Since everyone seems to have forgotten, why SHARDING is on the table: by Ion Hazzikostas

[Fesz-cenarius]

First, define “starting areas” and “few weeks” and tell us all how you define those terms.

Starting Area - The first zone of all major factions I.E. Elwynn Forest and Durotar

Few Weeks - Few is generally accepted to be a unit of 3 or more, mostly leaning towards 3.

Second, if they even once use sharding outside of those areas, will you be here making a public apology to those that have raised concerns that Blizzard will not be able to resistbusing sharding beyond those “starting areas for a brief time at launch”?

In the event they use Sharding outside of the Starting Zones, I can’t imagine it would even be for any amount of time beyond the intial “few weeks”

And sure, should they implement it outside of the “Starting Areas” I will apologize on this forum, tagging you in it.

I appreciate that you definitely defined what you consider “starting areas”. I also appreciate that you define “starting areas” as the 1-10 (or 12) zones.

I would not call your definition of “few weeks” anything remotely close to definitive. IMO, definitive would have, for example, “three weeks”. “3 or more” could easily include 12 or even more. IMO, three months of sharding is simply unacceptable.

IMO, sharding used beyond the 1-10 (12) zones and beyond the first two weeks maximum is unacceptable.

As for sharding being used beyond those “starting areas” two big potential, even probable, ones come immediately to mind. Those would be the AQ event and the frequent Southshore/Tarren Mill battles.

[Laeric-wyrmrest-accord]

Finally, someone else with proper priorities!

[Galdor-maelstrom (Galdor)]

Why should I do that?

[Katerina-mannoroth]

So far, Blizzard has refused to definitively define any geographical or time limits for sharding. Why not?

It couldn’t possibly be because they want us to buy into the “just the tip” lie and then “Oops.”, could it?

Because they’re leaving an out for “Unknown unknowns” and Classic’s Launch is certainly one that could fit Rumsfield’s notorious maxim about “There are the things we know we know. There are the things we know we don’t know. And then there are the things we don’t know we don’t know.”

1. They know Launch starter areas are very likely to be a mess without respect to what they do.
2. They don’t know what kind of response they’re going to get from the fans on launch day, or what kind of scale they need to prepare for.
3. Because the scale of potential responses from the fans differ by potentially several orders of magnitude. They don’t know how long it is likely going to take in order to “scale up” to whatever we, the fans, throw at their servers.

Scaling from 50K to 100K concurrent per region is one thing. Scaling from 50K to 500K to 1M Concurrency numbers, per region, is an entirely different thing. And they won’t know what kind of problem that kind of rapid scale up would be like until they encounter it.

[Katerina-mannoroth]

What worries me most about sharding the launch is that 6 months later the devs say: “cases we hadn’t anticipated result in sharding to occur in some situations. We are looking to address this, but it turns out to be somewhat tricky.”

I am not predicting that though, just a worry.

This is why you beta test.

[Matcauthon-steamwheedle-cartel (Matcauthon)]

So far, Blizzard has refused to definitively define any geographical or time limits for sharding. Why not?

It couldn’t possibly be because they want us to buy into the “just the tip” lie and then “Oops.”, could it?

Because they’re leaving an out for “Unknown unknowns” and Classic’s Launch is certainly one that could fit Rumsfield’s notorious maxim about “There are the things we know we know. There are the things we know we don’t know. And then there are the things we don’t know we don’t know.”

1. They know Launch starter areas are very likely to be a mess without respect to what they do.
2. They don’t know what kind of response they’re going to get from the fans on launch day, or what kind of scale they need to prepare for.
3. Because the scale of potential responses from the fans differ by potentially several orders of magnitude. They don’t know how long it is likely going to take in order to “scale up” to whatever we, the fans, throw at their servers.

Scaling from 50K to 100K concurrent per region is one thing. Scaling from 50K to 500K to 1M Concurrency numbers, per region, is an entirely different thing. And they won’t know what kind of problem that kind of rapid scale up would be like until they encounter it.

when it comes to classic the correct amount of sharding is MINIMAL.

[Katerina-mannoroth]

when it comes to classic the correct amount of sharding is MINIMAL.

That’s all well and dandy. I’m sure that’s their goal. But Goals and Objectives are one thing, implementation often turns out to be another matter entirely.

They might need 3 weeks to get enough additional servers online to handle the load players are placing on them without sharding. (Assuming players transfer)

It’s also possible they could run into a paradox not much unlike one urban planners often encounter when they add new capacity to a public roadway. In that the moment they add capacity, traffic “almost immediately” increases to take advantage of the new capacity. Which puts them right back in the situation of needing another couple of weeks to increase capacity yet again, so they rinse and repeat the cycle yet again.

[Matcauthon-steamwheedle-cartel (Matcauthon)]

when it comes to classic the correct amount of sharding is MINIMAL.

That’s all well and dandy. I’m sure that’s their goal. But Goals and Objectives are one thing, implementation often turns out to be another matter entirely.

They might need 3 weeks to get enough additional servers online to handle the load players are placing on them without sharding. (Assuming players transfer)

It’s also possible they could run into a paradox not much unlike one urban planners often encounter when they add new capacity to a public roadway. In that the moment they add capacity, traffic “almost immediately” increases to take advantage of the new capacity. Which puts them right back in the situation of needing another couple of weeks to increase capacity yet again, so they rinse and repeat the cycle yet again.

that sounds like a failure to plan on their part and does not constitute an emergency on my part.

[Fesz-cenarius]

when it comes to classic the correct amount of sharding is MINIMAL.

That’s all well and dandy. I’m sure that’s their goal. But Goals and Objectives are one thing, implementation often turns out to be another matter entirely.

They might need 3 weeks to get enough additional servers online to handle the load players are placing on them without sharding. (Assuming players transfer)

It’s also possible they could run into a paradox not much unlike one urban planners often encounter when they add new capacity to a public roadway. In that the moment they add capacity, traffic “almost immediately” increases to take advantage of the new capacity. Which puts them right back in the situation of needing another couple of weeks to increase capacity yet again, so they rinse and repeat the cycle yet again.

that sounds like a failure to plan on their part and does not constitute an emergency on my part.

That is very true.

I’m probably oversimplifying, but if they are using cloud servers and not physical servers, could they not have additional servers prepped and ready to be made live should the demandcexceed their estimates?

Hypothetically:

Blizzard estimates 500K people initially. They plan to cap the servers at 10K.

They have 50 servers for launch.

Would it not be possible to have another 50 servers prepped and ready to go live, should the demand exceed that 500K estimate? Would it not also be possible to have arrangements made for even more servers to be readily available if the demand exceed even one million players?

It’s just a thought, but it seems to me that would be far better than scrambling to acquire additional servers once classic has gone live and the demand has exceeded available capacity.

[Katerina-mannoroth]

that sounds like a failure to plan on their part and does not constitute an emergency on my part.

Because having “metrics” suggesting you need to plan for 50K and after deploying for 65K, you end up with 100K instead is a failure to plan.

[Fesz-cenarius]

that sounds like a failure to plan on their part and does not constitute an emergency on my part.

Because having “metrics” suggesting you need to plan for 50K and after deploying for 65K, you end up with 100K instead is a failure to plan.

IMO, that would be a failure to plan sufficiently.

[Katerina-mannoroth]

I’m probably oversimplifying, but if they are using cloud servers and not physical servers, could they not have additional servers prepped and ready to be made live should the demandcexceed their estimates?

Something tells me their “cloud services” aren’t quite the average bear. It isn’t like provisioning additional servers from Amazon Cloud is going to help improve the gameplay experience.

There are limits to just how much additional (permanent) capacity they’re going to have available to immediately provision into at any given time. (Sharding into idle (BFA) capacity not withstanding)

Data Centers don’t just leave legions of idle systems lying around consuming power and accumulating wear and tear while nobody is paying them to operate it. Blizzard’s contingency planning for Classic’s release would probably be a fascinating study in numerous things, but it still cycles back to “nobody knows what to truly expect” and the issue of “orders of magnitude” is a very significant bug in the ointment.

[Matcauthon-steamwheedle-cartel (Matcauthon)]

that sounds like a failure to plan on their part and does not constitute an emergency on my part.

Because having “metrics” suggesting you need to plan for 50K and after deploying for 65K, you end up with 100K instead is a failure to plan.

good thing they have sharding then huh? gives them that two week window to get more servers up.

[Katerina-mannoroth]

Because having “metrics” suggesting you need to plan for 50K and after deploying for 65K, you end up with 100K instead is a failure to plan.

IMO, that would be a failure to plan sufficiently.

Its a failure of the metric used, not a failure of planning. The problem is, much like with urban planners and highways, there are no reliable metrics for “pent up demand” on which to make plans with.

But one thing that would help them gauge pent up demand for Classic once it has launched? Sharding is going to accomplish that in ways that login queues alone never could.

[Fesz-cenarius]

I’m probably oversimplifying, but if they are using cloud servers and not physical servers, could they not have additional servers prepped and ready to be made live should the demandcexceed their estimates?

Something tells me their “cloud services” aren’t quite the average bear. It isn’t like provisioning additional servers from Amazon Cloud is going to help improve the gameplay experience.

There are limits to just how much additional (permanent) capacity they’re going to have available to immediately provision into at any given time. (Sharding into idle (BFA) capacity not withstanding)

Data Centers don’t just leave legions of idle systems lying around consuming power and accumulating wear and tear while nobody is paying them to operate it. Blizzard’s contingency planning for Classic’s release would probably be a fascinating study in numerous things, but it still cycles back to “nobody knows what to truly expect” and the issue of “orders of magnitude” is a very significant bug in the ointment.

Well, Blizzard has about 6 months to make those necessary arrangements.

It seems to me that that additional capacity does not have to be permanent at launch. It seems to me that the additional capacity could be initially set up as short term (in case demand does not exceed estimates) with arrangements in place to make that additional capacity permanent if demand exceeds estimates.

It seems to me that, additionally, if that additional capacity is needed, subsequent similar arrangements could be made for even more additional capacity should the initial additional capacity not prove to be sufficient.

[Enkki-eonar (Enkki)]

Why does everyone focus on this single fraction of a sentence like it’s the entire reason for sharding being mentioned as a solution ?

Because we don’t trust Blizzard to do what it says. Too many times we’ve been hit with “Nothing is written in stone; everything is subject to change.” So that “limited” scope and application can EASILY become the prevalent feature of the game, realm, zone, locale. Temporary can become permanent with the snap of Blizzard’s fingers.

It’s happened too many times in the past, and there are those of us still around who remember it. Blizzard CANNOT be trusted to do what it says or not do what it says it won’t do. And after it’s done, it’s too late to protest.

[Puddintusk-area-52]

Wild how everyone becomes armchair developers when this topic comes up.

Personally if the people who have been in this business for over a decade and have access to all of the numbers and information say they need sharding to make things work then I’ll accept that.

I highly doubt they will use sharding more than absolutely necessary if only because of the absolute sh*storm it will cause.

[Zaldera-moon-guard]

Sharding was not part of the experience of Vanilla WoW. It will not be a classic experience with sharding. That’s all there is to it.

Classic would be leaving all realms closed to their own individual populations.

We’ve forgotten why sharding is “on the table”? Blizz has forgotten what constitutes a classic WoW experience.

[Puddintusk-area-52]

I’d rather have sharding a few weeks than spend six hours trying to log in.

Sorry, not everyone has the option of taking time off of work to play video games on launch day.

Some of us are adults with full time jobs and limited vacation days who only get to play video games when we get home from work, and dont want to spend that entire time in a loading screen being told we are number ninety nine thousand in queue.

Not trying to imply that you aren’t an adult with a job btw, just making a statement on what exactly is useful about sharding.

I fully agree that it shouldn’t overstay its welcome. Only saying that you will get much more enjoyment out of actually playing the game than you will out of sitting in front of a login screen.

[Fateweaver-misha (Fateweaver)]

I’d rather have sharding a few weeks than spend six hours trying to log in.

Here is this false dichotomy again.