Monday, February 2, 2015

Standard Bank: Our DevOps Journey (part 2) [feedly]

Standard Bank: Our DevOps Journey (part 2)
// Chef Blog

This is the second entry in our ongoing, bi-weekly series examining our customer Standard Bank's DevOps journey. You can read the first entry here and continue below for part two.

Planning the DevOps Journey

Standard Bank brought in Chef to help plan and execute its DevOps journey. The first of a series of planned visits was three days long and, during that time, Chef and Standard Bank worked together to develop an approach. From a broad perspective, Dawie Olivier, who represented the development side and Mike Murphy, who represented the operations side, had brought together the sharpest minds at Standard Bank to answer these questions:

  • Was the group willing to make the changes required to adopt DevOps?
  • If not, was the group willing to accept the consequences (basically, irrelevance and obscurity) of continuing with the old ways?
  • If the group did want to adopt DevOps, would they work with Chef?
  • Would the group accept the responsibility of planning the journey?
On the first day, the Chef team spent time with about 35 Standard Bank executives. For many of the participants, it was the first time they had worked together. Chef took everyone in the room through basic DevOps, an explanation of automation, and how Chef fits into the story. A key point was that, in order to implement continuous deployment, some companies try to use that process on infrastructure, while others try and use it on applications. Chef integrates those two aspects. With Chef, you can create a single train for applications and infrastructure.

In the afternoon, the audience learned about the business reasons for using Chef and received a crash course in Chef fundamentals. For example, the group learned about resources, roles and environments. Although the people in the meeting weren't the people who would write the cookbooks, the objective was to give everyone involved a common vocabulary. DevOps is about inclusion and a shared language makes that possible.

At the end of the day, Chef asked, "Where are you now?" "Where do you want to be?" The Standard Bank group went off on its own to devise a plan. The group wanted to decide how best to use the next two days to answer those questions. Here is their agenda.


The group wanted to outline:

  • The vision and scope of the project
  • The function of the project and the assumptions behind it
  • Release train capabilities
  • The organization of the team that would work on the pilot project
  • The cadence of the project
  • The environments used by the prepaid feature
  • The budget
The group decided that part of the Chef team would spend the next two days training the engineers who would actually write cookbooks and oversee the release pipeline. At the same time, some of the executives would meet to decide on what area of the business would be best for introducing Chef. The goal was to automate an entire slice of the stack and to build an entirely automated pipeline.

The executives wanted to select a feature that would provide customer value immediately. It would be pointless to put lots of effort into something that didn't benefit the bank's customers as soon as it was available. Secondly, the feature needed to belong to a group that already had experience with disciplines such as continuous integration. These two criteria were best met by using a feature that was part of the bank's Internet portfolio. Those features are close to the customers and one of the teams that handled that portfolio—the mobile team—was one of the teams in the banks most truly driven by the need for velocity.

Of course, stability was also a concern. Although the feature should be a critical one, it shouldn't be something so central to the bank that, if it failed, the bank couldn't function.

From a range of options, the group decided to implement the prepaid feature, which would allow bank customers to prepay utilities such as electricity and broadband. The pilot project would create the mobile app, with a web front end following shortly afterwards. Eventually, it would also be possible to use the feature from an ATM.


The group also discussed the team that would work on the project and some guiding principles. This is the subject of the next blog post.

At the end of the three days, everyone came together and Dawie spoke to the group and summarized the situation. "We are going on a journey and we want to implement the prepaid feature. We want to use Chef. Every two weeks we will demonstrate success and have demonstrations. If anyone thinks this strategy is incorrect, you have to speak up now." No one disagreed and everyone had enthusiasm for the project. The group was ready for change and ready to take risks. It was time to put aside the old way of doing things. It was time for something new.

Next: The Chop Chop Team





Shared via my feedly reader

Sent from my iPhone