Supporting a life-saving humanitarian organisation to migrate Confluence from Server to Data Center.
Discover how we helped The HALO Trust improve its Atlassian instance with Quickstarts and AWS tools.
Requirements for moving from Server to Data Center with AWS
- Upgrade HALO’s single instance of Confluence Server to the latest version.
- Build two new AWS EC2 machines to properly host the application and database servers.
- Set up and run a sandbox environment for migration testing.
- Upgrade Confluence Server licence to Data Center licence and relocate infrastructure.
- Industry: Humanitarian NGO
- Number of employees: 11,000
- Tools and services used: Quickstarts deployment of Atlassian Confluence on AWS Data Center, and various Amazon services including EC2 Auto-Scaling Groups, EC2 instances, NAT Gateways, PostgreSQL RDS databases, EFS Filesystems, and Amazon Cloudwatch metrics.
Results at a glance
HALO benefited from our AWS experience and the expertise of the wider Adaptavist Group in ensuring a successful migration.
A dual-node environment deployed using Atlassian Quickstarts on AWS Data Center.
Increased reliability and performance of their Confluence instances and related apps.
Infrastructure that can scale to meet peak demand.
A more secure environment, with persistent data held separately from the application server.
A file storage system that grows to suit their needs.
About the customer
Founded in 1988, The HALO Trust is a humanitarian NGO, involved in the clearance of landmines and other explosive devices that have been left behind by conflicts. HALO stands for Hazardous Area Life-support Organization. Headquartered in Scotland, it currently has offices in England, Washington DC, and The Hague, with operations in 28 countries and around 11,000 staff worldwide.
The challenge
Initially, HALO had set up Atlassian's corporate wiki tool—Confluence—to benefit a few teams, explained Patryk Glok, HALO's Deputy Head of Information Communication Technology: 'Over the years, that pet project grew into a mission-critical system, but it was running on rickety infrastructure.'
That infrastructure hosted the Confluence Server application, which is sunsetting in February 2024, meaning that HALO’s team would soon face a higher risk of data breaches and security vulnerabilities, with no tech support, security updates, or bug fixes. As an organisation handling sensitive data about conflict zones, being open to these vulnerabilities for any amount of time was particularly problematic.
HALO had made the decision to move to Data Center, using AWS, and needed a company to help with migration, making the transition as smooth as possible. 'We knew what we wanted, but we didn't know the process, how to handle licensing, or what the infrastructure requirements would be,' said Patryk. 'And we didn't have the capacity internally to do much about it—the expertise or the manpower.'
The client wanted to get the most out of Confluence, so teams wouldn't be slowed down when everyone wanted to work on it. They needed more resilience too—so if one instance went down, another one would be spun up, and users' work wouldn't be affected.
Adaptavist is an AWS Advanced Consulting Partner
Whether you’re just starting out and want migration support or need expert advice leveraging AWS to meet your growth strategy and business goals, we’re here to help.
The process
HALO looked at several providers, and Adaptavist was number one on its list. 'Adaptavist is a massive multinational with lots of experience and resources ready to be allocated,' said Patryk.
Our first step was a series of workshops to understand HALO's existing infrastructure and what they wanted going forward. With 756 local users across 25 groups and 30 spaces, the version of Confluence they were running wasn't up to date (and on a trial server), so we coached them through upgrading their instance and setup changes to make it more secure and migration compatible.
Their Server licence was coming to an end and, given the sunsetting of Server, there was a concern they wouldn't be able to extend it long enough for the migration to take place. We used our partnership with Atlassian to ensure this wasn't an issue and the migration didn't have to be rushed.
Then we needed to get them off Server and into the AWS Data Center environment. HALO didn't have a sandbox, so we supported them to get that up and running too. This meant we could work on any issue we found during migration and present it in a live environment.
'Instead of forcing the initial migration across the line, we went through a series of three live migrations,' said Rich Blunt, Technical Consultant at Adaptavist. 'As much as you can test it in the sandbox, production constantly evolves. We brought in many resources from across the business to try to resolve any issues that arose.'
Crucially, at all times throughout the process, we were really open with the client. 'We didn't shy away, letting Patryk know where stuff was at, providing information, keeping him abreast of the problems,' said Rich. 'That's standard practice. With a small company, the responsibility is landing on one person's shoulders, so we made sure that communication was there.'
'During migration, we encountered issues unseen elsewhere,' said Patryk. 'We wouldn't have been able to get through those without the massive amount of help from the Adaptavist team. They weren't willing just to hand the solution over in a half-baked state, which they could have done.'
The solution
We proposed a highly available environment, to be deployed using Atlassian Quickstarts—these are Atlassian-maintained templates for creating standard Confluence infrastructure on AWS. The bastion host on a public network allows an administrator to SSH from it to an application instance on a non-public network.
In production and development environments in AWS, we deployed Auto-Scaling Groups, EC2 instances, NAT Gateways, PostgreSQL RDS databases, EFS Filesystems, and Amazon Cloudwatch metrics, along with all interconnecting resources. Everything was deployed using Quickstarts.
During migration, there was a drift between the deployed resources and the Cloudformation parameters. We updated the Cloudformation FQDN and Instance Type parameters, but this caused a couple of knock-on effects.
Ultimately, to resolve the issues with drift, we removed the bastion server resources and disabled that feature in the Cloudformation templates. Then we terminated all application instances in order to resolve a remaining Cloudformation dependency conflict. After this, the FQDN and instance type parameters were set to their correct values, and application instances were created—and they’re now performing as desired.
Finding smart fixes
At the point where we got them up and running on one node, teams noticed that URLs were linking to their old website. We were able to lean on the expertise of The Adaptavist Group's ScriptRunner team to produce a custom script that updated all their links and got them working properly. 'That would have taken us weeks and months to track down manually,' said Patryk. 'So that's a very good example of Adaptavist going above and beyond.'
Right at the end of the project, when we were making these final adjustments, Confluence discovered a major vulnerability. Luckily, we could incorporate the security patch into HALO's deployment to get that fix. It meant their instance was blocked off from the internet for a week. But without that fix, their instance could have been vulnerable for a long time.
"We wouldn't have been able to achieve where we're at today without the expertise from Adaptavist."
The results
'With Quickstarts, the servers you're using are part of an application load balancer,' explained Chris Sanders, DevOps Engineer at Adaptavist. 'It has the capability of spinning up new instances if the demand increases, or reducing the number if the demand goes down. The customer can specify how many they want, no matter what. And all their persistent data is stored somewhere else.'
For HALO, this meant when a high volume of people start work at the same time, we could schedule spikes to handle the increased traffic. It also gives the organisation more resilience. With two instances running at all times, if one has a problem, a new one will be brought up in its place—and it's completely seamless and invisible to the user.
Our client also benefits from AWS Relationship Database Service (RDS), which means the organisation's database is separate from the application server. So no matter what happens to the instance, the database stays secure. And with Amazon's Elastic File System (EFS), The HALO Trust now has a file storage system that will grow as big as it needs, without worrying about no-more-disk-space alerts.
'The average user isn't going to notice anything outstanding,' said Patryk. 'But for my team, it's the peace of mind of knowing that users will be none the wiser if one node goes down. And if we'd stayed on Server, we would have been in a really bad spot. Now we don't have to worry about patching for quite some time, so we can decide on the next steps.'
About Adaptavist
Adaptavist, part of The Adaptavist Group, is a global technology and innovative solutions provider, helping organisations boost agility and overcome the challenges of digital transformation. Five key practices shape our offerings: agile, DevOps, ITSM, cloud, and work management. Our global team spans over 1,000 employees, with a 22,000+ customer base representing more than half the Fortune 500. We are an Atlassian Platinum Solution Partner, and an AWS Advanced Consulting Partner.