It depends on number, speed, and size of attacks incoming.

Few large attacks (Ultraxion/Zon'ozz/Blackhorn/Spine Amalgs/Madness) you want stam.
Many small attacks (Yor'sahj/Boat adds/Spine blood) you have more freedom to choose avoidance.

Essentially, if you can't survive 3-5 regular blocks in a row you need more stam; if you can survive 20+ regular blocks in a row avoidance/mastery is superior.

It's all part of a Burst Time analysis. Since the hardest bosses this tier have large, slow swings (upwards of 200k before mitigation) and/or huge magic burst damage, stamina is better.

EDIT: Analyzing your chart, I notice that all the options have more than 200% "amount of healthpool received." I think that's where you're tripping up. When you have received 100% of your HP, you're dead. Everything after that point is irrelevant. The stam builds live about 1-2 swings longer under heavy damage situations than an avoidance build because they have a larger pool.

By comparison, avoidance builds offer the best damage reduction, assuming that your healers can keep you up. The problem with such builds is that bad luck still happens and tanks fall over after a string of blocks.

The fact that you don't fear dying says a lot. 1 in 24 swings (1-2 pulls) you take a ~210k damage spike over 4 seconds and 1 in 64 swings (3-4 pulls) you take a ~280k damage spike over 6 seconds.

What you seem to be saying is that you scare your healers on a regular basis. Your healers have to be constantly spamming you because you want to take less damage over the long term. Personally, I prefer the stamina route.

Untrue.

In a raid scenario, what kills tanks is burst damage. An unreliable heal for 3% of your health (maybe slightly more than 3% if Field Dressing affects the proc, which I assume it does) can't be counted on for Devastate+melee+shout combos on Blackhorn, for instance.

As a tank, you have three priorities in a raid.

Priority one: you must be able to hold aggro. With the current amount of threat that the tank classes are able to generate, this step is trivial unless your DPS overgear you by an insane degree, and pretty much every tank out there can safely ignore it. In the past, there have been encounters where threat was somewhat more important due to various fight mechanics, although that hasn't really been the case for any encounters this expansion, with the possible exception of Domo before they nerfed DPS aggro on that fight.

Priority two: you must be able to survive the worst-case scenario. Stamina, armor, resistances, and cooldowns all figure into this. Damage absorptions effects (Power Word: Shield) also grant temporary benefits here. Once you're able to consistently live through the burst damage that bosses will throw at you, then you have enough EH (effective health) and you can move on to step three. How much EH you need will vary depending on the content. For instance, during tier 10, bosses hit so hard that you never had enough EH, and as a result, tanks were never able to move past this step.

Priority three: you must mitigate or avoid enough damage that your healers are able to keep you up throughout the fight. Basically anything that reduces your overall damage taken throughout a fight will figure into this. Avoidance, armor, resistances, absorbs, and self-healing all play a part here. How important this step is will vary depending on healer mana regeneration and healing output. For instance, if a healer can top you off from 1 health to full every second without running out of mana, this step would be largely trivialized, because there wouldn't be anything you could do to affect what they did. In ICC, this was what happened, more or less, because healers never ran out of mana and healed for most of your health bar in a single cast.

After you've fulfilled those three requirements, you can choose to focus on DPS stats (hit/expertise) in order to help your raid push through various DPS checks more easily, because your survivability isn't an issue at that point. This is what you see a lot of tanks do on Ultraxion, because you're never really at risk of dying unless you're not managing your cooldowns properly and the enrage is an issue (or at least it was pre-nerf).

There is no possible situation where step 3 comes before step 2. It doesn't matter how much self-healing you have if you won't survive Impale, or Psychic Drain and a melee swing at the same time.

In Cataclysm, healer mana regeneration has scaled up from tier to tier so much that we're seeing a similar situation, especially in 25-mans where there's wider access to mana regeneration cooldowns like Mana Tide Totem. The only situations where healers are ever hurting for mana is when you're under-healing encounters (guilds are using three or four healers on encounters designed for six, and they're STILL managing it), and even then, it's the raid healing that will run them dry, rather than the tank healing.

This tier, the only tier where I'd probably use Souldrinker over Hand of Morchok for the survivability benefits is Yor'sajh, since the EH requirements for the fight are actually relatively low, while the purple debuff makes self-healing more valuable. There are other fights where I'd definitely use it, but that's for the substantial damage boost rather than the healing.

Stamina is generally a much better choice after CTC, due to the fact that basically all of the burst damage that you'll encounter in Dragon Soul is unblockable. For example, look at a fight like heroic Hagara, where you need to be able to survive Focused Assault. For that, the only stats that will help you are stamina and armor, and you can't stack armor. Other examples include Zon'ozz (Psychic Drain), Madness (Impale), or Morchok (Stomp).

EDIT: Read the "tank steps" in the post right before this one for a look at the reasons why.

@Muspel:

Drain Life ticks for 9.6k on my char, pretty significant heal I think. It scales well with gear and cooldownsl. It is a DPS increase. Most importantly, it no brainer improves your bust time to a degree that ~200-250 stam simply can not compare on most content. I bet its better for TtL too.

So I've seemed to have noticed most prot warriors are stabilizing their health pools unbuffed around the 220-230 range, not really going too much higher than that, is this a result of the nerfs or is there a sweet spot that tanks should be aiming for anyway?

I'm also noticing a return back to pure Mastery gems, is that b/c with IP and SSV coupled with the EH from the 410 gear, your HP is large enough?

Edit: When i say a return to mastery gems I'm referring to the switch back to yellow mastery gems from the green stam+mastery cut, not replacing any pure stam gems.

Morchok is admittedly trivial in terms of EH requirements after the nerfs, but for guilds that are still progressing on him, it's likely that they don't have the best healers and may need the extra cushion. I was just throwing it out there as an example of a fight with burst damage, not as a major consideration. Had this been T12, I would have mentioned pre-nerf Shannox, as his melee+arcing slash combo was actually surprisingly painful for such an easy boss, especially if it lined up with a bleed tick.

On heroic 25-man, Impale hits for 1.2 million damage. Or it did, before the nerfs, it's somewhat lower now. For the impales where it lines up with a bolt impact, you have a particularly high chance of dying.

You reforge for the hit and expertise caps and wear DPS trinkets. You can even wear some pure DPS gear if you want to and/or use certain DPS enchants, like Landslide. The 100% uptime on Shield Block due to the buff he gives you means that block capping is trivial, and unless they fixed it, SB will actually stack with itself every time you refresh it, meaning you reach 100% critical block chance by the end of the fight.

...I assume you mean effective health when you say "bust time". And 9.6k is going to be less than 5% of your health, which I'd argue is the exact definition of insignificant. Nor does it scale particularly well. Going from maybe 150k health in 346/359 gear to 230k in 397/410 gear will increase the size of the proc from ~6.4k to ~9.6k. Now, percentage wise, that's a 50% increase, but in the grand scheme of things... 3k more healing when bosses are hitting more than twice as hard is a drop in the bucket.

If so, then you don't understand what EH is. There is no possible situation where a heal can increase your EH, random or not. It's literally outside the definition. It's like arguing that Shield Wall lets you get out of fire faster (it might let you survive more ticks, which makes you less likely to die, but that's an entirely different scenario). Your effective health is simply a measurement of how much raw damage you can take before you fall over dead, with no healing included.

And like I said earlier, TTL/overall mitigation is never more important than reaching the EH threshold for an encounter. Let me illustrate this with an extreme example. Imagine that you have a boss that swings once every minute for 300k damage, and you currently have 250k health. Depending on how you gear, you can either reach the point where you heal 300k damage over the course of every minute, or where you have 305k health.

In that situation, EH is always better. It doesn't matter if you never require any healing if you don't survive in order to get that healing in the first place. Granted, most encounters are far less cut-and-dry, but you see my point.

The only reason to gear for TTL is if you don't die to burst damage and there's some kind of limitation on how much you can be healed (healer mana or the Yor'sajh debuff, for instance). Blizzard's initial design theory for Cataclysm was "healer mana will matter, and tanks will have so much health that bosses will gradually whittle them down while the healers try to top them back off efficiently rather than getting bursted down", but they failed to deliver on that pretty spectacularly. In that environment, where EH isn't an issue and conservation matters, tanks would probably end up actually gearing for TTL. As it is, it's just not worth it because there aren't any meaningful benefits to getting a higher TTL when all that kills you is burst damage.

Now, to reiterate: I actually do recommend using Souldrinker on most fights. The DPS increase comes at a tiny EH cost, and that increase is far more likely to help you score a kill than the bonus stamina from HoM. But if you're running into situations where you're dying to burst damage, then Hand of Morchok is a better choice to help you survive. Period.

First, EH should not be used to describe tanks survivability. Its a meaningless stat without considering other factors, ie swing size or timer.

Second, BT model is based on probabilities. Souldrinker's proc does not need to be 100% to be useful. It still contributes, no matter the proc. BT windows on the majority of bosses are large enough to account for more than one proc on average.

Third, the notion that "5% is the definition of insignificant" is loltastic. I suggest you start 24-manning HM progression or switch off the latest DS nerf to appreciate the difference. Not to mention that this "5%" routinely comes out top 3 heal with <30% overheal on some encounters. Which is sorta handy for any tank, let alone a warrior.

Last, I have no idea what "EH-threshold" has to do with this discussion. It's not like you desperately need Hand of Morchok to reach it and switching to Souldrinker gonna suddenly make you a one shot wonder on any bosses.

ps Armsaun, BT does NOT describe a worst case scenario. It describes the probability of a worst case scenario occurring during a course of a fight. Souldrinker helps to bring this probability down.

I think the notion that healer mana is completely trivial doesn't necessarily hold true for the extended fights at the end of the tier. Having seen healers running on fumes for Spine trying to keep pace with searing plasma during progression, I'm inclined to agree with kopcap here on the SD issue(I assume a similar healing stress for madness as well, though I was DPS on our guild's second kill and wasn't as aware of mana bars).

I don't believe anyone here is disputing EH > mitigation in how we gem and reforge for the majority of the tier, but the basis for this is largely because 1. Most of what is coming at us in this tier is either magic or unblockable/parryable/dodgeable damage and 2. Healer mana pools are quite generous right now. It was even said earlier in the thread that Blackhorn does provide incentives to gem more for mitigation due to the large portion of incoming damage melee/devastate is. If we accept that there are some fights this tier where healer mana is a concern, what we have now is a unique way for plate tanks to extend burst time regardless of damage type, provided that our EH is at a point at which we can survive encounter-specific bursts. From a holistic hand-wavy point of view, I would contend that that point is much easier to hit than we think(tank damage mitigation does not exist in a vacuum free of healer CDs), and as such would merit giving more weight to helping healers over the long term of the fight.

If the stamina contribution of H Hand was more significant or we need more physical mitigation stats for Blackhorn I could see a case for the mace, but I think the EH argument has been taken a bit too far in extreme in this particular case.

My mistake, I was under the impression that Burst Time analysis was a worst-case scenario. It specifically measures 1) the average number of swings until tank death and 2) the amount of time that will take, extrapolated from the boss swing timer.

We can alter the formula to account for crit blocks as well as regular blocks, just like we can put in a term for SD procs.

I'd also like to talk about the tanking priorities.
Muspel stated very clearly the 3 accepted priorities: 1) hold threat, 2) survive worst-case scenarios, 3) mitigate as much damage as possible.

I agree that threat is a non-issue in this tier. I also agree that mitigation should take a backseat to EH when tanks can't withstand predictable burst. I'm not certain that I agree that more EH is needed to survive worst-case scenarios. With so many CDs available, we can take many worst-case scenarios with ease. As one example, H Hagara's Unrelenting Assault is easily mitigated with a Shield Wall or healer CD, especially considering that healers expect to spam the tank during that time.

I am almost willing to say that we don't need more than 225k health for any fights. The added HP pools definitely provide a buffer, but at 225k a warrior can survive 3 regular blocks from most bosses and during other times, CDs are expected or required. Should we be spending all of our free itemization points on getting HP above 250k or should we make sure our healers can keep going at the end of those long encounters?

I'm not trying to say that the priority is wrong, just that we may want to revisit our assumptions about how much EH is needed.

EH needed always depends on content you are doing. If you feel you have too much health, you probably do.

If you are GCD'ed while being topped up - by all means use Hand of Morhock. Or a cooldown.

Its an old article, written before even DKs were introduced. Tanking models have vastly changed since then. Heals, the articles relates to, are direct, landed, external spells cast reactively for purposes explained within the article. Things like HoTs are passives are omitted for simplicity sake, its silly to argue that they don't contribute to tanks survivability or BT. There is nothing that stops you from accounting for Souldrinker procs if you want a more accurate model, you just need to expand it a little.

Especially when its the last 5% of your health.