sim.plified.com

We’re winding down our second week with Pattern Tap, but the site is winding up.  It’s 11pm on Friday, and we’ve already surpassed 9k unique visitors for the day.  To say that this is unexpected is an understatement.  The popularity has been welcome, but there have been some difficulties along the way. Neither Matthew, nor myself, had had to manage a site that got this popular, this fast.

I have to hand it to our Server Engineer, Jay Janssen, who is squeezing every last drop of potential out of our server configuration.  We are all learning about the value of caching, light http and breathing a sigh of relief that we hosted all our images on S3 (with a backup plan of course).

Here’s a few things I’m starting to learn.  Not a lot of answers here, just my questions:

Build for Server Caching

How can a dynamic page be designed so that the dynamic portions of the page can easily be cached by the web server? Having all the parameters appear in the query string seems to be helping.  Another idea we’ve toyed around with is to have a “progressive enhancement” type approach.  Where the dynamic portions of the page (like that username at the top of the page) would be populated by Ajax, rather than just automatically appearing.  This way, most of the page can be cached on the server while the minor portions of html are retrieved asynchronously from the server.

Be Ready to Fix Bugs

As much as we try, we don’t catch all the bugs.  We count on users giving us feedback (cheers to the guys are user voice for a great feedback system).  In launching a site I needed to be ready to crush those initial bugs that showed up… I wasn’t ready.  The need to fix bugs fast is escalated when your site becomes popular very quickly.  I can say that for me, this was something new.  I’m used to having request trickle in (because no site I’ve launched has gotten anywhere near this popular) not 8k visitors hitting the site in the second week.  Whether or not your site becomes wildly popular in the first week, you should probably plan some bug fixing time into the first couple weeks of release.

Be Ready to Add that Critical Missing Feature

As much as we try to anticipate what our users will want, there is probably that one missing feature that needs to be added to tune the users experience.

Plan for Scalability

Our plan for scalability was loose. This was mostly our inexperience.  I now see the value of looking at scalability in terms of hard data.  For example.  If your your starting on shared hosting at how many visitors/requests do you migrate to a virtual server.. to a dedicated server.. to a server array?  Can you migrate quickly?  Working through some of these questions ahead of time might help when trying to make a couple of quick jumps early on after your sites release.

I’m sure there will be more to come.  If you’ve got some ideas about some of the issues above, please share your thoughts in the comments.

Technorati Tags: PatternTap,Pattern Tap