Back in 2010, Australian enterprise software company Atlassian Corp. Plc had the usual problem of fast-growing companies: too many customers — or rather, too many customers for its information technology systems to handle well.

When Atlassian launched its first software-as-a-service offerings for its two largest products, the software issue tracker Jira and the team collaboration app Confluence, it gave every customer its own container – basically a custom virtualized server. Mike Tria, Atlassian’s head of infrastructure, said that worked great initially, but it started to run into problems with scaling to tens of thousands and then hundreds of thousands of systems that needed to be managed.

In short, when customers scaled up, Atlassian couldn’t scale up with them. That required a 2,000-user limit per customer for a lot of products released in the cloud.

Now, thanks to a just-finished project to migrate some 78,000 Jira and Confluence customers to Amazon Web Services Inc.’s cloud — 70 percent of its overall 112,000 customers in 185 countries — that problem is gone. Indeed, the Jira and Confluence teams are still trying to discover what the limit might be in the new cloud setup — “if there is such a limit to be hit,” added Tria.

Challenges and benefits

Atlassian’s journey is one that many large enterprises are making as the flexibility and economics of cloud computing sweep the information technology industry. While some large companies such as Dropbox Inc. are actually moving workloads back from the cloud to their own data centers, others such as Atlassian — and, just this week, domain-name registrar GoDaddy Inc. — are embracing cloud computing in a big way to allow their systems to scale up and release software faster.

It’s far from an easy task. For Atlassian, that was apparent as soon as the project started in the planning stages in 2015. The company code-named the program Vertigo after the team described it to one of the cloud development managers, who responded, “I really like the idea, but it’s giving me vertigo.”

But it culminated recently with the migration of all the Jira and Confluence customers in the space of a year. SiliconANGLE recently talked to Tria about what it took to migrate two major products and tens of thousands of users into the cloud — and, even more important, the benefits that came with the massive move.

Most of all, moving Jira and Confluence onto the AWS cloud has also greatly increased Atlassian’s ability to stay ahead of customer needs and roll out fixes and new functions quickly.

“Every time Jira was deployed, in the old model, the customer needed to be shut down, deploy the code, and then bring them back up,” Tria said. “The problem here is that how large the customer was would map to how long it took to shutdown and update. This meant that really big customers would have to go offline for days.”

Now, with the AWS cloud, Atlassian has moved all of the software primarily into an environment made up of stateless microservices, which are simple, single-purpose virtual machines that act on their own data and write it back when complete. That means microservices can be updated without needing to take anyone offline.

The result: Software updates that used to be done perhaps once a week can now be done two or three times a week — with improved reliability, uptime and performance. Tria told SiliconANGLE that recently, Atlassian even managed to push out nine releases in one day for Jira.

Deploying all the time

Atlassian’s Mike Tria (Photo: Atlassian)

“In the new world, we’re just deploying things all the time,” Tria said. “With microservices, we were worried we’d end up with a less reliable system to our customers, but as it turns out with the way the cloud works, we just ended up with a more reliable system.”

At the time that Project Vertigo was in its planning stages, the cloud industry had a lot of contenders beyond Amazon.com Inc.’s cloud. For example, at the time the Google Cloud Platform, Microsoft Corp.’s Azure Cloud and VMware Cloud offered their own potential solutions.

Tria said Atlassian chose AWS for two big reasons. First, the company already had some dealings with Amazon and “the number of cloud-offerings that were needed to support Jira and Confluence were very few.”

At the same time, he added, Jira was being deconstructed into microservices. “It would be impossible to ‘lift and shift’ migrate the system, it needed to be re-architected and Amazon had the numbers Atlassian needed,” he explained.

The second reason was the operation’s depth of experience. “Atlassian really needed durability and reliability above and beyond shiny things, bells and whistles,” he said.

James Kobielus, lead analyst for application development at Wikibon, a sister company to SiliconANGLE, said that AWS at the time defined and set the bar for the public cloud market, making it an obvious choice for Atlassian. In his trip report for AWS re:Invent 2017, he wrote that AWS had seen “staggering growth and runaway momentum in the global cloud market.” During that same year, its revenue rose by 42 percent, to $18 billion.

Beyond AWS

Nonetheless, Atlassian hasn’t neglected other cloud environments, which some of its customers may already use for other applications. Kobielus said in his trip report for Atlassian Summit 2017 that the company spent that year announcing even more cloud offerings, including Jira Software Data Center for Azure. “Matching the capability introduced for the AWS cloud last year, Atlassian now lets customers deploy Jira Software Data Center in Azure,” he said. “This provides Jira Software Data Center customers with high availability, disaster recovery and performance at scale with the deployment option that best fits their needs.”

So what’s next for Atlassian? “There’s always more we can do to make our cloud foundation faster and better for customers, so there will be incremental work that will continue,” Tria said. “We are also taking the learnings from this recent project to apply it to future updates we make to our cloud infrastructure.”

Yes, he’s being coy. But it’s apparent that Atlassian’s journey to the cloud isn’t over yet.

Image: Lucian Savluc/Flickr