Since it finished its cloud migration, Bendigo has focused on improving its architecture to increase cost savings and improve the performance and availability of its digital offerings.

Balancing performance, Bendigo started running its nonproduction workloads on Amazon EC2 Spot Instances, which let customers take advantage of unused Amazon EC2 capacity in the cloud.

The bank also achieved resiliency and scalability by implementing other AWS features for fault-tolerant workloads.

Bendigo claims it has reduced its compute costs by approximately 60%, increased the resiliency of its workloads by nearly 30%, and improved the performance of its banking system by roughly 20%.

Meeting open-banking regulations by migrating to the cloud on AWS
Bendigo holds over $83.4 billion in deposits, and it serves over 110,000 shareholders while employing over 7,000 people across 317 branches.

In May 2018, the Australian Government passed the Consumer Data Right legislation, which required all major financial institutions to comply with open-banking practices by July 2020.

This legislation also meant that financial institutions must provide their customers with access to and control over their personal data. These industry changes increased demand on Bendigo’s on-premises infrastructure. To face this, Bendigo underwent a digital transformation on AWS.

Bendigo chose AWS as its cloud service provider because of the maturity of its services, and in March 2020, it kicked off its cloud migration using Amazon EC2.

The bank also began running its workloads using Amazon EMR, a cloud big data solution for running large-scale distributed data processing jobs, interactive SQL queries, and machine learning applications using open-source analytics frameworks.

To meet open-banking mandates, the company configured clusters that would help it maintain a availability across its banking systems.

After undergoing such drastic change, Bendigo wanted to improve its cloud architecture and reduce compute costs by running its clusters using Amazon EC2 instance fleets.

"Cost savings wasn't something that we were thinking about initially," says Bendigo development infrastructure engineer Adam Hobbs. "But we realised that we wanted to improve the performance of our architecture."

In 2021, the Bendigo team entered the second phase of its digital transformation journey, and the team began experimenting to meet the bank's computing needs.

Using Amazon EC2 spot instances to reduce compute costs
Bendigo engaged with the AWS team to test Amazon EC2 instances at a lower price.

Because Bendigo's developers use its nonproduction environment for iterating features for its digital offerings, the bank identified this as an area where it could reduce compute costs.

Bendigo says the bank runs its nonproduction workloads entirely on Spot Instances.

"Amazon EC2 Spot Instances have been an effective way for us to save on compute costs while maintaining an elastic workload," says Bendigo data lead for open banking project Leandro Silva.

The bank also tested different Amazon EC2 instances for running its Amazon EMR clusters.

By diversifying its instance types, the bank has reduced the likelihood of service interruptions and downtime and made its workloads more fault-tolerant.

"We went through stages of configuring different instance types," says Hobbs. "We spent some time with the AWS team identifying which instance types have the highest availability, which has delivered a much better experience."

Bendigo's Amazon EMR clusters consist of different node types, which use Amazon EC2 instances for cluster capacity planning. "We have been setting up core nodes," says Hobbs. "We went from a 5-node cluster to a 30-node cluster."

To facilitate cluster scaling, the bank implemented task nodes, which it can spin up or spin down to increase capacity as needed and save on compute costs.

Since then, the company has grown to use 60 core nodes, and with this, Bendigo's Amazon EMR clusters can use other available task nodes if one fails, minimising service interruptions and downtime.

The bank also started using Managed Scaling for Amazon EMR, which automatically resizes its clusters.

Using Managed Scaling, the company can manage its compute costs more by setting up minimum and maximum limits for its compute capacity.

"We gain deeper insights about our workloads, which helps us monitor our usage and true costs," says Bendigo service owner cloud platform team Ash Austin.

"Working on the cloud helps us enhance our risk and vulnerability management in a highly secure, robust, and controlled environment.”

Building a data lake to enhance machine learning capabilities
Bendigo is modernising its core banking system on AWS to unify its branches and use a consolidated banking system to provide transactional updates in real time.

The bank is also building a data lake using Amazon Simple Storage Service (Amazon S3), an object storage service. By storing its unstructured data in the cloud, the bank is laying the foundation for integrating machine learning and artificial intelligence alongside its digital offerings.