No latency here: Improving performance, scalability and cost with a future proof cloud migration strategy.

Planning for a future proof AWS Cloud Migration

When it comes to meeting their high performing business needs, speed, flexibility and being future focussed for Racing and Wagering Western Australia (RWWA)’s data system is paramount. Their TAB wagering system processes a high volume of data requests across Australia and to keep up with the growing demand from their users the team needed it to work faster than ever without any latency. There was also the requirement to increase or decrease usage (and cost) based on need as well as a subsidiary consideration to review current licensing limitations and expenditure with their existing Oracle database and identify any additional opportunities where possible. Recognising this, RWWA developed a cloud migration strategy for moving to AWS cloud while retaining their Oracle systems and making them perform well in AWS without attracting additional Oracle investment. This is where Blue Crystal Solutions came in.

“We came to this project with some immediate needs around usage scalability, decreasing operational costs, and increasing data processing speed for our national wagering platform. Then there was the long-term posturing around diversifying our thinking without being tied to a solely on-prem system,” says Grey Properjohn, Head of Technology, RWWA.

Holding onto an expensive legacy within Oracle

Although the RWWA team were happy with their current Oracle databases, the motivation for change didn’t originate from a lack of capability, performance, or support; it came from the need to evolve their systems to best cater to fluctuating demands.

“Our immediate challenge was to keep RWWA operating within their current license constraints without having to invest more in Oracle,” says Daniel Cox, Chief Technology Officer, Blue Crystal Solutions. “We then needed to make this high-performance system run in AWS with half the CPUs to remain compliant with license restrictions in the public cloud. Latency was also a risk when leaving some components on premise. The key to making Oracle work was understanding the workload, tuning to reduce workload and ultimately design an AWS solution to suit. We also had to factor in what on premise applications (Oracle and others) had to stay and move, and the impacts of doing that.”

RWWA was seeking the flexibility to employ different technology solutions. Sure, this came back to cost but, beyond reducing licensing, it also meant scalability. In their world, there’s a big difference between the first Tuesday in November (aka Melbourne Cup Day) to the first Tuesday in December.

Melbourne Cup versus any other Tuesday

It’s not just Melbourne Cup; the pressure on RWWA’s TAB betting and wagering system can fluctuate massively throughout the year, but it’s somewhat predictable based on global sporting events. Despite this foresight, the current on-premise data centres – which were handling up to seven terabytes of wagering and customer data at any one time – didn’t allow for any kind of usage scale. This meant they paid the same during Spring Carnival as they did during any other, quieter, times of the year, and given the recent growth in its digital channels, continuity like this is not feasible.

“RWWA needed a cloud-based system that could scale up and down as the needs arise, rather than being stuck within a system that caters to their busiest time of year, all year.” Daniel said.

Paying the same premium all year round was something RWWA was keen to change but it all had to work without any latency. In fact, they wanted the data processing time to improve.

“Moving to AWS’s consumption-based model meant that we could meet increased demand but didn’t have to consistently operate at that level when demand was low,” says Properjohn. “This dynamic resourcing could clearly drive cost efficiencies, but the real question was, how could we mitigate latency introduced by the distance from our on-prem datacentres in Perth and the AWS Sydney region?”

Proving the latency risk

It was the first time RWWA had attempted moving Oracle workloads to AWS cloud; therefore, they needed a partner who understood both platforms and had developed the relevant pre-cloud migration tools which could give them a clear roadmap for future cloud migration. This clarity began through an analysis stage, then a proof of concept (POC) once there was a potential design.

To begin with, BCS utilised their data-driven insights tool, Primesize, to gather statistical data across RWWA’s environment, including workloads during the Melbourne Cup and provided a Target Architecture Report.

A POC was then meticulously designed to make sure that the data needs could be maintained in the cloud environment and test the latency risk back to the on-premise systems.

“RWWA is one of the few companies that understand the importance of latency, which is especially important for them when it comes to betting and wagering transactions, as any latency, even to the millisecond, can cause huge problems,” Daniel said.

BCS also played a part in upscaling their cloud migration strategy as the scope had broadened. Additional systems had to be included in the cloud migration project and considerations around the geographical distance between systems in WA and NSW and the impact it has on latency. Working together, they designed a target environment that catered for the requirement to reduce latency and is also future ready.

Beyond the latency risk, the POC was also focused on understanding how to correctly configure AWS EC2 for Oracle. This was all about matching the performance of the equivalent Oracle on-premises environment and demonstrating scalability for expected future growth in demand.

As Oracle specialists, BCS were able to test latency concerns and demonstrate scalability; then came the launch.