All case studies / Unily
Unily reduced customer downtime from 1 hour to 9.5 minutes with Octopus
About Unily
Unily's employee experience platform helps enterprise customers connect, inform, and engage employees. Its solutions bring global employees, stakeholders, and teams closer together, whether they work in the office, remotely, or on the frontline. Unily's out-of-the-box enterprise intranet solutions help some of the world’s most successful businesses by providing tools to communicate, collaborate, and innovate.
Challenges
Unily needed to automate its deployments to scale
Unily’s platform is SaaS-based and hosted in Microsoft Azure. As Unily’s team scaled, their manual deployments became more complex and time-consuming. Even with only 20 customers, deployments took an entire day initially. The team had to deploy on a Saturday to minimize downtime, usually starting in the morning and taking several hours to push updates out. After 3 months, Unily purchased Octopus and changed to a more managed deployment process, providing them the ability to deploy during the week. As a result, they completed roughly 200 deployments a month by late 2020. As the company scaled, it needed a tool to increase its deployment frequency and speed as the requirements grew to do more deployments.
Unily also wanted to improve its downtime for customers. Customers choose their upgrade windows so the downtime window works for them. The Unily customer success team manually managed and booked these, which became time-intensive. Some customers were overdue on some complex deployments, and downtime was over an hour. Unily needed to automate deployments and reduce downtime while still providing customers with flexible upgrades.
Unily’s deployments had to meet the needs of a range of customers ranging from small to enterprise requirements. The team wanted a tool to increase efficiency that let developers focus on more critical tasks.
Sometimes, I had deployments that almost broke me. I’d be there on a Saturday and still be pulling my hair out on a Monday. It was so time-consuming.
Solution
Octopus helped Unily increase deployments from 200 a month to 3,000 a month
Unily implemented Octopus in 2016 which let them automate deployments to all customers. In 2020, the team sought to further increase automation, minimize downtime, and improve efficiency by creating a dedicated deployment team.
Now, Unily uses Octopus to automate its deployments as part of its automated CI/CD pipeline. The team manages customer changes via ServiceNow. This ensures processes are compliant and automatically starts deployments when a customer requests an upgrade. Octopus integrates with Unily’s homegrown automation platform, automating other tasks via the Octopus API. The team can also choose to deploy different versions to customers and when to deploy to a customer group or a single customer. Using Octopus provides a seamless experience for Unily’s engineers, so they spend less time pushing buttons and more time on critical tasks.
Using Octopus, the team have a process to minimize downtime across their deployments using Octopus’s custom step templates. Their first 23 steps run while the client site is still live, and then the site goes offline to complete the critical work. This has minimized downtime, as they can complete steps in parallel without bringing the site down.
Unily's CI/CD pipeline
)
)
Value
Reduced downtime from 1 hour to 9.5 minutes
Reducing downtime was a main goal for Unily when automating its deployment pipelines. Before using Octopus, downtime was approximately 1 hour for customers, in some cases much more than an hour. The team also often worked on weekends to reduce the impact on their customers.
Unily reduced its downtime in stages, initially reducing it to an average of 47 minutes. Now, having completely automated its pipelines, downtime averages only 9.5 minutes, and the team have also reduced their failure rate to 1.2%.
Our downtime improved so dramatically that customers contacted their customer experience officers to check their downtime was correct.
3,000 monthly deployments take an average of 25 minutes
Before using Octopus, Unily’s deployments were manual and took hours to complete. Because of this, the team only deployed around 200 times a month. As they scaled, they needed a solution to increase their deployment speed and frequency. Now, they deploy 3,000 times a month, which takes only 25 minutes on average.
I now know if we need to deploy something, we can do it every time. We only need a good cup of coffee and some patience, and Octopus does the rest. It’s been a winning story for us, and it’s mind blowing to see how far we’ve come.
Automating self-service upgrades for customers with ServiceNow
Unily uses ServiceNow to manage customers' software versions. Octopus lets the team manage complex deployments to customers by integrating with ServiceNow and their homegrown automation tool. Customers book upgrades at a time that suits them, and using ServiceNow, their deployment gets triggered in Octopus. Octopus provides the details of the deployment time stamps and closes the deployment in ServiceNow when it’s complete. This provides guardrails for deployments that aren’t entirely automated by checking upgrade versions against the production version and requiring an ID from ServiceNow before deploying. The team has automated processes while still providing flexibility to customers.
ServiceNow is our client service profile, so having Octopus integrate with our own automation makes the entire process seamless.
10x increase in deployments without growing the team
As Unily scaled, it had to consider hiring more staff for its manual and time-intensive processes. The customer success team also spent a lot of time scheduling upgrades.
With Octopus, Unily has automated its deployments to reduce unnecessary work for its team. They’ve increased deployments without increasing team size. The team spends less time on deployments and more on important work. The team has also upskilled in new areas.
With technology constantly changing, it’s very important to adapt. I wanted to automate our processes and get engineers away from pushing buttons. We then taught them to upskill and focus on new areas.
Reduced virtual machine costs by $40,000 a month
The team at Unily uses Octopus’s Triggers feature to automate deployment tasks to services for when Virtual Machines (VMs) are shutdown. They use Azure Automation to turn VMs on and off to increase efficiency and run them only when necessary. This process has saved Unily around $40,000 a month on its VMs. Sometimes service deployments must happen while some VMs may be turned off. By using the triggers in Octopus Deploy, services are automatically updated on the VM when it is turned on again.
The cost of our infrastructure is substantial, so saving $40,000 a month initially on our VMs is a great saving for us.