Since 1993, Somos has operated the SMS/800 Toll-Free Number Registry under the FCC tariff, enabling Toll-Free Service Providers to reserve and manage Toll-Free Numbers. Somos also helps businesses enhance the value of their Toll-Free Numbers by adding text messaging, multimedia routing services and other capabilities.
Somos’ mainframe systems and processes served them well over the years but the costs of operating that mainframe were constantly rising, it was difficult to add new features quickly, and quality was suffering because there was never enough time to do thorough QA testing. A typical production release required 30 to 40 engineers between 12-36 hours to complete. At the same time there was constant pressure from customers and business units to release value faster and reliably. If that wasn’t challenging enough, a number of key mainframe personnel were retiring in 2018. Somos needed a way to meet those demands and gracefully modernize their operations.
“To help our customers move into the 21st century, we had to future-proof our business and our operations by moving off the mainframe to a more modern, adaptive environment,” says Gary McKay, Scrum Master at Somos. “Once we made the decision to embark on that journey, we decided that a distributed microservices architecture would make life easier for our customers while enabling us to release new features when our customers demanded them.”
Somos’ plan was to safely adopt new technology into their pipeline, like containers, but they also realized that they couldn’t completely eliminate their heritage mainframe applications and interfaces. “We still have a lot of business rules and capabilities on that platform that we didn’t want to lose,” says McKay. “We wanted to encapsulate that functionality in a microservices architecture. Our prime directive was to do no harm and reduce interruptions for our customers.”
The team’s goals were to:
- “Abstract away” the mainframe interfaces so those applications could continue to work and be incorporated into the microservices architecture
- Create a bridge from the rigid mainframe software release process to a flexible DevOps culture
- Provide approved automation as a service for predictability and more testing time
- “Unclog” the pipeline to accelerate releases
McKay first did value stream mapping to decompose the legacy code into microservices, and to re-engineer and improve their processes. It also helped build trust among all the different groups in the organization as everyone could see and contribute to the process – all while continuing to run the business.
McKay had used another well-known release orchestration tool in his previous organization and had practical knowledge of its shortcomings. He found the other tool prescriptive and proprietary. McKay said, “my goal was to get an ARA tool where it didn’t restrict me to the way it did things.” The tool in his previous organization locked him in to the way it did releases and deployments and that was a problem for him. “I wanted to find something that was flexible as well as powerful enough to do what I needed to do in our environment.”
McKay turned to CloudBees Flow to standardize environments, automate deployments, and predictably orchestrate all their releases. By publishing reusable pipeline and application components in a self-service catalog, developers can now deploy a QA environment themselves in a few minutes, for instance, without requiring an entire team to do it for them. McKay was also able to collect and share metrics that made it easier to improve processes, to build trust amongst all the teams, and to provide better oversight and control of 3rd party vendors.
With CloudBees Flow orchestrating its entire pipeline, Somos reduced overall release times to one hour from over 12 hours and dropped its cloud services deployment times from 12 hours to 20 minutes. “With the time we saved, we are able to better leverage our existing resources,” says McKay. “Rather than using third party vendors for everything, we are now more self-reliant which, in turn, provided better cost optimizations.”
Somos is now able to add services based on customer demand. “Previously, a customer would have to wait up to 18 months for us to roll out a new feature they needed,” says McKay. “But now we can roll out a new feature in a few days or weeks. It’s allowed us to grow into new markets. For example, we just expanded to texting and smart services, opening up a new revenue stream for us.”
Somos’ microservices architecture and DevOps culture, driven by CloudBees Flow, has changed the way they do business:
- Releases take an hour, with five people, instead of 12-36 hours with 30+ people
- Deployments to AWS take 20 minutes, instead of 12 hours
- Able to deliver new customer value, in new markets, to keep up with demand
- Product quality soared, reducing Help Desk loads
- Marketing can work with customers to plan new promotions and features
- Less reliant on 3rd-party vendors
“By having everything in the pipeline, releases become a service rather than an event. That makes my life a whole lot easier,” McKay observed. “I like not having to work nights and weekends anymore.”