Wednesday, December 9, 2009

Patch 3.3 First Impressions


Blizzard you are never prepared!

Echos of the most famous saying, you are not prepared, in burning crusade the day of launch are heard...every...single...content..patch. Because there are so many problems. I know what your thinking "But Skonged Blizzard does their best and everything is all complicated...blah...blah..blah" Sorry but Blizzard is not getting a free pass from me. I also work in IT, have for many years, when something big is going to happen you want to be ready. What could be bigger than the final content patch where we face down the last remaining bosses until Aurthus, last talent changes for at least 6 months, and everyone reactivating their account and logging on to try the new LFG system. That a lot of activity and here is my suggestion to better handle the situation.

Triple everything.
-Cluster every server with at least 3 of the highest end servers.
-Upgrade network to fiber channel with load balancing.
-Buy more bandwidth or if topped out move to data center with higher bandwidth availability.

Clustering is the new hot high availability in server rooms now. Blizzard needs to , with highest priority for customers, spend the money to rebuild wows foundation on the fastest, most robust, network infrastructure money can buy. Stability and speed come down to one thing in the server room. Cost. To rebuild will take millions, but they can afford it. In fact they lose money when things don't go as planned. Spend the money Blizzard. Every content patch.

Test Test Test
-Simulate max ammount of people connecting, simultaneously.
-Simulate max load and see where you can upgrade.
-Simulate max raids/instances groups

Testing content is important but your priority should be testing the foundation of wow then content. Don't want to put the horse in front of the bull.

Overall
-Awesome LFG tool
-Great instances
-Content is superb.

I loved the LFG tool and it has been a long time coming. The New instances are a lot of fun. Content is always top notch.

Just prepared next time.

2 comments:

  1. I imagine that they considered "tripling everything" and decided that a) the extra complication and potential downtime all the hardware upgrades would introduce would be worse than the problem it solved

    but mainly,

    b) The cost of adding enough hardware to completely eliminate any lag, instability or queuing for the first week post 3.3 was not worth it. Much the same way that the servers are not balanced for the worst-case scenario when the entire server population all stands on the same spot. It's an 80:20 thing, they get 80% of the way towards perfect, for 20% of the cost.

    It's very easy to just say "olol Blizz has loads of money, they should just buy more servers" but the law of diminishing returns applies to this as much as anything else. The WoW server architecture isn't perfect but it's a *lot* better than it used to be and, in my view, easily "good enough" for the fee that they charge.

    Luis Villazon

    ReplyDelete
  2. For a good network admin hardware/networks are never too complicated to handle. There are ways to horizontally build a new network foundation without disturbing the current one. Once data has been moved and settings are in place. You redirect users to new network.

    When providing a service cost should never be a factor. Cost will be recovered and you get a faster/better network foundation. Considering customers are 100% of revenue 80:20 rule takes lower priority here.

    There is no diminishing returns when it comes to new technology. In fact Moore's law (http://en.wikipedia.org/wiki/Moore%27s_law#Futurists_and_Moore.27s_law) backs this up.

    If you make something good, then with a little more time you can make it great.

    ReplyDelete