Pickles Auctions in public cloud push
- 18 April, 2017 15:40
Pickles Auctions has spent decades building what its CIO Claudio Salinas describes as ‘monolith systems'. It wasn’t a good scenario for an Australian auctions and valuation organisation looking to pick up the pace when dealing with customer demands.
Salinas arrived at the organisation in 2014 and discovered a bespoke ERP environment where core finance, CRM, and inventory functionality had been built from scratch using the PowerBuilder development language.
“In this monolithic environment, a change to a simple function or update to our systems would mean full end-to-end regression testing,” Salinas told CIO Australia.
“Even though the creation of a particular function might only be a couple of days or hours’ worth of development change, regression testing would take a significant amount of time.”
“That would create downstream bugs – every time we’d fix a bug, we’d have to go back through that testing cycle. So what started as a simple change, resulted in a week’s worth of effort by the time we make our way through testing and resolving bugs and so forth.”
This prompted Salinas to initiate a transformation strategy, which has so far seen the organisation migrate around 25 per cent of its tech infrastructure to the public cloud. The foundation of this strategy has been to phase out and replace some legacy systems with Microsoft Dynamics 365 running in the Microsoft Azure cloud.
“I kicked this off by starting to understand the real key business drivers and I took the conversation away from the technology and focused it back on the business ... asking what were the key things that were hurting us,” he said.
“Being able to create marketplaces is fundamentally what Pickles is about and being able to reinvent those marketplaces and improve and enhance them is the challenge.”
Salinas said Pickles will run a hybrid cloud environment for the time being but the organisation expects to go full tilt at the public cloud, inside Azure data centres, in the future.
“We are looking to shift all workloads, our core auction and operations systems, and websites to [the public cloud]. New workloads such as some of the innovations that we are building will be based in Azure.”
These workloads include an optimised image management capture solution which will go live next month. A key product will be the roll out of 360 degree images of vehicles being sold online.
Some legacy technologies will remain in a hosted private cloud, such as those which run Pickles’ mobile apps, said Salinas.
A low-risk transformation
Salinas and his team have architected the organisation’s IT transformation so it can be completed with the least possible risk. This is achieved by replacing legacy systems in small chunks, said Salinas.
“If we had the desire to be more aggressive with timeframes and do it in 12 months … that introduces a whole bunch more risk and potential disruption to our customers in the business,” he said.
Salinas said the speed at which new functions and applications are being released in this new public cloud environment is of huge benefit to Pickles.
“Our ability to just turn services on and be part of a community and leverage things like Microsoft cognitive services – we are seeing advantages in that,” he said.
Salinas said Pickles was at the beginning of its journey taking advantage of machine learning and cognitive computing services.
“We have built some PowerBI apps and dashboards so we are leveraging some machine learning across some of the data sets and building visualisations into Power BI.
“We’ve got lots of ambitions and doing some really interesting work with our data. A real fundamental thing that we’ve been able to do now is really just leverage the data sets we have. We’ve come from an environment where we have had data locked inside applications in islands. We’ve been able to leverage a data lake and that has given us exposure to be able to cross-pollinate data sources,” he said.
Salinas said Pickles sits on the richest source of used car sales data in the country and the company wants to mix that information with other data sources to provide additional offerings to its customers.
“We are looking at releasing some data products into the marketplace that have been enabled by that technology,” he said.
Follow Byron Connolly on Twitter: @ByronConnolly