) (2500 overblock)
Mob hits Paladin for 1500
Mob hits Paladin for 1500
Mob hits Paladin for 0 (1.5k blocked) (2500 overblock)

This version of Block leaves more room for scaling, it more effective than the current iteration against large boss damage, and instead of making Block tanks even more required on trash, actually brings them closer to the others.

Block can be made stronger without making it even more overpowered against trash. Block can be made more significant without homogenising and turning it into a renamed Armour. This change will not make Block scale against incoming damage, but if you can imagine trash damage and boss damage on a scale from left to right, Block is sitting too far on the left now. Guaranteed and minor damage reduction. I propose we just shift it to the right a little. A little less likely to happen, but larger reductions.

Anyway, I think I'm about done. Going to take a drink. Thanks for your time if you've read this far.

P.S. Ghostcrawler, if you're reading this. Please. Please. Fix the broken Block items in the Naxx loot table. Specifically Wall of Terror from KT, the belt from Patchwerk Heroic and the BoE Block Value boots from Heroic Naxx. They're itemised using pre-BC stamina budget.

Oh, and finished now. Flame away.


---------
End of his post

Anyone who read it will realize some of the suggestions he made aren't even needed anymore because of stuff they changed, but there you have it.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 17. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:11:13 PM PDT
limit-reached limit-reached
I've found 2 serious problems so far.

1. No gimmick fights rule. Every boss is a gimmick fight nowadays, since any boss without a gimmick is just a simple tank&spank.

2. No PvP gear rule is unfair to Druids, since some PvP items are arguably best-in-slot pre 3.1.
80
View All Posts by This User Toggle Ignore / Unignore This User
  • 18. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:13:05 PM PDT
limit-reached limit-reached
Was waiting to mention these must be pre current 3.1 PTR iteration tests, but you mentioned that at the end. Very interesting read.

"Sorry moron, but I can't out-tank stupid"
80
View All Posts by This User Toggle Ignore / Unignore This User
  • Black Dragonflight
  • 19. Re: Pre 3.1 Exhaustive Tank comparisons   03/08/2009 03:14:14 PM PDT
limit-reached limit-reached
Goodness gracious this is a long read. Going to take me a bit to finish reading it.
1 . 2 . 3 . 4 . 5 . 6 . 7 . 8 . 9 . 10 . 11 . 12 . 13 . 14 . 15 . 16 . 17 . 18 . 19 . 20 . 21 . 22 . 23 . 24 . 25 . 26
Forum Nav : Jump To This Forum
Blizzard Entertainment