Advanced Computing in the Age of AI | Friday, August 19, 2022

Q&A with Dell’s Top Server Exec: Ashley Gorakhpurwalla 

(Source: TheCube, YouTube)

If he wasn't busy before, Ashley Gorakhpurwalla, vice president and general manager, Servers, at Dell, no doubt is even busier today, now rumors of a merger between Dell and EMC swirl among tech pundits and financial whizzes. A few weeks before these discussions hit the headlines, Gorakhpurwalla sat down with EnterpriseTech's managing editor Alison Diana to discuss the company's server initiatives, including Dell Datacenter Scalable Solutions, unveiled in August.

With Dell World on the horizon, Gorakhpurwalla teased about the potential release of more new products later this month, his preference for holding onto new solutions a little longer than his peers would like, and the evolving face of the datacenter – and Dell. Read on for the complete interview:

EnterpriseTech: It's been a couple of months now since you announced Dell Datacenter Scalable Solutions. Can you update us as to what's going on?

Ashley Gorakhpurwalla: We were pretty excited about it. I was actually talking to some of the team today about how, if they hadn't twisted my arm a little bit, we wouldn't have announced it because it's usually my way to keep things a little bit tighter to the vest relative to unique things or innovative things we're doing. But in this case, the progress I've seen over the last month [or so] has been really promising because we've had a number of customers, our own actual sales folks around the world, and partners reach out proactively and want to know more… and they've all been really receptive to tell us more, how our involvement becomes exactly what we've been talking about or want to talk about, and they've been very happy to take advantage of our products or our flexibility in this space.

I won't throw numbers out but we've had an intake process, of people coming to us wanting to know more and wanting to get involved with business arrangements that's increased quite a bit.

We're still emerging in this space, compared with our other businesses. If you think about 20 years in servers and almost eight now in [Data Center Solutions] DCS and so we've almost completed a year here in DSS and that year continues to have quarter over quarter increases at a pretty impressive rate. I still think what I said is accurate; it may happen sooner than later actually, that this will be one of our fastest-growing business units in Dell Enterprise, taking advantage both of the market growth in this area and the positive response to the changes.

As to products, we will have products. We are working on products. We have some right now that are performing real well in the labs. We've talked to just a couple of our initial early customers about them under NDA. When will we release them? It's still the variable we're working through. It's going to be this fall and I suspect we'll probably try to intersect something with Dell World if we can. It's a good location for us to be able to interact with customers, press, and analysts as well. If we have an update we'll probably not do any real update until that time.

EnterpriseTech: What has the feedback been so far? What have you seen from prospective customers?

Gorakhpurwalla: The theme we've seen has been pretty consistent and that's it spans a couple of segments that we can probably parse out a little bit to think about. One is those is customers who are building IT infrastructure but that need to scale very quickly or it needs to be able to be able to built, managed, and deployed at a scale that's above even a large enterprise. The other is high performance computing. And I know [EnterpriseTech] has a passion for that. We have a passion for that. If you think about the similarities there, it's not at the same scale as some of these mega-web-tech companies but it's got a really similar profile, which is building IT infrastructure typically for a singular purpose or app or capability but doing it at a pretty tremendous scale, having almost a project mentality of building it out around this workload or application, and then taking advantage of the scale or having a challenge of scale which is you really have the opportunity to control and optimize and tune to get the most out of what you're doing. And almost across the board, a lot of these customers really have flipped kind of over the fence on the paradigm from 'IT is keep the lights on,' 'IT is wringing a few dollars of OPEX out,' 'IT is you know things are working unless someone calls you with a trouble ticket,' over to the other side of the fence which is 'IT is a strategic weapon in our competitive landscape. We can do things faster than our peers. We can grow quicker. We can get into different markets faster. We can flat-out compete, based on how good our IT is.' So it's become a business strategy. Those cases you typically are dealing with a very different infrastructure and a different set of how to build that infrastructure.

The other variable here that people tell us is important is, 'I need to have the IT infrastructure follow the software I need to run.' It's a bit of form follows function. In this case, function follows software and so many of the grids that are being set up – HPC, clusters, standardized IT infrastructures – are being dictated by either large pools of storage, that's a fairly common theme. 'How do I build that? What do I need, some customized capability or do I need you to deliver it in a way that other people can't deliver, or I just need some optimization in the system so I hit the most out of that storage. Or compute clusters for HPC, in terms of tuning fabrics or building out certain elements of the metalware or in terms of compute?' We see a lot of folks starting to build out either Open Stack or other applications that allow cloud-like agility. They may not be using it for cloud but it's building up that agility of scale, how you manage it, the resiliency of the software layer versus, perhaps, only at the hardware layer. Those things mean customers – as a matter of fact in my inbox, which I was reading on the way over here, two customers both in this case coincidentally large financial services companies who said, 'This is what we've been asking for because we don't mind actually being on the leading edge of technology adoption and in some cases that means moving away from, even, our internal enterprise IT methodologies of wait 'til something is mature, validated, build a very standardized image, hold that image extremely stable for a very long time – as a matter of fact, as long as you can – and then only change when depreciation of the equipment has run out.' They want to move to the other side of the continuum which is, 'I just heard of another technology or software. I want to try it. And I want to try it at scale very quickly because I have the ability to do that these days where I didn't have that ability in the past.' And so we've got the ability to say, 'Okay.' You know, when we developed this platform, this technology did not even exist yet. What we can very quickly and have dedicated subject matter experts who can start to work with them and help build out a solution that brings those newer technologies to a really stable, high quality platforms.

EnterpriseTech: So enterprises can have their cake and eat it too, so to speak?

Gorakhpurwalla: Well, nothing's for free. The level of uniqueness, if you will, that somebody needs is typically going to drive the level of capability that we can deliver or the time in which we can deliver or in some cases the cost it takes we can deliver. So we always want to add that technical conversation with somebody if it's necessary. But we can all say, 'Yes. But is it the right thing to do? Can you build it to scale? What's the risk? Do you really need 99 percent of what you're asking? Can you do 80 percent for 20 percent of the cost and effort?' That's not a typical conversation around, 'Here's what we offer. I hope you enjoy it.' You have to change to, 'Here's what we can do, together perhaps, collaboratively, and this is why. This is what we know about it.'

EnterpriseTech: That's a very different conversation for Dell and the customer.

Gorakhpurwalla: When talking about high performance compute, it's a pretty difficult conversation if you say to somebody, 'I hope you figure out everything you need, and if you would give me a schedule and a bill of materials I will go and try to fulfill that for you.' Well, okay. I mean that could happen. In some cases we'd be happy to do that and deliver the best quality on time and ongoing support, but I mean there's not a ton of value in that. Where the value or differentiation or why you would come to Dell is because we'd start with the problem of, so, We're looking to build a cluster that is sized like x number of watts or maybe it's based on compute power; we're trying to simulate oil and gas reserves; we need to be able to store the data on this; it needs to be available in this case to these different locations. We have a disaster recovery plan that needs to fall into these SLAs and we've already find this to be true sometimes. And we've already made up our minds about this one piece of this puzzle and we're building around that piece of the puzzle – it was already funded, it was already given, we already have it – and so that's a much more interesting discussion. You need a different level of expertise to have that, and that's what we've built.

EnterpriseTech: Ashley, you've been leading the server business at Dell for about a year now: Is this an initiative you started or was this something that was already there when you came to the role?

Gorakhpurwalla: It's a little bit of the latter, in that we have called it slightly different things. It didn't have a fancy code-name like Strikeforce at the time when I took over, but it's been brewing for a while. You could almost say it started in the DCS days, where we started to think about how do we take the things we're learning and move them into the enterprise space. What was different maybe seven years ago when we had this conversation … was scalability of infrastructure, open source software, things like that, not all that came applicable to standard IT, IT in enterprises but there's this new emergence of web-tech or high-scale computing or high performance computing or different ways to say the same things – which is scale-out – but scale-out perhaps not at the hyperscale level where you have custom datacenters and the entire stack, so it's from there that we've been talking about how to segment, perhaps even further, so we can be focused and really, really good at it – which is to spend time focused on the scale-out solutions group.

The Dell PowerEdge 1900 was released in 2006.

The Dell PowerEdge 1900 was released in 2006.

It's been something that's been a part of our DNA for a while; we just haven't organizationally structured it as its own business until recently. It's something we've all been working on but for me, it seems like it's a standard thread. It's had continuity since I've been at Dell for 15-and-a-half years on servers. It feels like a natural progression and how we do things, which is how do we sometimes push the envelope of technology or innovation, perhaps a little bit quicker than we should sometimes. Sometimes I'm a little bit disappointed that we don't have the ramp or the uptake that we would have projected but that's okay. One of the beauties of being a private company and a company under a technology giant like Michael [Dell] is that those are okay things to try. You can fail a little bit of going too fast on some technology or being there too soon and so we take advantage of that permission.

EnterpriseTech: As a technologist, was that the most noticeable change when Dell became private?

Gorakhpurwalla: The most noticeable change and the most immediate change is just an immense sense of stability. And I think that has to do with the period before we went private, probably, was the opportunity for a lot of people to speculate on outcomes for Dell and Dell wasn't really allowed – both the company and the person – to speculate or put out too much of what was really happening, and so we really moved into this really calming effect of long-term strategy, stability, purpose, that is pretty focused from the top down, and that's the been the biggest change for the people who work here, independent of whether they're in technology, or sales, or the field, or procurement, or IT.

From a technology standpoint, what's been awesome is we are already fact-finding out five to seven years. We have really, really excellent teams in technology, technologists, who think through that, and we have to, because in some cases we're influencing the roadmaps of our silicon partners who have to think five years ahead. But the real gem of it now is our ability to invest forward toward these goals somewhat independent of what happens every 13 weeks or every quarter, in terms of the market dynamics or sales dynamics. It's easy to say that when things are going good – and the market's been going good – but I think even when we're in a position where the market's been challenged by currency fluctuations or regional issues we really haven't blinked at all and continue to invest forward toward our next generation of technology or our next generation of customer engagements. So if anything, my biggest challenge this year has been too many opportunities to chase, how to prioritize. And the second one, well I really need to spend a little more time on, we have these pretty lofty hiring goals as we invest forward and I wish I could find more talented folks in the enterprise space to join Dell.

EnterpriseTech: That's a common complaint for everyone in tech…

Gorakhpurwalla: Although it's different to [a recent conversation] you had [regarding HP layoffs]. I wish I could hire more and more competent technologists and engineers and capable people, but we're behind our goals today, but we're certainly spending a lot of time on it.

EnterpriseTech: There seems to be a lot written these days about the death of the datacenter. How valid is that statement?

Gorakhpurwalla: What's the old saying, "News of my death has been greatly exaggerated?" I think that's the case here. Think about a couple of vectors people usually take when they talk about that: Infrastructure as a Server, cloud, and software defined datacenters. But go back seven years ago, eight years ago, and people would have said, hey isn't hypervisor going to kill the infrastructure business because doesn’t' the levels of ever increasing consolidation mean this business is going to shrink enormously? And it's time to get out.

The same could be said for the demise of the datacenter, says Gorakhpurwalla.

The same could be said for the demise of the datacenter, says Gorakhpurwalla.

And I think we've seen that there's no shortage of data to store. There's no shortage of analytics to provide better insights to your business for big data. And there's certainly no better way to differentiate your business than higher performance in terms of your customer interface. So I don't see that stopping any time soon. I see that accelerating. We certainly have no shortage of opportunities to build out datacenters more and more these days. What software-defined datacenters or infrastructure as a service have shown people is that there are new ways to build their IT around pod structures or agile, standard infrastructures or methodologies that allow people to scale their businesses much easier than the old days of every year set up a budget, build a couple of projects, and then next year start all over again. That agility, whether they learned it from a little bit of experience in a public infrastructure or they have got used to it from consumerization of IT, is just creating more and more opportunity, so we certainly are not in the space of thinking the datacenter is going away.

I would say the other thing that is interesting is we recently had some results back from a survey we did. There is a move afoot of also, for various reasons – whether it be security, data being held in geographies, or the difference between on-prem management or off-prem management –people repatriating their business data or their business applications out of infrastructure as a service, public cloud, and having a hybrid methodology. There's a pretty high percentage of people who say they're considering it. There were 80 percent of the IT decision makers considering it, in trying to decide how to best run their business. If you think about somebody who has a very static load or static application, it's kind of difficult to think of infrastructure as a service or public cloud as the right choice for something like that. But if you have a very dynamic one, I think you're going to look at the peak of that load and say, 'How do I best accommodate that without having to build to the peak?' so that's where these hybrid models become very, very useful.

 

About the author: Alison Diana

Managing editor of Enterprise Technology. I've been covering tech and business for many years, for publications such as InformationWeek, Baseline Magazine, and Florida Today. A native Brit and longtime Yankees fan, I live with my husband, daughter, and two cats on the Space Coast in Florida.

Add a Comment

EnterpriseAI