The problem
The underlying goal of the oil and gas company’s plan with Azure was to adopt a one-cloud strategy that would allow them to migrate all data and applications to the cloud, eliminating the need for data centers. Utilizing a central data hub would accelerate both development itself and the velocity at which new applications could be deployed — two crucial differentiators in this competitive industry.
The company swiftly discovered they could create interesting prototypes with Azure products, but they were not able to quickly deploy them into production, nor could they properly configure a continuous deployment pipeline to fully automate this process. The company spent a year and a half struggling to achieve their objectives with little progress and were at the end of their rope.
“Three years ago, the development pipeline and big data story in Azure was still a little rough around the edges,” said Blueprint Vice President of Technology Chris Carter. “At that point, the documentation wasn’t seamless. It was like buying a car and having to assemble it yourself, but all you’re given are instructions on how to inflate the tires and how to put the engine together. There’s nothing about putting the engine in the car or putting the wheels on the axles.”
One application the company had hoped to build and automate through Azure would enable automation of the drilling process. Normally an operator uses instrument measurements and readouts to monitor where the drill is in the strata of shale that will produce hydrocarbons. The company wanted to build a machine learning model that would use incoming feedback to determine if the drill is in the “pay zone” and guide the drill better than a human.
They were working on the first prototype of that application when Blueprint was brought on as a partner. One developer had Jupyter Notebook on his laptop and was using CSV files to build an algorithm that could predict where the drills were going as they were guided to the “pay zone.” The results were promising, but the manual effort involved in gathering CSV files from numerous sources, running them through the algorithm and then sending the projections to the site operators mean the application wouldn’t scale for production usage. The process needed to be operationalized and automated but optimizing that pathway from the developer’s machine to the cloud to the drill site was where the application’s development had stalled.
“They had no way to automate or streamline that process – or to ensure that as they were doing all that experimentation and work, it was being properly managed from a budget perspective” Carter said. “Part of why they were so frustrated when we arrived was because a month before, they blew their whole operating budget for that team because someone spun up a bunch of Azure resources and forgot to turn them off. Their Azure bill was more than their labor cost.”
The Blueprint Way
Blueprint brought a wealth of experience and knowledge to this project. On previous engagements, Blueprint had addressed many of the same pain points this company was going through, allowing Blueprint to quickly demonstrate which Azure products worked well and which products still needed some work. By directing the company’s focus to the products that Blueprint knew would work well for their needs, progress was finally made after a year and a half of futility.
“It was really about bringing our learnings from previous Azure work and showing them they could achieve their vision,” Chris said. “We have such breadth of knowledge in Microsoft’s product suite and know the best ways to bring their products together to drive the most value. We also have a really straightforward process for adopting Azure, which can be a big scary ecosystem because there is so much there, and it changes so fast.”
After only 48 hours, the company had a functioning development pipeline delivering an application into an operational cloud environment. Over the next 90 days, Blueprint then delivered a fully configured model for taking applications from development to production on Azure, which the customer development teams were able to quickly scale out across the organization.
Blueprint then worked three more months helping the company build automated pipelines for more applications and educating the company’s teams on how they could transition to doing the work themselves. The company’s developers were used to building applications in a traditional manner where they get deployed to a server and manually pushed by a human. By transitioning from such manual development, the company was able to build a completely scalable process that takes advantage of all the benefits of an elastic cloud model.
