Each IT domain has its own constraints.

​

I've currently worked in 2 main ones : Software editor and Bank.

​

Investment Bank domain is a world were a big volume of money flows everyday passing by the managed infrastructure. That's why it require reactivity, good communication and respect of the processes.

​

Incidents must be well identified/prioritized and appropriate fix must be deployed after having identified the impact it can have in the current service level.

​

 In BNP Paribas, we met this kind of situation in an application named Fish.

Application was load balanced (Active/Active) between differents backends and we had an incident in only one backend. Before restarting everything, I requested what was the service state to the client and answer was "Application works randomly, Please fix it ASAP, It's Urgent! It's a critical application, Business is impacted!". In this situation we wanted the less service interruption possible. After having identified the failing backend, we got it out of the Load balancer configuration.

 

It offered 2 mains advantages:

 - All the requests came to the working backend so issue workaround was found

 - As the failing backend was isolated, application team and us were able to work on it without any production pressure.

​

​