Visualizing and managing cloud expenditure with DoiT technology

With DoiT console, the 20 Minutes team can now see exactly what cloud resources it is using and their cost, both for its serverless infrastructure on AWS and for its data analytics workloads running on Google Cloud.

“The DoiT platform lets us analyze on a day-to-day basis the cost of our AWS Lambda usage, our Amazon S3 storage costs, and all our workloads,” says Jérémy. “We check it daily. It’s practically the first thing the CTO does when he arrives in the morning to understand what we spent the day before and whether we need to adjust anything to keep our costs in check.”

Something that the 20 Minutes team regularly adjusts to manage its costs is its caching. The news site pre-generates a large number of pages, which it caches, ready to be used when a journalist writes an article. This way, it can avoid spiking costs by creating its pages all at once. However, that pre-generation and caching can also be expensive, and 20 Minutes previously found it hard to understand which option was more cost efficient. Now, with the advance reporting from DoiT, Jérémy and his team can strike the right balance.

“The console allows us to see the big picture of all our costs,” he explains. “If we look at our dashboards and see the costs are rising, we can increase the cache for say 10 minutes, and then we will immediately see the costs coming back down within the day.”

Similarly, 20 Minutes can use the forecasting function in the console to predict its costs over a certain period. This helps it to keep its costs down in the face of traffic spikes by increasing or decreasing the cache accordingly.

Spotting spiking costs quickly with Anomaly Detection

As well as monitoring its cloud usage in the console, 20 Minutes uses Anomaly Detection to spot unexpected spikes in its cloud costs, enabling it to act quickly to resolve the issue. For example, every time a journalist posts a story, 20 Minutes uses AWS Step Functions to automatically sync the content with its partners. When a bug in its Step Functions configuration led to spiking costs, 20 Minutes immediately received an alert flagging the anomaly and prompting the team to look into the problem. They realized that the system was making too many calls to Step Functions, causing a sharp rise in costs.

“Thanks to Anomaly Detection we spotted the issue quickly and fixed the bug, and the next day we could see that Step Functions was going back to normal, saving us from having to wait until our monthly invoice arrived to spot and fix the issue,” Jérémy explains.

20 Minutes is also able to reduce its cloud costs further with DoiT Flexsave™, which automatically allocates discounted rates to the company’s AWS compute workloads, reducing its on-demand costs and helping to lower its overall cloud bill.

Freeing up developers to develop new services

20 Minutes doesn’t just save money with the DoiT technology, but time. Without full visibility of its cloud usage, 20 Minutes had to spend a lot of time trying to diagnose and fix the cause of rising costs. For example, when the company experienced a problem with the development of its API, it was initially hard to understand what the issue was, and it took a lot of time and effort to resolve. “If we had the DoiT console at the time, we would have been able to identify the issue quickly and save around two weeks of development time,” says Jérémy.

With these time savings, the 20 Minutes team has been able to dedicate team members to more valuable projects, developing innovative services for its staff and readers. For example, two of Jérémy’s team were recently able to dedicate their time to developing a new generative AI tool to help journalists quickly reformat news stories received from the news wire service. This in turn allows the journalists to dedicate more of their own time to creating original content.

Access to cloud expertise on demand

20 Minutes maintains regular contact with the DoiT team, with a direct messaging channel for instant communication, as well as monthly check-ins to go through any outstanding issues. Whenever they need further support, the 20 Minutes team turns to DoiT Cloud Solve and a Cloud Reliability Engineer (CRE). “The DoiT CREs always react quickly and are precise both in their analysis of the problem and in the solutions they bring,” says Jérémy. “They help us to fix issues in a matter of days.”