Resilience: when technology works to enhance the user experience

Behind every click on SNCF Connect, there's a promise of seamless navigation, even when systems are put to the test. To achieve this, SNCF Connect & Tech relies on a resilience strategy led by Resilience Officer Xavier Debonte who is passionate about the subject.

"My role is to ensure that the service holds up no matter what happens!" explains Xavier Debonte, Resilience Officer at SNCF Connect & Tech, who leads a team of four DevOps engineers. Their mission is to guarantee service availability, even when the unexpected happens. "When a partner is unavailable or our IS encounters a disruption, we need to be able to continue offering a high quality experience - even in downgraded mode," he adds. 
 

This is essential for an application used every day by millions of people. "There's no room for error. What we do is highly technical, but at heart it's simple: we want every user's experience to be the best it can be."

While resilience is receiving increased attention today, it's not a new concept at SNCF Connect & Tech.

The technical foundations have been solidified, and extensive awareness-raising work has been carried out among the teams. "The aim is to get every developer thinking about resilience right from the design stage. Part of our mission is to train teams and share this culture internally," adds Xavier Debonte. 

So, what does resilience actually mean? "It's a set of methods that enable us to be ready in the event of a failure" explains Xavier Debonte. These include:
 

• Self-scaling
   

"When users flock to the website, the system has to keep up. We work with self-scaling mechanisms: the infrastructure automatically adds resources when demand increases. Before the Olympics, for example, we reconfigured everything to absorb a surge in load without breaking down."
 

• Circuit breakers
   

"We use tech circuit breakers when a partner site bugs or responds too slowly, to temporarily cut off access. Rather than penalising the entire system, we isolate the problem. It's transparent for the user and we avoid a snowball effect."
 

• Chaos engineering
   

"Last but not least, chaos engineering. We simulate failures in a parallel environment to observe how the system reacts and check that our monitoring is effective. We also train our teams to deal with these incidents."
  

Every quarter, several dozen people take part in an exercise. "It's demanding, but vital. Resilience isn't just a technical matter, it's a team sport!" concludes Xavier Debonte.

"We have raised awareness among our delivery teams and built resilience into every new feature," continues Xavier Debonte. Critical periods (flash sales, seasonal peaks) are prepared together with all services involved: payments, buyer's journey, infrastructure, etc."Everyone has a role to play in guaranteeing the robustness of the whole."

The next step is to optimise energy efficiency without losing robustness. "We fine-tune our tools to provide the best possible response, including reducing the footprint at night or weekends." Thanks to its choice of 100% cloud, SNCF Connect & Tech is exploring options like spot instances. "It's less costly, consumes less and we maintain the quality of service. In short, everyone wins!"