Innovations in DCIM: How Will Blockchain and AI Change the Game?
by Josh Anderson
DULLES, VA – Data Center Infrastructure Management (DCIM) is a foundational, if dynamic, element of any Data Center Summit held by CapRE, and that’s why CapRE’s 2019 Data Center & Cloud Infrastructure Forecast East featured a panel titled Innovations in DCIM: How AI, IoT & Blockchain Are Changing the Operations Game. This featured a snapshot into the state of DCIM – which is changing faster than any single discussion can measure. But that didn’t stop six esteemed panelists from blowing the topic wide open.
“In the previous discussion, they were talking about workloads being pushed to the Edge,” began Moderator Patrick South, Vice President of Development at the Chamber of Digital Commerce. “So if you look at the oscillating models of compute – and where we started over the past few decades — such as the centralized construct that we had with mainframes, moving to the decentralized architecture in client-server, back to a decentralized architecture with the Cloud, and now as we being to push out to the Edge, and we look back to a decentralized construct, how does that impact the kind of decentralized technologies that we are talking about today?”
South then asked another question. “With the emergence of autonomous vehicles, and AI, what are the means they’ll have at the edge for these “mini” data centers if you will?” he added. “The stream that IoT and IIoT will require will certainly have an impact on decentralization. And blockchain in particular. Blockchain is a tricky subject. It’s a bit bifurcated. If you look at crypto-currency on the one hand and the enterprise Blockchain on the other, those are some of the things we’ll talk about today.”
“Across the panel, include decentralized compute and storage capacity, which can be leveraged from unused capacity that individuals and enterprises have today, how blockchain can be used for data integrity (and we see that being used in the defense space today),” summed up South, before handing off the microphone to Mr. Alex Jacome, Chief Executive Officer of Achievion Solutions. “So with that, I’ll turn it over to one of our panelists, who probably has the hardest question today. Can you provide us with a brief overview of blockchain in AI and the spectrum that exists in AI?”
“When I try to explain the concept of blockchain I usually like to use this analogy,” began Jacome, first delving into the foundation of crypto-currency. “Think of Blockchain as a distributed, secure and immutable database. So what does that mean? Imagine that you have a database sitting on a single server. Now replicate this database across thousands of servers and nodes throughout the entire world. On top of that, make sure that only the person who added the records to the database is able to see it and read from it. and finally make sure that all of the new records can be added to the data base. Old records cannot be deleted. This gives you the basic foundation of the blockchain.”
Next, he explained that there are two major types of blockchain out there – public and private Blockchains. “Public blockchains are the ones that sit on a computer and belong to regular people like you and me. A private blockchain is also called an enterprise blockchain,” he listed. “Those are the Blockchains typically deployed to the IT infrastructure of a company or enterprise. The question is – how do these decentralized infrastructures work in coordination, together?”
And according to Jacome, Blockchain’s answer to this is the consensus algorithm, of which there are several types. “The most popular one is called proof-of-work. What does that mean?” he asked the gathering of a few hundred conference attendees. “It means that each of those nodes, which runs an instance of the Blockchain, has to institute a complex mathematical equation. Whoever finds the right answer first wins the right to add the nest block to the blockchain. But the nature of those mathematical tests are fairly resource-intensive. All of the nodes, all of the computers on the Blockchain, have to have very powerful GPUs or very powerful CPU capabilities. So that’s one of the business advantages for the data center. That’s Blockchain in a nutshell.”
Next, Jacome switched over to artificial intelligence. “From an AI perspective, the first mention of AI was back in 1956. This is not a brand new technology. It’s been with us for over 60 years,” he shared. “However, since it was mentioned in the industry, there have been two AI “Winters.” The definition of that is a period of declined investment, research activity in the field. The first was in the 1970s and the second was in the 1990s. The current period we’re in is called the AI Summer and the reason is that by the late 2000s, early 2010s, the industry and overall world has accumulated lots of big data. And AI is very data hungry.”
“So now in the 2010s, we’re seeing a lot more powerful machines than we had in the 1970s and 1990s, and obviously the advancement in AI research,” he continued. “When we work with AI applications, we like to break them down into four major layers. Though it sounds straightforward, AI has a lot of aspects to it.”
With that, Jacome listed all four. “The first foundational layer is usually called AI and Machine Learning Infrastructure – this is where your AI-optimized infrastructure, servers, networks, hardware is located. The second layer is called AI-enabling technologies. these are technologies like advance robotics, machine vision,” he intimated. “The third is the horizontal applications – the pieces of software that can be applied to any kind of business or organization. Finally, the fourth and biggest layer is what we call industry-specific vertical AI applications. This is the biggest layer out of all four and the majority of advancements, of new projects, fall into this layer.”