Saving Comic-Con Before It Starts Part 2 - Breaking Up the Site

Comic-Con Survival Guide Saving Comic-Con Before It Starts Part 1 – Separating the Geeks

Today’s entry is pretty simple as it follows along the same logic as yesterday’s post. It’s time to split up the technology (or where people access the tech) so that one location isn’t bearing the brunt of the attack…

Breaking Up the Site

Again, a slow (not crippled) server and maybe a few minutes page load time is acceptable to most of us as long as it yields concrete, dependable results (good or bad). Six hours of browser refreshing is something that should never happen again. Same goes for the site telling you that all passes are sold out when they’re really not. Regardless of who Comic-Con partners with next year to get the job done, it has to speak the truth to us, even if that truth is hard for us to hear. It’s better than the mind-numbing frustration and confusion that was felt by many on February 5th. Is one server system enough to handle 100,000+ geeks when they hit a single site with the power of a hundred “denial of service” attacks*?

Send the 4 Day Buyers and the Single Day Buyers to Completely Separate Sites – It’s apparent that no matter where Comic-Con has been sending people each of the 4 times they attempted to sell memberships, it’s not enough. To lessen the burden on the one server (actually, a cluster of likely load-balanced servers), the users should be pointed in two different directions from the official Comic-Con site if all passes need to be sold starting at the same exact time. Those who want 4 Day memberships should be linked to one site while Single Day membership registrations get linked to a totally different site, supported by a completely separate backend system from the first, with no crossover. They should be as separate from each other as you can get, so that one doesn’t have the slightest effect on the other when it all goes down.

Yes, the 4 Day membership site will still be incredibly bogged down, but hopefully not as badly, since the single membership folks would be making their purchases on a site that isn’t fighting to serve the full week convention-goers. However, the process and the company that provides the online purchasing service should be consistent between both sites, especially if a prior registration/login is required.

This suggestion need not be an either/or situation in relation to yesterday’s idea. Implementing both of these ideas (separating the times/dates and separating the backend systems) could help to ensure a more successful registration day for the users and the provider. No, not everyone will get a ticket, because Comic-Con 2012 WILL sell out again within the same day the tickets are made available. In fact, if handled correctly, there’s a very good chance it would sell out even faster than 2011 with these processes in place, mainly because people might not be banging their heads against their monitors for hours on end as they try to decipher whether there are still memberships available or not.

The biggest negative for Comic-Con International would be the added cost in implementing the registration service on two separate systems and keeping them afloat during the onslaught. Even split in two, there’s no telling if even half of the number of people attempting to buy memberships on each site might still be too much for the separate systems to handle. Of course, that’s where testing comes in (tomorrow’s post).

Once again, I’m thinking that a version of this idea would also work for the usually insane upcoming Hotel Registration Day, but that’s a completely different post (seriously… coming soon).

What do you think? Would it help? Would it make no difference at all? Feel free to make comments here in the comments below or on the official Unauthorized San Diego Comic-Con Survival Guide Facebook page and let’s hash this out. I don’t have all the answers and ideas. I may have written a book about San Diego Comic-Con, but unlike some people, I don’t claim to know it all. Maybe you have something to contribute that’s even better. Let’s hear it!

Next: Saving Comic-Con Part 3 – Testing For Good

* “Denial of Service” attacks are what some hackers use to take down websites. They essentially write a script that automatically sends so many “requests” or visits to the site that it basically can’t handle it all and gives up. It’s like a teacher being expected to answer the questions of ten thousand students all at the same moment… and his head exploding as a result. Except, when we do it with Comic-Con, our requests to the site are legitimate attempts to complete a task, not a detrimental attack. Same result.