As enterprises try to deploy infrastructure to support AI applications they generally discover that the demands of this application can disrupt their architecture plans. This episode of On-Premise IT, sponsored by Pure Storage, discusses the disruptive impact of AI on the enterprise with Justin Emerson, Allyson Klein, Keith Townsend, and Stephen Foskett. Heavy duty AI processing requires specialized hardware that more resembles High-Performance Computing (HPC) than conventional enterprise IT architecture. But as more enterprise applications leverage accelerators like GPUs and DPUs, and become more disaggregated, AI starts to make more sense. Power is one key consideration, since companies are more aware of sustainability and are impacted by limited power availability in the datacenter, and efficient external storage can be a real benefit here. This is still general-purpose infrastructure but it increasingly incorporates accelerators to improve power efficiency. One issue for general purpose infrastructure is the concern over security, and enterprise AI applications will certainly benefit from broad access to a variety of enterprise data. Enterprise use of AI will require a new data infrastructure that supports the demands of AI applications but also enables data sharing and integration with AI applications.
On-Premises for Today’s Spotlight Podcast:
Pure Storage Panelist
Justin Emerson, Principal Product Manager and Technical Evangelist – FlashBlade at Pure Storage. You can connect with Justin on LinkedIn or on Twitter. Find out more about Pure Storage and FlashBlade on their website.