Auto Trader Engineering Blog

  • Text Mining Our Dealer Reviews

    At Auto Trader, knowing our customer is very important to us, and we invest a lot in researching what makes them tick. In previous posts, we described how we used text mining to understand our customers through our customer support descriptions. We have another very valuable source of data: our dealer reviews.

    Read more…

  • Supporting Building a Data Lake

    Data is king. Information is power. It’s not just about storing lots of it though. There is no point having years of data without the ability to interrogate it and surface the information required in a timely manner. We at Auto Trader recognised the power of data some time ago. As our data set grew there was a realisation that we needed a better structure, we needed quicker queries.

    Read more…

  • Lead Developer 2017

    I was given the opportunity to attend the Lead Developer conference in London, which occurred on the 8th and 9th of June 2017 in the Queen Elizabeth Exhibition Centre in London. This was obviously exciting enough on its own, but it also coincided with the UK General Election, and the QEII Centre is situated right in the heart of Westminster, opposite the cathedral and behind the Supreme Court, just a stone’s throw from the Houses of Parliament.

    Read more…

  • It matters what you measure

    It’s not enough to simply measure things, you need to measure the right things! When I first introduced release reporting at Auto Trader I based the success criteria on the things the business cared about at the time. It was very basic and a release was deemed to have ‘failed’ if it was either backed out or needed to be fixed. As the way we measured success evolved we noticed how this didn’t really represent the direction we wanted to follow.

    Read more…

  • Resilience4j - a lightweight, flexible circuit breaker

    We knew that our application would break if the database was down. More precisely, we knew that our service end-point would time out when we were logging to the database some not-essential-but-useful information for each item in the request’s large batch.

    Read more…

  • Exception Handling Conundrums

    Our squad recently came across an unhelpful error message* in response from one of our APIs. After some digging around in the code, we discovered that the message was due to an exception that had not been handled appropriately and instead had bubbled up. This prompted us to sit down as a team and agree on some best practices around how we will deal with exceptions. This post will cover what we agreed on, in particular how to handle crossing knowledge boundaries, using examples from our codebase.

    Read more…

  • Adding GitHub Organisation Webhook Support to GoCD

    The bulk of our active codebases, over time, have made their home our GitHub Enterprise server. We also have a GoCD (continuous delivery pipeline) server that is polling these repositories to work out if it has something to do. The upshot of this is that, every minute, for each of these codebases, GoCD polls Github for changes. This consumes a lot of unneccessary CPU cycles (especially because some of these sources haven’t been updated recently) and is one of the reasons our GoCD server is slower than we’d like it to be. This blog post will talk about how we improved this and my experiences while contributing code back to the open source community.

    Read more…

  • Meeting Smells and How To Avoid Them

    As developers, we’ve probably all heard of code smells, but did you know there are also meeting smells? For those of you who didn’t instantly stop reading, you may be thinking “Oh no! You’re that person!”

    Read more…

  • How we halved page load times using Tree Shaking and Ahead Of Time Compilation

    In March 2017 we released a new homepage for our trade users using a new client side technology framework utilising a Component based Angular 2 structure. Currently, measurements of our users’ page load speed tell us that 75% of users see a fully loaded page in under five seconds with 50% loading the page in three seconds, but it hasn’t always been this way. Initial measurements using Google Chrome’s Lighthouse performance benchmarking tool saw page load times of ~19 seconds on mobile devices on 3G and 10 seconds on a desktop broadband connection. Using Lighthouse as a benchmark we set out to reduce page load times for mobile devices on 3G to under 10 seconds.

    Read more…

  • ReactRiot!

    On the weekend 24-25th June, three developers from Auto Trader (Leon Pelech, David Carter and myself) participated in the first ReactRiot hackathon. It was a two-day event with teams competing from all over the globe.

    Read more…

  • Monitorama Portland Conference

    Four of us were lucky enough recently to visit Portland Oregon to attend Monitorama, a conference geared towards (mostly) Open Source monitoring and related topics. This post is about what our key takeaways were, and the new themes we observed. I had previously attended an as yet unrepeated Monitorama EU in Berlin in 2013 and had found it to be a wonderfully inspiring set of talks.

    Read more…

  • Postmortem on the application outage on Fri 26th May

    On Friday 26th May, we experienced a major service outage for 75 minutes around one of our core products, VDS (Vehicle Data System). The result of this was some applications not being able to return vehicle-specific data.

    Read more…

  • Data Modelling for MongoDB

    This is the second of a two-part post looking at how your data model affects MongoDB’s performance. This post describes how we arrived at a problematic data model, some of the changes we made to improve it and describes what we learned along the way.

    Read more…

  • Investigating Query Performance in MongoDB

    This is the first of a two-part post looking at how your data model affects MongoDB’s performance. This post describes a performance issue with one of our applications and how we came to the decision that our document model was ultimately the cause.

    Read more…

  • Using AssertJ

    When writing tests we aim to use Descriptive and Meaningful Phrases (DAMP). This also reduces the time needed to understand the intent of the test. Making assertions read like the sentences we see every day is one way to achieve this. AssertJ provides a comprehensive set of fluent assertions for Java.

    Read more…

  • Developing Your Career

    Many development managers dread that one day one of their staff in a regular catch up will ask “What do I have to do to get promoted?”. This is daft really, because having staff that are motivated to develop themselves is a critical part of a high performing team. Of course, such a question could really be referring to a desire to be recognised—perhaps they don’t feel their title reflects their progress so far. Perhaps they are not satisfied with their salary. Or perhaps they are simply asking where their manager feels they should be concentrating their efforts to learn. This post is about what we have done recently at Auto Trader to help development managers work through this question with their staff.

    Read more…

  • Let's have a conversation

    At the Auto Trader Hack day last year we built a prototype skill to read back stats for a retailer’s stock via Alexa Voice Service API using the Amazon Echo Dot. We wanted the user to be able to ask for information about a vehicle, but also take some sort of action after finding out information. The initial goal was to find out how we could do this and understand what sort of challenges and limitations that we might present. This is my perspective of the journey before the hack and how we got to showcasing the prototype.

    Read more…

  • Using syntax trees for text mining

    So far, machine learning techniques such as data cleansing, frequency mapping and clustering have allowed us to peek at the topics behind a huge quantity of calls from our customers. We’ve been able to show statistics on what our customers think from unsupervised learning. But what if we could employ a deeper understanding of the meaning of the text itself in our text mining?

    Read more…

  • Text mining our customer calls

    At Auto Trader, our support staff receive up to 15,000 calls every two months. It’s a huge operation to co-ordinate everyone involved and get an objective measurement for how we’re doing, and what is making our customers tick this week. Currently, support managers go through a subset of recorded text from the calls we receive, manually classifying the emails in order to understand the volume of calls relating to each topic and identify changes needed to our products or our training. A few months ago, we formed a working group with a few other developers from around Auto Trader to see if we could solve the issue of text mining our calls for an objective and measurable understanding of our customers.

    Read more…

  • Recommendation Engine Hack

    Here's what we learned on the hackathon when we developed a fully functioning recommendation engine for vehicle adverts.

    Read more…