Our customer is an umbrella insurance company — they own and manage several brands that provide insurance services across many markets within and outside of Australia.
The organisation has a number of large on-premises datacentre environments, with all the regular challenges of building, deploying, and operating reliable systems in a self-hosted situation. While they also had a mature and effective cloud presence in AWS, it was more closely aligned with customer-facing components of the business, such as websites, customer portals, and the like.
Our customer’s central technology division wanted to be able to take advantage of the speed, performance, repeatability and security of the cloud with AWS as applied to backend systems and components which, to date, had not been considered as appropriate for running in a cloud environment.
Much of the challenge with the current on-premises environments related to the lack of visibility and control over key aspects, like security, networking, and identity management; all of which had been outsourced to various vendors, and provided in a contractually-directed, time-constrained manner. As an example, firewall changes happened twice a week and the vendor was not obligated to validate that the changes made actually implemented the changes requested.
Successful change implementation could therefore take several attempts, leading to weeks of waiting for rework, while requiring significant troubleshooting and verification efforts from staff.
Our customer already knew that they wanted to investigate migrating workloads from their on-premises datacentres to AWS, but first needed to find out what the cost and effort would be, as well as identifying which aspects of their current system would need to be changed in order to take full advantage of operating in the cloud.
After investigating the requirements, they decided to engage AWS’ Professional Services (ProServe) team, specifically because of their “30in50” migration program which aims to migrate 30 workloads in 50 days.
While it’s understood that this is a significant challenge, especially in an organisation that could sometimes take in excess of 50 days to release a change to a single workload, the goal of the program is to deliver as many workloads as possible in the elapsed time, and using the effort as a forcing function to identify those aspects of the current system that would have to change in order to build a cloud operating model that could be successful.
AWS ProServe didn’t have sufficient capacity to assist with all the details of the migration approach; and so the search began for a suitably-qualified partner who could take on the work of identifying workloads, designing solutions, implementing them, documenting the challenges, and bringing internal teams up to speed.
As a long-time AWS Advanced Consulting Partner, Cevo were well-placed to provide the technical capability and capacity to help with the 30in50 migration. In addition, our prior work with the customer has allowed us to build relationships and understand their specific needs. This prior work, including the creation of a number of tools and approaches that remain current within the organisation, helped them feel comfortable that we would be able to assist in the program.
The 30in50 process itself runs at a breakneck pace. First, candidate applications and workloads are identified; they’re run through a set of checks to determine their dependencies, the type of data (and any confidentiality concerns) involved, and the availability of subject-matter experts within the team.
Next, a solution design for each workload is created. Because of our customer’s own internal process requirements, these designs had to be reviewed by the architecture team, their security team, and AWS as well as the workload owners and subject matter experts.
Then, implementation begins; as each workload is migrated, additional layers of checks, audits, and reviews are invoked to ensure that the documentation is complete, the workloads are secure and reliable, and the teams responsible for supporting them in the new world are knowledgeable and effective in keeping the systems running.
Sometimes, workloads turn out not to be suitable for a variety of reasons: data classification was probably the primary reason for rejecting the suitability of a workload, closely followed by a lack of knowledge internally as to what it was, and how it actually worked. For this reason, we had to continually search for new workloads to consider, to keep the candidate population strong.
Ultimately, 26 workloads were fully-migrated from on-premises data centres to AWS. Those workloads are now in production, and the on-premises versions have been switched off.
Several factors contributed to not getting to the full 30 workload target: the time for internal processes to adjust to the sheer pace of the process; removal of candidate workloads during the process as a result of data classification; and the customer’s desire to incorporate capability building for their staff into the process, which Cevo wanted to support to the highest level of quality and comprehensiveness.
Despite not quite reaching “30in50”, the programme was declared a success. As a result of the migration our customer has realised a number of benefits including:
- building an understanding of what needs to change in their existing processes in order to enable cloud adoption for the central technology workloads
- developing skills and knowledge amongst their central technology team (many of whom had never run a workload on AWS before, or used Git, or even looked at a build pipeline)
- refining their roadmaps for migrating from on-premises systems to cloud, including adoption of cloud-native approaches to architecture, resilience, provisioning, operations, and cost management
- reduced time to value, increased resilience, and better scalability and security for migrated workloads.