The long view on computing is that every computer system starts with a limited set of functions and ultimately opens up to new functionality as creative individuals use the system in unexpected ways.
Cloud computing, its own kind of computer system when looked at in aggregate, is going to experience that renaissance just as other systems have, says John Roese, the chief technology officer, products and operations, of computer giant Dell Technologies.
"If you look at the compute paradigm, and you look at a mainframe or supercomputing environment, there's an interesting thing that happens: the system starts with a limited set of components and limited optionality so that you have manageable parameters to get to a system outcome," explained Roese in an interview with ZDNet via Zoom.
"What happens, though, is as the system becomes stable, and the system gets utilized, what occurs is diversification of the creative element of what we use it for — people figure out what to do with them which nobody contemplated when we first built the system."
The result of that burst of creative use, said Roese, is "it puts stress on architectures that are too curated, and leads eventually, inevitably, to a shift to an open system architecture."
There is a parallel now with cloud, says Roese.
"Since we went through it in the computing space — we [Dell] opened up computing — if we apply that to the cloud world, to the total IT stack today, our role is effectively to make sure that happens" with cloud.
"The master plan for Dell's place in the universe," said Roese, "is not to create another upstream silo but to actually look at which layers we can start chipping away at on the infrastructure side and turn them horizontal."
What does that mean, exactly?
Roese sees multiple initiatives that Dell is undertaking that may take layers, meaning, groups of functionality, and turning them into products and services that span multiple cloud computing providers.
For instance, Project Alpine, part of the company's APEX family of products, is Dell's program to bring its file, block and object storage management software into the public cloud as a service. Alpine will let companies deploy applications across storage in different locations, including on-premise facilities. The offering was first announced in January.
"Project Alpine is interesting because we said, We do storage really well, but if we do storage and it instantiates under a proprietary Amazon or a proprietary Google or a proprietary VMware, that isn't really binding these systems together into a multi-cloud," said Roese.
"And so Alpine was all about saying, What if that storage layer was actually horizontal, what if it was compatible with them but Dell provided that?" explained Roese. That creates tension because "lots of parties make money off of storage," he observes, including the cloud providers themselves.
But it comes down to the fact that "customers' biggest challenge in multi-cloud is, I would like my data not to be coupled to what compute substrate I use," he said.
In an analogy with a computer chip, Dell's storage software, said Roese, becomes "the shared memory substrate of the multi-cloud environment," like the cache memory in a processor.
This is a way, at the same time, to leverage distinct parcels of Dell's patented technology.
"By the way, from an IPR [intellectual property rights] perspective, we own the IPR in the storage layer, I don't care what a public cloud tells you," remarked Roese. "When they have a hard storage problem, they buy our gear and put it in their environment."
Another layer that could be chipped away at is edge computing, says Roese.
"Our edge strategy is creating an environment where an edge is not a mono-edge, in fact, the edge is a platform in which you have access to all the compute of all the clouds," said Roese.
"In fact, we have in our labs systems that on the same edge cluster are running [Google] Anthos, K3s, [Microsoft Azure] Arc, Vmware — and they're just software packages!" he said, citing four major container-management suites.
The same horizontal approach has Dell partnering with every so-called data processing unit, or DPU provider, including Nvidia's Mellanox and AMD's Pensando family of chips.
Working with Snowflake and other young companies, said Roese, makes the movement to data-in-motion, including Apache Kafka and other programs, a potentially horizontal affair.
"It's not, here's some storage under Snowflake," said Roese. Instead, "Their [Snowflake's] primary storage is in AWS, Azure or Google, and we are their edge." To be that edge, in this case, means that Dell uses connectors such as Kafka "so people can copy their data directly into Snowflake as a feature of a [Dell] storage array." Likewise, he said, by using a technology called remote database tables, "the Snowflake compute can now execute against remote data on prem."
Security may be the final frontier of such horizontal effects. For Dell, the field of security is a relatively new product area. Dell's offerings include data vaults, which it provides in both public cloud and on premise, as a way to "harden" data against ransomware.
But beyond a single product, "the right play for Dell is to lean hard into acclerating and simplifying the adoption of zero trust," he said, which is "not a product but our ability to organize the ecosystem."
Zero trust is "not hard," as a concept said Roese, it is simply constant authentication. That is, however, a very hard paradigm shift to engineer, he said. Being an entity that oversees zero trust, said Roese, can be an important position spanning different cloud computing instances.
Zero trust "is a set of business controls, knowing what you want your systems to do or not do, but then it also has a separable control plane," he said of the standard zero trust systems architecture.
"That control plane is in charge of diverse infrastructure underneath it," observed Roese. "If you were going to build for the multi-cloud world, that is exactly what you would do."
The process of chipping away at layers, said Roese, is a blend of partnerships.
In the traditional cloud compute world, "Today, we have pretty good relationships with all of them," said Roese. "We have a deep relationship with IBM and Red Hat, a deep relationship with Microsoft, deep relationship with Google, a growing relationship with AWS, a fairly good relationship with VMware, and on the telco side, we work with Wind River really deeply."
In the younger area of data-in-motion such as open-source Kafka, there is a growing number of relationships with young companies, with whom the existing partnership with Snowflake may serve as a model, said Roese.
"You can imagine us doing the same thing with Datastax and Databricks," he said. "So, it's a whole collection of people building new data architectures, and our role is still subordinate and underneath them."
Trying to open things up is one reason, said Roese, that Dell divested its stake in VMware last year. Although a virtualization platform could have advantages, he said, "there was also an encumbrance by owning them," he remarked, in terms of ending up with another silo competing with the cloud providers.
"We said, in order for us to be an enabler for [cloud service providers], do we want build things specific to them, or take it horizontal?" Jettisoning VMware was a vote for horizontal development, he said.
(VMware is in the process of being acquired by chip and enterprise software conglomerate Broadcom.)
From a business standpoint, all of the horizontal efforts are meant to create new opportunities for a company that already enjoys top position in its traditional compute and storage equipment markets.
"It's this age-old question: How do you grow a $100 billion company?" said Roese, referring to Dell's total revenue in 2021.
"You can consolidate, but if you run out of addressable market, it's a problem." Hence, his job, he says, is to divine, "How can we expand into adjacencies that make sense," a process that has the dual benefits of expanding the market opportunity but also creating "a lot more diversity at the company, which is good as technology ebbs and flows."
The company has six initiatives as the moment that are of primary focus to expand Dell's reach beyond its traditional bailiwick into adjacencies. That determination comes out of what the company calls its "tech radar," which for Roese means, "every technology that comes at us, we take a look at it, we decide if it's core [to Dell], adjacent or periphery, whether it's heading toward us or away from us."
The six, along with security such as zero trust, include as-a-service products such as APEX; edge computing, broadly the move away from centralized data centers; telecom technology, a new area for Dell where it sells to phone companies, called the Telecom Systems Business; artificial intelligence and machine learning; and data management including the real-time data-in-motion field that's emerging.
"The punchline" of those six initiatives, said Roese, is to "take the original TAM [total addressable market] of the company," which had been six hundred billion, and turn it into the current total addressable market of $1.4 trillion.
To the age-old question of business growth, said Roese, that larger opportunity has the effect of "expanding the surface area for Dell to grow."
Just as important to the business, said Roese, is that the six areas of focus, and the approach to them, sets Dell apart from the approach taken by competitors.
"At a macro level, we, kind-of, are emerging as doing what others aren't doing," he said.
"HP [Hewlett Packard Enterprise] is, kind-of, in a different vector, they're trying to be VMware, I think; Cisco can't afford to do it because you have to be able to play the commodity game; Huawei can, but they're crippled right now; the public clouds definitely don't want to do all the stuff I've talked about, like putting IT into the real world."
As a result, "I'm cautiously optimistic that we're playing a very meaningful role to pull things together, but we're doing it not just as an integrator but owning layers of the technology stack," said Roese.
"That makes us a bit of an odd duck, but we're a $101 billion odd duck that seems to have a lot of value and is, kind-of, providing more and more foundational technology."