Tenancy.co.nz provides a suite of online tools for the top real estate firms and property managers throughout New Zealand. Their tenancy application form–which we helped develop–processes over 50,000 applications per month alone.
With the need to run a complex offering of online products at scale, there are many interrelated components using multiple AWS services, across multiple environments, including testing, staging and production.
Tenancy.co.nz has seen huge growth and the hosting infrastructure and usage bills grew to unsustainable levels. So our AWS experts completed an infrastructure review in order to see how we could help.
We have worked with CustomD for a number of years and they are extremely knowledgable in what they do. Their understanding of AWS systems and architecture is incredible. Their efficiency in diagnosing issues and identifying optimisation potential is paramount to both our required up time and customer satisfaction.
Luke Siemonavich - Development Manager, Tenancy.co.nz
As a result of dealing with many different systems at Custom D, we’ve developed knowledge, tools and strategies to help effectively manage costs on AWS, while at the same time either maintaining or improving performance, reliability and security.
With this in mind, we quickly identified some big hitter changes that could be made, such as moving from on-demand to reserved instances, which drastically reduced their monthly bill by thousands.
AWS's usage-based billing, the way in which an application is programmed, and it's system architecture can all have a massive impact on monthly costs. So we also completed a technical audit of the system and underlying code, to make further suggestions with cost-saving implications.
Through the process of looking at the current infrastructure, our experts were able to consolidate different AWS services and simplify the setup, which in turn reduced cost. Examples included migrating multiple databases and environments to Aurora serverless instances, and removing unnecessary EC2 instances, Redis clusters and OpenSearch domains.
Logging & Monitoring
While looking at how we could optimise the infrastructure we discovered there were multiple CloudWatch dashboards, and a ton of unnecessary logging (500gb a month).
We reduced log levels, and went through the source code to remove non-critical informational logging and optimised reporting of errors by sending that information to sentry rather than CloudWatch alarms.
This all had a flow-on effect to reducing the monthly AWS cost.