Logicalis urges businesses to build for AI, not just refresh infrastructure
Organisations need to rethink architecture, governance and security rather than simply refresh ageing infrastructure.
Many organisations have deferred infrastructure investment since COVID but as AI adoption becomes a business imperative it’s now overdue, according to Peter Cardassis, technical services director, Logicalis.
“Infrastructure is really key to setting up the foundational layer for everything that is coming with AI and how we’re revolutionising organisations,” Cardassis said.
He advises against simply performing another traditional infrastructure refresh. Instead, they should use this point in time to reconsider the fundaments — architecture, governance, security, operational models and future workloads.
Looking at the bigger picture, organisations are facing decisions about infrastructure investment, legacy systems reaching or passing extended support, decisions over whether to refresh, modernise or migrate applications, increasing demand for AI and growing cybersecurity threats.
“Now’s the time to make that decision — to be bold — but rather than just refreshing, it’s about setting your organisation up for the future,” he said, speaking at this week’s Dell Technologies Forum in Sydney.
Cardassis explained that organisations still need to support large legacy applications that remain critical to the business. AI infrastructure, therefore, needs to coexist with existing environments rather than assuming everything can simply be replaced.
“That means being able to run those large legacy monolithic applications that still run huge parts of the business, while at the same time having AI-ready infrastructure to take on new workloads and new ways of working with all of your staff using AI,” he said.
As AI adoption expands, the overall technology footprint and management requirements also grow while at the same time, cyber threats are rising and resilience expectations are higher than ever.
“AI still has to sit on something, and that ‘something’ has to be managed, secured and governed,” he said.
AI is also challenging the previous trend towards centralisation in the data centre or cloud. In practice, this means that as datasets become larger, data itself becomes a form of gravity influencing where applications should run.
AI infrastructure needs to be considered across the entire environment — from endpoint devices to branch offices, edge, data centre and cloud. Architecture needs to determine where workloads and data should be processed based on the value and requirements of each use case, he noted.
In many cases, data will need to be processed at the edge and used where it’s most likely to give you the best value.
“Organisations should consider processing data where it is generated rather than constantly moving it between the edge and central infrastructure,” he explained.