Wabot Project
The Wabot project is a project to create an HPOO PLC to industrialize our installation processes at the SGCIB.
​
Context
​
Member of the middleware team at Société Générale (SGCIB), My role is to ensure the smooth running of production.
It means :
​
- Ensure that critical or non-critical applications provide their service and that they are not interrupted by various and varied incidents.
- Participate in the implementation of new applications by providing expertise on the technologies I have expertise.
- Respect the protocols established by the bank in terms of communication so that any event impacting the technologies I manage informs or alerts the right people at the right time.
​
The fact is that I work in a large bank, the responsibility of different expertise required for the proper functioning of the Information System are divided between different teams.
​
The DBA team manages the databases
The UNIX team manages Linux / Unix servers
The Network team manages network infrastructures ...
Our clients are the application teams whose role is to make live the applications they manage but also to implement new ones.
​
Why did we choose to automate?
​
In a process of continuous improvement of our service provided, we took the time to analyze and discuss with the customer about the process of provision of environment ready to use.
​
​
​
​
On average, between the time of the customer's initial request and the delivery of the platform to the customer, it took about 2 to 6 months!
The average installation / configuration time of a container was half a day when everything was fine.
​
This was explained by several elements:
- A certain number of teams had to intervene manually in the delivery process of the platform and in a defined order
Unix / Windows teams install and configure the server and give rights to the teams that have made the requests (Middleware / DBA ...)
Middleware teams check access and prerequisites (usually with several exchanges with the previous team to correct more or less important omissions)
- The Infra teams exchange very little between them (they take care of their perimeter but the decisions are not taken taking into account the evolutions of the other teams).
- Processes are little or not automated.
The years 2013/2014 saw a clear increase in application platform installation requests. So we had to question the service we provided.
​
At that time I was technoleader Apache / PHP that is to say that I was responsible for the evolution of this base (technological watch, management of CVE, development of the installation methods of this Middleware base).
An initiative is born from all these observations: automating our installations.
​
Of course our facilities were already scripted but it was to automate at a higher level:
- We did not just want to execute commands sequentially but to have a system to warn in case of error, or even to perform automatic remediation.
- We wanted to shorten the delivery time to the maximum
- We wanted to interface this system with the tools we already used (Ticketing tools / Supers ...)
- We wanted it to free time so that we could focus on important topics.
​
The Project
​
I did not know HPOO at all before participating in this project.
So I took part in setting up the Apache HTTPD installation module and I created the one to automatically monitore any newly created Apache.
​
The ability to have a monitoring module for each of our technologies was critical because it allowed us to be able to deliver viable platforms in production with the same speed as non-production.
Wabot was a very nice team project.
​
Indeed we (all the technoleaders of the team (Apache / Tomcat / Weblogic / IIS)) developed our modules at the same time which allowed us to discuss the difficulties we encountered in order to solve them more quickly.
​
The result of this project is that the modules of installation / supervision of our containers have made it possible to go from half a day of manual installation to 30 minutes by removing the risks of human error.
This project was therefore a big success.
Subsequently, we set up a gateway between Wabot and our ticketing management tool to provision platforms via a simple ticket.
​
Future
​
Wabot plateform is now decomissionned ... But that's not important at all.
​
The important thing to remind is that this project permited to my coworkers and I to fully understand the big value of automation in a big compagny context.
​
Today I think differently and we put the automation to an higher level (#HAPI).
