One thing I find fascinating is trends and statistics. I have some monitoring set up using munin so that we can keep track of all sorts of server stats….number of web requests served, MySQL queries, memory usage, etc. It’s all that good stuff that any good (or paranoid) web developer wants to know about.
Looking at the number of database requests, it’s very obvious where the weekends land since the queries are way down.
Munin is pretty flexible, so I thought it’d be really useful to see some visual trends for how our users are logging roasts. With just a little work, I was able to start recording and charting the number of roasts we receive every five minutes. We just finished collecting just over a month of data, so here is what the chart looks like for the past 30 days.
This chart shows the number of new roasts in the previous five minutes. The gaps (or lower numbers) right after the numbers on the x-axis (24, 25, 26, etc.) are Friday through Sunday. As soon as Monday hits, everyone seems to get back to work. Apparently Friday is a big day to turn the burners off and do some maintenance. You can even visually line this chart up with the database chart above and see how they’re related, in terms of the timeline.
I started doing some DB queries to see some other things. Here is the number of roasts per day from January 1 until yesterday, July 19.
You can see how, as we’ve added new customers, the average number of roasts each day is increasing. In fact, yesterday, July 19, 2010, was the biggest day we’ve every had with 124 roasts logged.
So, what is the most popular day to roast? I keep an eye on these charts every day, and empirically felt like Tuesdays always had the biggest spike. With a little database query, here are the results for number of roasts grouped by day:
+---------------+----------+ | DAYNAME(date) | count(*) | +---------------+----------+ | Sunday | 200 | | Monday | 1272 | | Tuesday | 1385 | | Wednesday | 1048 | | Thursday | 1199 | | Friday | 366 | | Saturday | 132 | +---------------+----------+
Also note that all of this data was looked at without knowing (or caring) who was doing what. I’m just looking at trends here, not keeping tabs on our users. This is also accounting for timezone differences. Apparently roasters on different continents keep very similar roasting schedules.
Numbers are fun.