Navarotto will speak at WAN-IFRA's Digital Media World conference in Vienna during a session on Monday, 10 October, titled Technologies & Tools for Modern Newsrooms, part of the World Publishing Expo. Here, he tells us about a few of Monrif Net's recent projects, including the implementation of Platform as a Service and how the company is benefitting from it.
WAN-IFRA: Your company recently made a complete review of your digital delivery infrastructure. How did this undertaking come about and what did it entail?
Cesare Navarotto: Over 10 years ago, Poligrafici chose to host the digital infrastructure internally, taking advantage of the server farm which was already in place for hosting the newspaper editorial system.
This setup gave us some short-term savings, but to the detriment of scaleability and flexibility.
Last year, we had to decide whether or not to invest in a full upgrade of our in-house hardware and connectivity, and we made the strategic choice to move to the cloud instead.
First of all, we wanted to migrate our front-end infrastructure to the cloud, in order to get nearly infinite scaleability and to be able to cope with those traffic peaks that many times in the past had saturated our servers and bandwidth.
At the same time, we wanted to keep the web CMS, Atex Polopoly, in our internal server farm because it is tightly integrated with a number of legacy applications (for example, the newspaper editorial system) and there could be unforeseen impacts in moving it altogether to the cloud.
Front-end in the cloud, back-end in the internal server farm: this means effectively going for a headless CMS approach where the front-end's servers contact the CMS via an API and then distribuite the content to the final users.
In order to fulfill our migration plan, we needed to completely re-design our software and we chose to follow a microservice architecture approach: There would be a service responsible for content retrieval from the backend CMS, a service responsible for templating and rendering, plus a number of other ancillary services each dedicated to specific functions, all working independently towards a common goal (i.e. publishing our sites!).
Once we had designed our new architecture in microservices, we had to decide how to implement them and we wanted to do it in a fully portable way, so that we would not be tied to any particular cloud provider. Docker containers were the answer.
As you can imagine it was a complete revolution of our old way of working, and I am really proud of what my team accomplished in less than nine months of work.
You’re making use of Platform as a Service, but since this is a pretty new concept for most of our readers, could you please describe what it is and does?
Gartner defines PaaS (Platform as a Service) as a broad collection of application infrastructure (middleware) services (including application platform, integration, business process management and database services).
We can say that PaaS automates the configuration, deployment and management of applications in the cloud: If, for example, you need a database instance for your application to connect to, with PaaS you will be able to start it with one click, already pre-configured and ready to go.
Most of all, when necessary, you will be able to scale everything up in a second, just moving a slider control in the PaaS admin panel (but be aware that the more you slide, the more you pay!!).
On the negative side, I can mention two points: costs can soon become steep (after all, you must pay for the services that PaaS are providing) and, depending on the provider you choose and on how your applications are built, you can risk falling into a vendor lock-in situation where it would become very difficult for you to change providers if you ever needed to do it.
In order to mitigate these problems, we adopted an open source PaaS solution (have a look at www.convox.com if you want to learn more), running over Amazon cloud, minimizing costs and at the same time, taking advantage of Amazon's reliability.
What have been some of the benefits you’ve realized since making the change to Platform as a Service?
One big advantage is that operations work is reduced to an absolute minimum. We used to have an extra consultant just to keep our servers safe, taking care of updates, logs and backups, whereas now our developers manage the entire infrastructure by themselves.
Another advantage is that this architecture allows us to deploy new features and upgrades to the production environment in a fully transparent fashion, with the possibility of easy rollbacks in case anything goes weird.
How many staff do your newspaper operations have working with digital delivery and did your move to Platform as a Service require much training in order for them to work with it?
Our digital operations team currently counts only five full-time developers, and with such limited resources, working in an agile fashion was more a necessity than a requirement.
In order to fulfill our migration project, the team needed to learn a number of new technologies, so we hired an experienced consultant who worked part-time with us on the project, helped us to put it on the right track from the beginning, and then acted as a sort of oracle any time we got stuck with something. So we could say that it was basically training on the job, but with an experienced guide.
What developments are you planning next?
We need to get to know our users better; only with this knowledge will we be able to offer them a better experience and increase their engagement and thus their intrinsic economic value.
Working towards this goal means to have in place a complete customer identity management solution which would collect and enrich user profiles, interfaced to a data management platform that would activate them from the editorial (ad hoc content recommendations) and business (selling user segments to advertisers) points of view.
Last but not least, we are partnering with an academic spin-off startup to develop a semantic engine which will help us categorize and tag (i.e., extract entities representing people, facts, locations, organizations) all our content.
Each of those projects alone would far exceed the scarce resources of my team, and therefore, our approach is always based on selecting and integrating the best solutions that we can find on the market.
In some cases, "best" is the market leader, in other cases "best" is the only solution that is also available as open source and therefore carries no vendor lock-in risk. We are lucky to be nimble enough to be able to make bold choices and at the same time have a large enough audience to get meaningful results from our tests.