Mark Gritter (markgritter) wrote,
Mark Gritter
markgritter

SimCity launch

A lot of people are framing the SimCity launch problems as a DRM issue, and there's some fairness to that. But SimCity's design is "online always" (even your cities are saved to the cloud) so I think it is more apt to compare the game to the launch of a new MMO. One that went horribly, horribly wrong.

(Whether or not you want an MMO SimCity is a different issue.)

The leaks that have come out so far are not particularly technical in nature. But it seems to me that EA made at least three errors:
* No QA at scale. This is excusable since it's very hard to simulate 100,000 real users. But many MMOs have an open beta period to get some feel for how well they scale and whether there's a problem looming. EA did some beta testing, but either did not take the right sort of measurements, or ignored the warning signs. (Internet-scale services are especially difficult to QA because the heterogeneity of client behavior easily dwarfs anything you can capture in a test plan.)
* Insufficient operational planning. Any time you have to "roll out three additional servers" as part of your response, it's a sign you got cheap on the back-end. There was a fundamental disconnect between expected user base, and the capabilities of the service. These systems should have been overprovisioned to start, but potentially cut back post-launch if the service was working well.
* Non-scalable architecture. Most serious SaaS companies can spin up additional servers in the cloud, and have architectures that permit this. (But not all.) The ability to use public-cloud resources is a powerful scaling tool. The fact that EA was cutting game features in order to handle the load suggests that their system was not able to partition load effectively across a pool of resources. (Or, they were unwilling to pay for additional resources for the pool.)

From a distributed systems perspective, the failure is perfectly understandable--- but amateurish. Few startups could successfully go from 0 users to tens of thousands on the first day, and all experience growing pains across all three of the above axes. EA has little excuse, though. It feels like they thought they were just building a DRM infrastructure, when really their design required a much more aggressive operational capabilities.
Tags: distributed systems, games
Subscribe
  • Post a new comment

    Error

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.
  • 3 comments