Amaysim addresses 'tech cost' of growth spurt

Amaysim’s rise from zero to over one million customers in six years has come at “some kind of technology cost”, but one the telco is righting with constant re-evaluation, cloud and a gravitation towards a microservices architecture.

Head of IT operations Peter James told iTnews Amaysim’s application stack currently consists of “a combination of an older monolith architecture as well as a newer microservices architecture that we’re gradually transitioning to".

Though the company has only been around a relatively short time, it has some monolithic elements from its early days as a telecommunications start-up.

“We’ve always been really keen on providing the best user experience for our customers,” James said.

“Especially early on, we partnered really heavily with a bunch of organisations [for the IT and application stack] to enable us to focus on experience.

“At the time we didn’t have engineers or developers, and we didn’t have an operations team at certain times, so ... we took [architectural] decisions to get to market quickly and to have a stable product.”

In the years since then, however, Amaysim has become a major player in Australia’s telecommunications space, both as a mobile virtual network operator (MVNO) and now as a reseller of the national broadband network.

“Obviously over time we’ve had quite substantial growth,” James said.

“That has come at some kind of technology cost in that we need to constantly re-evaluate what we’re doing.”

Cloud shift

Three years ago - just prior to James’ appointment - Amaysim took an axe to most of its on-premises infrastructure.

“We were heavily invested in physical infrastructure, it was not doing what we needed it to, and it wasn’t scaling with the responsibilities that we had as we grew,” James said.

The company decided on a “lift and shift” of the majority of its production systems into AWS.

“We have a very tiny amount of on-prem software that we have to use to talk to one of our key partners,” James said.

“Everything else is in AWS, from the sales funnel to processing core data files to self-service to the way we interact with some of our third parties.

“Obviously we resell [services on] the Optus 4G network so the way we interact with Optus is all on AWS infrastructure.”

The telco remains heavily invested in the AWS ecosystem. James said the configuration had been resilient to “all but the most critical of AWS failures”, and that Amaysim also liked the pace at which AWS launched new features, which outran the telco’s ability to adopt them.

“We’ve had a look at multi-cloud in the past and the conversation probably comes up every 6-12 months,” he said. “Right now though we’re very AWS-centric.”

Moving into microservices

Next on the table is a gradual shift in the way Amaysim architects its applications and workloads. Like many others, it is adopting a microservices architecture where applications are broken into a series of small, often standard, services or components.

The shift to a microservices architecture will, however, be gradual.

“We don’t want to stop selling SIMs for three months while we go to a complete microservices architecture,” James said.

“It’s always going to be done with the underlying business priorities [in mind].”

That means the first applications and workloads to use a microservices approach are largely new ones.

“Basically everything we do from now on is [in what] we call it the microverse,” James said.

“We basically say we’re going ‘microverse first’ - which is our internal name for going microservices first for anything new.

“So, for example, the broadband funnel for customers to buy NBN from us is pretty much a microservices approach.”

Amaysim is also hoping to replace some of its monolithic architecture but that is likely to occur at a slower pace.

“We’re gradually looking at key services or functions in the monolith which we can gradually peel off and change to a microservice, but we’re not running a complete re-architecture program,” James said.

Expanding original tech usage

Not everything from Amaysim’s early days of IT is set for replacement or re-architecting.

The telco is a long-time user of New Relic software for performance monitoring on its applications and infrastructure, and its use cases have slowly expanded.

“It’s become a tool that we’ve started to use more and more,” James said.

“The first iteration at Amaysim was before we had an operations team. It was just monitoring a couple of front-end web servers.

“It wasn’t owned by anyone in the [IT] team - we just knew we had it - and I think it probably wasn’t tied to any alerting platform.

“But as we operationally matured we started looking around at different tools and solutions.

“As we moved into AWS and really focusing on building a DevOps culture, we started to look at changing the way we actually improved and monitored our software.”

Amaysim has championed customer self-service since its early days, and as it grew, James said he wanted IT to be able to maintain focus on improving the end customer experience. The use of New Relic tools increased to automate more system administration and monitoring tasks.

Usage has continued to expand. In addition to the standard application performance monitoring and management (APM), Amaysim uses New Relic’s Synthetics and Insights products to gain deeper visibility of its own stack, as well as what is happening upstream in its partners networks, which could have a flow-on effect.

“There’s been several times - as recently as a few days ago - where we’ve noticed something going wrong with one of our partners [networks] before they did,” James said.

He uses Insights to monitor real-time impacts on IT assets.

“For example, we can look at stats for how many times the help pages get hit if there’s something that’s been reported in the media,” James said.