Guild Wars 2

How does updating the game and the servers (and worlds) of GW 2 work exactly?

I noticed that even in GW 1 you never had to take anything down like other mmos for maintenance (unless something royally was broken—which occurred at least twice already in GW2’s history) to patch both client and game server.

How does it work and what methods are used, if any, to keep it all running smooth between old and new builds and versions of clients?

If its proprietary then give the basic terms like one would learn in a database admin class or something.. (which I took but forgot the terminology.)

Game design and programming is something I’d like to do so i’m very interseted in how this works.

Also other questions: is this a dedicated server or a listen server method of connection or is it something entirely more advanced? I know its definitely not peer to peer.

I’d love to have a dev or an experienced member of the game community answer these questions.

If it must be moved please move to a proper place.

Asuran technology.

Check out this:

http://www.gamasutra.com/view/news/180776/Video_Guild_Wars_2s_programming_tricks_revealed.php#.UTPsClded4M

Some interesting information there.

This quite impressive… I understand how it works a bit more. Its really cool how we can stay up at the same time as a new version or build is going in.

Basically, they have a series of computers acting as a cloud that dynamically creates instances of maps and balances processing between them. An inactive map may use 1% of the processing power of 1 server, whereas a heavily active map may use more than 1 server’s worth of processing. All worlds are virtual and exist in the cloud, compared to the older concept of each world having its own dedicated server. Both NA and EU have their own clouds.

For simplicity, you can think of maps as a virtual machine and the cloud as a computer. The computer can run many virtual machines with different types of operating systems (in this case, maps for the specific build ID). When an update occurs, new instances are created with the new build while old ones are set to be destroyed. They don’t have to worry about old instances being created, since it can’t happen, and it doesn’t matter if an old instance crashes, since it will just force an update. If an update is major however, to a point where an old instance can no longer communicate with the system, such as a modification to the character database, then a full reset is required, which is when the servers go down entirely. From the virtual machine perspective, this would be when an update to the software running the virtual machine is required.

When logging in, you connect to the login server which processes your build ID and either passes you through or requests you to update.

So in other words… other MMOs do not do this advanced method of side by side updating via cloud computing and virtual spaces and that we must still suffer through 5-10 hours of down times every week (looking at you SEGA (PSO2), WOW and <insert game name here>).

Is it similar to the hot swapping and stuff of hardware and software in database admin and such where you don’t need to shut down? I still can’t put my finger on those terms. Trying to find what would be closest to regular cloud computing and running databases if I am to start my own game from the ground up as I would ahev to start with that anyways.

At times like these, I wish I kept my books, and still were in touch with my professors of old. I took stuff like sql and database security admin/oracle database courses.

Yumiko Ishida.3769:

So in other words… other MMOs do not do this advanced method of side by side updating via cloud computing and virtual spaces and that we must still suffer through 5-10 hours of down times every week (looking at you SEGA (PSO2), WOW and <insert game name here>).

Is it similar to the hot swapping and stuff of hardware and software in database admin and such where you don’t need to shut down? I still can’t put my finger on those terms. Trying to find what would be closest to regular cloud computing and running databases if I am to start my own game from the ground up as I would ahev to start with that anyways.

At times like these, I wish I kept my books, and still were in touch with my professors of old. I took stuff like sql and database security admin/oracle database courses.

As I understand it, ANet leases server capacity in bulk from Akamai, at least in the U.S. (not sure who they use in Europe). ANet has a “port” of sorts into that lump of capacity and (as Healix points out) can run as many virtual machines from that lump of capacity as they need in order to handle the game. For the most part it is all transparent as far as ANet is concerned, and you (or anyone else) could do it the same way. I’m not sure about Akamai, but I know that places like Amazon will lease capacity to individuals as well as very large corporations (I thought I read that Netflix leases capacity from Amazon) and you can use it like a virtual machine from your own desktop.

I think the posts so far has been right. Basicly, the server capacity depends on the amount of people being online on the server. So what Anet does is to keep the ‘old’ virtual servers running with the old build, but forcing people to get the new build (with an announced kick). they keep running till the last person is out, but the capacity is brought down with the amount of people leaving. the new build is running on ‘new’ virtual servers and that capacity is increased when mor epeople get the new build. The update server is separate and get the hard hits during the update times.

Off course this all uses a lot more resources then regular gameplay. but they have the cloud to get more resources during update times.

I would love to see a dev talk here too, but we seem to have already figured out how GW 2 ticks.