As part of the continuing series of articles giving some insight into the work that goes in to building and running one of the largest online dating companies in the world, it’s my turn to explain exactly what the fairly ambiguous job title of “Architect” has me spending my time doing…
What it actually means is that it’s my job to ensure that the WhiteLabelDating.com platform is, and continues to be, built in such a way that means we can continue to provide exceptional service to our millions of members and thousands of partners, whilst simultaneously building new features, such as the recently launched mobile interface.
The day began a little strangely: I was in the office by 5am, because I’m off to Japan on Wednesday to compete for Great Britain in the World Ultimate and Guts Championships, and luckily Global Personals are flexible enough to allow me to attempt to timezone-shift as much as I can before I go.
Actually, that previous paragraph started with a lie; I was at the office by 5am but decided to test the efficiency of our awesome 24-hour moderation team; I’d forgotten my external door pass. With security not starting until 7am, I was kind of stuck. I ended up sending an SOS in a profile update which I knew the mods team would read and, if they were as good as I knew they were, would come and rescue me. Five minutes later, I was sat at my desk with my customary morning bowl of porridge, thanks to the well-stocked GP kitchen.
Strangely, in this world of electronic communication, I get relatively few emails (and most of those I do get are GitHub notifications!) so it doesn’t take me long to scan through the few I’ve received or not dealt with over the weekend. In this case, these included some ups on a few actions from Friday’s development team management meeting. I also spent half an hour continuing the organisation of the second of our internal hack days, which we’re aiming to hold in August.
We’re currently in the process of reviewing our application stack for significant performance limiting processes, so much of my time at the moment is spent investigating problem areas, employing additional monitoring and reviewing the systems we have in place to work out which parts need tuning. I spent some time this morning adding extra metrics to our ColdFusion web development platform to allow us to break down request timings through our performance monitoring programme, New Relic. I also looked at the data collected over the weekend so I could make suggestions to the operations and development teams as to where they might like to direct their attention.
Next I turned my attention to reviewing changes that some of the team have been making for moving the internal API services that power the WLD mobile application to JRuby, a java implementation of the Ruby programming language. I’ve also been looking at additional changes to the caching within the internal service so that in 90% of cases we can completely skip the application server and just rely on nginx serving cached content directly from memcached. Part of this involved debugging some changes we’d made to the hashing algorithm.
By this time, the office was starting to fill up, and I had a chat with one of the leads on the the UX development team about our current use of Facebook Connect in the primary web application, and how we can look at reducing front-end page rendering times with a few relatively simple changes. That was followed up with some discussions with our Project Management team as to where team resources were required this week, and on which projects my input was required. I also had a chat with one of our developers who’d come up with an excellent suggestion on how we could use a HMAC-based solution to provide even better security with the links we send out in emails to members.
After discussing the current state of the performance work we’re doing with Ian (our Technical Operations Manager) and a couple of the lead developers, it was time for our weekly development meeting. We get the whole team into the boardroom, the guys working in the London office Skype in and we spend half an hour going through the current projects we have running and anything significant that needs to be discussed with the team. This is generally really useful to make sure that everyone is aware of what’s going on within the department.
For the rest of the day, I worked with Operations to implement a few of the performance changes we’d been investigating and a very productive couple of hours it was! With a few changes, we were able to shave around 20% of response times for all our core application requests.
With a full day done, I headed off home just as most people were heading back from lunch!
Back to blog list