GPT's breakneck cloud transition

A lonely AS/400 is all that's left of GPT Group's on-premise hardware after it migrated to AWS

For an enterprise to completely ditch on-premise hardware for the cloud is a radical step. To do so in the space of a year might be considered a little unhinged.

However, according to Greg Baster and Michael Lockhart —the CIO and infrastructure manager, respectively, at GPT Group — a decision by GPT to go 'all-in' with the cloud has paid back handsomely when it comes to cost avoidance and increased agility.

Baster told the Amazon Web Services Summit in Sydney yesterday that the seeds of the transition to cloud were planted in 2012. The ASX-listed property investment group's data centre contracts were up for renewal in a year or two, and the need for a refresh of all the company's servers and storage was rapidly approaching. In examining GPT's possibilities, cloud was included as an option.

"And I'm really glad to say we included the cloud, we included Amazon in that analysis," the CIO said.

"And not only did Amazon tick all the boxes of all the factors that we looked at, but also from a cost, agility and simplicity perspective it was clear standout in the market."

The company now has all of dev, test and production in Amazon's cloud, Lockhart said. ("The only exception being an AS/400 system that we literally can't do," he said.) AWS now hosts "all our core systems, our SAP, our Hyperion environments, our Lync telephony environments, our Active Directory and all our authentication systems, our SharePoint..."

GPT worked with an AWS partner and also brought in a project manager and a solutions architect with AWS experience. AWS was also heavily involved from the business case through the design phase and after the transition, Lockhart said. His infrastructure team of seven rounded out the rest of the group working on the project.

"During the migration as we moved particular systems, we'd go back to the business with different application teams and owners and bring in their specific skills, but in general we primarily kept the whole thing in-house," Lockhart said.

As part of the project, GPT went through a WAN redesign. "A big component of that was integrating [AWS] Direct Connect into the WAN to ensure that we provided the best end-to-end throughput resulting in the best user experience that we could to make the project a success," Lockhart said.

Read more: Inside Coles' 'cloud first' policy

Keeping the transition in-house was a conscious decision because of the tight timeframes wouldn't give much time for a managed handover from a vendor, he added.

"By our internal teams managing and seeing the whole process through from start to finish, we were able to run straight from project into production."

The project was "essentially lift and shift": Taking GPT's on-premise systems and data and standing it all up in the cloud. "We didn't want to revisit our systems and re-architect or redesign them for the platform," Lockhart said.

"We took the approach that affectionately became known as the virtual forklifting."

The final clean-up from the project and decommissioning of GPT's previous data centres was completed in January this year, Lockhart said.

"We now have a very simple and easy to manage infrastructure platform that will provide the capacity, the growth, scalability, flexibility — whatever is required for GPT moving forward, as it looks to achieve its broader business strategic goals," Lockhart said.

From an opex perspective, for Q1 GPT is tracking 15 per cent ahead of where it budgeted to be.

"We think that through the year, as we continue to optimise and look to use the platform more for what it's designed for, we believe we can push that towards 20 per cent and possibly a little bit higher," Lockhart said.

"From a capex standpoint we will achieve straight of the bat in 2014 cost avoidances on the infrastructure that ... was due for refresh this year of more than what we actually outlaid to run the project last year. So that's a great result."

"The business case called for around a 50 per cent return on investment — that's including the capex avoided — which we've more than achieved," Baster said.

"We've been able to reduce the cost of technology per employee with GPT by about 35 per cent and that cost will actually reduce over time as GPT grows," the CIO said.

In addition the agility of using cloud has already been proven by the speed with which GPT has been able to spin up new services, he added.

"From our perspective it's been a fantastic journey. A tough journey in terms of the amount of work that we did in a short period, but the results are phenomenal for the company," he concluded.

Read More:

Join the Computerworld newsletter!

Error: Please check your email address.

Tags amazonAmazon Web ServicesAWSCloudcloud computing

More about Amazon Web ServicesAmazon Web ServicesSAP Australia