reborn55 - Fotolia
The rise of search engine optimisation (SEO) and social networking sites have drastically changed the way many internet users access and consume news content, while making the process of engendering brand loyalty that bit harder for online publishing houses.
In response to these pressures, regional and national newspaper publisher Trinity Mirror Group overhauled its technology strategy and the management of its IT delivery teams to ensure the 100 million unique visitors to its news sites each month keep coming back.
For Neil McIntyre, director of engineering for digital at Trinity Mirror Group, and his 70-strong team of developers, product testers, project managers and system administration staff, this has meant adopting agile working methodologies, cloud-based computing services and new user experience monitoring tools.
“We are responsible for delivering the desktop and mobile websites and the mobile app for our core newspaper business. We ensure that the technology strategy and delivery align with the wider business and product roadmap,” he said.
“The goal for us is to be as agile as possible, so we can deliver new code as frequently as we can and measure the effect of that, because we have to make sure everything we do has a positive business and technology impact as we develop our platforms,” he continued.
A move to AWS cloud
The preparatory work for this started around 2011- 2012, when Trinity Mirror Group canned a nine-year partnership with its previous hosting provider, and set about moving its sites over to the Amazon Web Services (AWS) cloud. Around the same time, it also embarked on a push that would see it swap its bespoke content management system (CMS) for an off-the-shelf, commercial offering.
The latter move was embarked on as part of a wider redesign of its news sites – which include The Daily Mirror, The Daily Record and The Manchester Evening News – that fill the company’s product portfolio, said McIntyre.
“We host everything in Amazon, and the migration over to them was also part of the move over to the new CMS,” he explained.
“We were using a dedicated hosting provider before, which had a single point of failure as it just had one datacentre, and we suffered a few outages that ended up being quite expensive. One of the remits I was given, on the back of this, was to look at our current hosting strategy.”
“There was nothing that could compete with AWS from a cost point of view, and their self-service tools for creating instances were way ahead of what other providers in the market were offering,” he said.
In light of the organisation’s commitment to delivering a positive user experience and its past history of outages, AWS’s promises about resiliency and failover datacentres were also a big lure, added McIntyre.
Performance monitoring challenges
As the roll-out of the CMS and its site redesigns progressed through 2012 and 2013, McIntyre said his team started to realise the tools they had in place to monitor the performance of its web properties were not fit for purpose any more.
Particularly when it came to ascertaining, in real-time, what the user experience was like for visitors to the news sites as his team began rolling out website code, he added.
“We realised the tools we had for measuring the performance of page loads and the advertising content in the pages,vfor example, weren’t really sufficient. The tools couldn’t give us an accurate baseline of where we were and how we could improve our page load times going forward,” he said.
For example, if users reported slow page loading times at a specific time of night on a particular mobile device, it would be difficult for the firm to ascertain the cause.
“What we couldn’t say with any degree of certainty was if it was a piece of code we released that was causing a problem for a specific mobile operating system, which is the level of detail we required,” added McIntyre.
This kicked-off another period of product evaluation in 2014, as the company set about finding a suitable tool to fit the bill, in anticipation of another wave of website redesigns.
McIntyre said the priority was to find a customisable technology that would allow it monitor the performance of any part of its front-end operations, including page loading times and the activities of the third-party tool – called Omniture – it uses to measure website traffic.
Neil McIntyre, Trinity Mirror Group
The organisation had some ambitious page load speed targets to hit in the wake of its redesigns, McIntyre said, and ended up opting for SOASTA’s performance analytics tool, mPulse.
“The nice thing about the mPulse product was the depth and the amount of data we are able to obtain from it,” said McIntyre.
Having this in place, he explained, paved the way for the organisation to adopt a continuous delivery approach to rolling out new code across its sites.
“We release code around three times a week, and sometimes that isn’t frequent enough, but having this product in place allows us to understand what changes we’re making and where we need to make improvements,” he continued.
Reorganising the team for continuous delivery
“We’ve now set up cross-functional, product delivery teams, featuring business analysts, developers and testers that are all working towards achieving specific business goal, as we move to adopt a more services-based model,” he said.
However, there are still some areas still to address before his team is in a position to release code changes throughout the day, but – with mPulse and AWS’ technology powering the charge – a lot of the groundwork has been laid.
“The goal for us is to be as agile as possible – everything we do has a positive business and technology impact”
Neil McIntyre, Trinity Mirror Group
Next on McIntyre’s agenda is working out how to break up its CMS – which is used across 29 of its core websites and is built using a single code base – into a microservices architecture.
This is the process by which a larger web application is built using a compositive of smaller, modular services that are easier to update individually and on an iterative basis, rather than trying to upgrade the entire thing at once.
“We’re nowhere near being able to deliver the continuous delivery model, as it does rely on us taking a more microservice-based approach to what we do. One constraint we have with doing that is the CMS, which is shared across many of our products,” he said.
“What we are looking to do to get round that is break it up architecturally from a technology perspective into microservices. The more we can do that, the more we can start to release code that doesn’t have any dependencies across other product streams.
“There are other players in the market who have managed this, such as The Guardian, Spotify and Netflix. It is a model we are aspiring to, but we have some technology changes to make before we get there.”
Read more about agile software deployments
- Enterprises need to realise there are no quick fixes when it comes to tapping into the DevOps software development trend, warns Etsy staff operations engineer Jon Cowie.
- The topic of DevOps dominates discussions at many of the major IT conferences, with suppliers and analysts warning enterprises about the business risks of failing to adopt a more agile approach to software development.