Hardware Offloads, Not Everything Is X86

Alastair Cooke takes a look at why x86 isn’t the only option anymore for organizations looking to solve computing problems. This trend has been going on for a while, with GPUs finding new roles handling highly parallelized workloads in ML and AI. But he also highlights how NGD Systems is putting ARM cores inside SSDs to provide computation storage. He’s also looking forward to hearing about the Pensando Distributed Services Card at Cloud Field Day, which appears to offer hardware offloading for software-defined edge services, aimed at telcos. We can’t wait to hear what he thinks of their presentation!


72: GreyBeards talk Computational Storage with Scott Shadley, VP Marketing NGD Systems

This episode of the Greybeards on Storage podcast features a conversation with NGD Systems’ VP of Marketing, Scott Shadley, who presented at Storage Field Day earlier this year. They talk about their smart SDDs, which will become generally available by the end of 2018. These SSDs each have a 4-core ARM processor running Ubuntu. This allows any workload that can run on Ubuntu to be performed directly on the storage layer. They look at what these drives are capable of, and how they could be deployed by organization in the near future.


EP10 – Computational Storage: A Paradigm Shift In The Storage Industry with Scott Shadley and NGD Systems

On the latest episode of the TECHunplugged Podcast, hosts Max Mortillaro and Arjan Timmerman discussed computational storage with Scott Shadley of NGD Systems. This followed NGD Systems’ presentation on the subject at Storage Field Day last month.


NGD Systems from #SFD17

Aaron Strong got a look at how NGD System’s computational storage solutions hope to deal with the coming deluge of large data sets. At Storage Field Day, they showed how putting ARM CPUs directly on the storage layer, combined with their In-Situ suite to expose the compute, can help offload and distribute workloads, saving latency and energy.


NGD Systems Help Make Cat Searches Go Faster with Better Results at SFD17

NGD Systems’ computational storage solution certainly left an impression on the Storage Field Day delegates, including Jeffrey Powers. In his post on their presentation, it looks at how moving compute to the storage layer provides for better scaling and can be useful for complex application workloads.


NGD Systems Are On The Edge Of Glory

Dan Frith got to see NGD Systems at Storage Field Day last month. In this post, he writes up his thoughts about how NGD is approaching the problem of edge computing. Their approach isn’t simply to move the compute layer to a remote location, but rather to move compute directly onto the storage layer. This provides a cheaper alternative than moving the data to a central location, with massive latency and parallelism benefits as well. You don’t have to be “Gaga” to see the benefits in that.


The Rise of Computational Storage

At Storage Field Day last month, Max Mortillaro got a deep dive into one of the emergent companies in the computational storage space, NGD Systems. They seek to move away from traditional Von Neumann architecture and process data directly on the storage layer. Each NGD System drive is equipped with an ARM processor, RAM, and an ASIC for processing specific data sets. In their testing, NGD was able to reduce the number of servers needed for a given workload to 25% when using their computational storage solution. Make sure you check out Max’s TECHunplugged Industry Insights research paper on computational storage for a more comprehensive deep dive.


Tech Stand Up Episode 9 – SFD17 – Initial thoughts NGD Systems Podcast

Luigi Danakos discusses his thoughts on the computational storage shown by NGD Systems at Storage Field Day on the most recent episode of the Tech Stand Up Podcast. Joining him are Joe Houghes and Howard Marks. They discuss how the devices network, scale, and what use cases can best use the tech.