Tobold's Blog
Thursday, August 29, 2019
An unsolved problem

If you bought a mobile phone in 2004, chances are that is was a Nokia with a shape and functionality that looks primitive in 2019. The first iPhone wasn't even released yet in 2004. 15 years is a long time in technology. Yet if we look at World of Warcraft Classic in 2019, it is still plagued with the same technical problems as original WoW in 2004: When a lot of people want to play, servers get overloaded, disconnects happen, people have to wait in queues. But adding more servers isn't a solution either, because inevitably some people who want to have a look at WoW Classic in August 2019 will be gone by Christmas. The fundamental architecture of MMORPGs being played on named servers just doesn't scale very well, neither up nor down.

This isn't to bash WoW Classic. I think that WoW Classic is a good idea, even if I am personally not interested in playing World of Warcraft again, vanilla or any other taste. I think that over the years I mentioned several times on my blog that while free-to-play MMORPGs count everybody who ever made an account in their "player numbers", subscription-based games only count people with an active subscription. There are far more ex-WoW players than current WoW players. The potential pool of customers for WoW Classic is huge, and many of them feel some sort of nostalgia for the early days of the game. Even if the average player just subscribes for 3 months before leaving again, that is a huge pile of money for Blizzard.

However one has to think that vanilla WoW still was the growth phase of World of Warcraft. There were more new players coming in every month than there were players leaving. It is unlikely that this will reproduce with WoW Classic. It is far more likely that in three months the population on the WoW Classic servers will start to shrink. And that brings us back to the server scalability.

It has to be pointed out that this problem is particular to open world MMORPGs. Multi-player games like World of Tanks, League of Legends, or Fortnite don't have that problem, because the actual gameplay happens on small maps with a limited amount of people for a limited amount of time. Servers just need to be able to create a sufficient number of instances, and that scales up or down rather well. It is MMORPGs in which the first thing you do when creating a character is to select a named server that don't scale well.

I don't think the problem is unsolvable. There must be some sort of solution that combines the advantages of being able to meet your friends in game with the flexibility of instanced games. But WoW Classic certainly doesn't offer a solution at this point in time, and the negative consequences of that are foreseeable.

"Yet if we look at World of Warcraft Classic in 2019, it is still plagued with the same technical problems as original WoW in 2004: When a lot of people want to play, servers get overloaded, disconnects happen, people have to wait in queues."

I would agree with everything in this post with the exception of the above quoted text. Replace "technical problems" with "human induced technical limitations" and you have a solution, albeit not a "technical" one. Blizzard woefully failed to judge the demand for Classic. On name reservation day(Aug 12th), Blizzard only offered 11 US Realms and 2 Oceanic Realms for a total of 13 Realms.

And with over 2 million people(estimated) creating characters over the two weeks leading up to Launch day, Blizzard failed to react appropriately - resulting in the queues and server loads that have occurred. Now, just 3 days after launch, Blizzard finds itself with 30 US Realms and 4 Oceanic Realms, for a total of 34 Realms, which is almost three times more than they originally planned to launch with.

Reminds me of the old saying: "A lack of planning on your part does not constitute an emergency on mine."
Didn't the original Guild Wars (2005) have an elegant scalable solution to this. They used numbered instances for their shared city zones rather than fixed realms. This meant they could easily add additional instances at peak times presumably by hiring server space from AWS or similar. Players were automatically assigned to the next instance with capacity at log on but they could easily switch instances so it was easy enough to meet up with friends for quests.

Could that work for a game like WOW?
Cryptic games (Star Trek Online, Neverwinter) and ESO have both solved this. There's One True Server (maybe two, one EU one NA). The server has any number of instances, spawned on demand. If you want to group up with a friend, you group up and the game moves you to the same instance.

I think Cryptic did it better since it uses your account name as the unique identifier, meaning name collisions don't matter. It also lets you explicitly switch instances (if you like it emptier or fuller, for example). But both of them are almost-infinitely load flexible from the server's perspective, without making the users need to care about What Instance Should I Play On?
Named servers have their issues, but I think they are actually *necessary* if you want to make an actual RPG. (Unless total population is very low, that works fine too but is not conducive to profit.)

The thing is, being able to join your friends doesn't make a game an RPG. For an RPG, you also have to be stuck with your enemies, your friends and your not-yet friends. If you're not making friends in the game world, it's not an MMORPG.

In classic WoW you had a name and reputation that followed you. Your character had an actual existence in the world. We've lost that.

I think they should simply double up named servers e.g. make a Hades2 if Hades is too full, with the avowed intention of compressing them together again when the initial rush dies off. But in any case, server migration should be less annoying to players if it happens relatively soon in the game's life. Classic players will understand the need.

It's not a technical issue at all. I think Blizzard can stuff a lot of people into the servers - but they don't want to.

What they want is to have a stable server population after the tourists leave, at about the same size as a vanilla server. I think they estimated the number of long term players, and set the number of servers at what they think is a good number. (Too many servers and each server is a ghost town. Too few and they have to keep the layers up indefinitely.)

I also think that the queues as they are are not a technical issue with the login servers, but again a balance problem with blizzard. I don't think they want too many people running around creating resources on infinite layers on one realm. Either that, or they are still confident in the retention rate, even if they failed to estimate the number of players, so they're now aiming for more servers in the long term and want the players to spread out.

Which they don't, there are plenty of servers with no queues at all, but players stubbornly stick to the one where they have created characters. They don't want to lose their reserved name or guild, and more understandably they don't want to start over from level 1.

This is not a technical issue at all. It's a game balance and sociological problem.
MBP, there's a radically different feel between WoW (Classic) and GW. The latter really didn't have an open world where you met other players, just those numbered city zones that acted more as a lobby. The questing areas were all instanced to your group.

I can't really speak to the technical differences other than to acknowledge they exist, but from a player perspective these experiences are nothing alike.
IMO, it is a solved technical problem; the unsolvable problem is Luddite players. MegaServer tech like SWTOR, ESO, GW2 could have ten million players on one realm. As more people log in, they bring more physical computer servers (real or virtual) online and run multiple perhaps many instances. At 4AM 9 months after launch, those ten million accounts might need one instance for Wetlands/Hoth. But people who "value their reputation" are grumpy at megaservers.
Well Hagu I think the reputation guys have a point. It was a lot of fun running into that horde guy who ran his mouth on the realm forums and then smiting his ass, or knowing you're the 3rd best geared tank on the serve, or being in a guild that people respect or beefing with a horde guild or what have you. A ten million person server has none of that. The competition was something. It's not a fear of technology as much as objecting to the removal of sense of community.
This is like people in 3000 AD swallowing tasteless nutrient pills and laughing at Luddites who eat traditional food.
It's an easily solvable (Now, not when WoW was originally introduced.) problem from a technical aspect.

The problem now is so many people have a view that the server is some kind of community that must be preserved so you're always in with the same players.

Well, that doesn't work. Neither does treating RPGs like fictional stories where the hero miraculously always lives. In Star Wars, Hans Solo was right to laugh at C3PO's concern that the odds of successfully navigating an asteroid field are 3720 to 1. Hans might as well have looked at the camera and said "For them, sure. I'm the hero, I can't be killed here."

There IS a solution to those faceless 10 million person servers. Indy games. You want AAA graphics and 'quality'? You get 10 million person servers.
Server size has no relation to graphics quality. It's a design choice.
It seems like several people haven't quite followed what the company is doing.

They've invented a new technology called layering which breaks up each server into several separate instances. So when I made my character there were only subjectively about 100 people in the starter zone even though there's thousands on the server. It actually feels about right - busy but not overcrowded.

At some point they will collapse these layers so everyone will subjectively be in the same server.

There are many reasons to ctiticise Wow Classic but this isn't a valid one. Of course they knew there would be a "Wow tourist" effect. Who better to know.

The issue won't be technical, the issue might be that people rapidly lose interest. But even there Wow Classic might confound expectations. It feels like it's in a growth phase with people building social structures, reaching for friends they can persuade to try it, etc. I've not seen that in a game for a long time.
@Stabs: So if WoW has this new tech that solves the problem, why are there server queues? It seems to me at best a partial solution.

Blizzard enacted an update in which they were able to increase the number of people on servers without much issue. Couple this with the Layering tech that Stabs mentioned(has nothing to do with queue's though), and the server queues have been greatly lessened the past few nights. Not non-existent mind you, but nothing like it was.
I'm not having any of this. It's like WoW has only been revived for the sport of destroying it all over again.

It didn't needed fixing the first time and it certainly doesn't need it while we still have the 'fixed' one.
Post a Comment

<< Home
Newer›  ‹Older

  Powered by Blogger   Free Page Rank Tool