Compressing Storage Costs: How Modern Solutions can Lower TCO

Companies are generating more data than ever, and storage costs are rising. Six Five Media hosts Mitch Lewis and Camberley Bates are joined by Broadcom Product Marketing Engineers Tom Nagelmeyer and John Nicholson, for a conversation on how modern solutions can significantly lower the Total Cost of Ownership (TCO) for storage.

Their discussion covers:

– The evolving landscape of data storage and its impact on TCO

– How Broadcom’s innovative solutions are setting new standards for storage efficiency

– Real-world examples where modern storage solutions have made a difference

– The role of software in optimizing storage costs

– Future trends in storage technology and their implications for businesses

Learn more at Broadcom.

Watch the video below at Six Five Media, and be sure to subscribe to our YouTube channel so you never miss an episode.

Transcript

Camberley Bates: Hello everyone, and welcome to Six Five On The Road. I am here with the team from Broadcom VMware, and let me introduce you to the folks that are on here that we’re going to be talking about VMware and TCO models and storage analysis, etc. So introducing first I’m going to take Tom Nagelmeyer. He is the product marketing engineer with VMware. Welcome, Tom.

Tom Nagelmeyer: Hi Camberley.

Camberley Bates: All right. And I’ve got John Nicholson, another product marketing engineer with the same group over there at VMware.

John Nicholson: Howdy, y’all.

Camberley Bates: And third, I have Mitch Lewis who is with The Futurum Group and Signal65, and he specializes in TCO and financial analysis. Welcome, Mitch.

Mitch Lewis: Hi Camberley. It’s great to be on here.

Camberley Bates: And I’m your host, Camberley Bates. I’m the Chief Technology Advisor over here at Futurum Group. And what we are doing today is looking at how do you compress storage costs, how modern solutions can lower your total cost of ownership? And this is coming from some real deep work that this team has done in terms of looking at what does it look like to deploy a large VCF cluster and support that with different types of storage systems and how do you best optimize your storage in that environment? So I want to kick off because the word TCO, ROI and a lot of other things get kicked around and sometimes it gets a little bit confusing to people. So I’m going to turn to you, Tom, if you would maybe talk about what total cost of ownership is, how does that differ from price and those kinds of things?

Tom Nagelmeyer: Sure. Yeah. So total cost of ownership and price are things that we intuitively know, but sometimes when we put it into an IT context, it can be easy to not understand. So I’ll start out with some simple examples. Most things that we buy and consume in life, we look at the price and say, “Well, that’s what it costs.” Anytime you go to the grocery store and you see something on sale there for say $2.99, some food that you eat, the price is equal to the cost. I know what my food cost is in life and there’s not a lot of total cost of ownership analysis that needs to go into that.

But when you start to look at, we say durable goods, things that have costs on top of what you pay, that’s when a total cost of ownership analysis can be helpful in helping you make a purchase decision. So a great example that everybody knows and thinks about is vehicles. Vehicles have a lot of associated costs beyond what you pay at the dealership. So once you get that car home, you have to get insurance for it, you have fuel or energy costs depending on the type of vehicle you have. There are maintenance costs like oil changes, tires. And then at the end of the day, even the residual value, how much you’re going to sell it for can all add up to what you would call a total cost of ownership.

Going back real quickly to an IT context, when you’re looking at different storage systems, it’s very easy to think of, “Oh, well, there’s a certain price or there’s a certain dollar per terabyte that I’m going to pay for this system,” and it’s easy to forget all of the associated costs that go along with it. When you’re comparing a like-for-like system, maybe those heuristics like dollar per terabyte can be helpful. When you’re looking at vastly different technologies, say a traditional array versus something like a hyperconverged infrastructure, that’s when a TCO analysis can really help you make a good decision.

So going back to that car analysis, if you’re looking at say the difference between say a EV versus a traditional gas car, all of those things that I mentioned beyond the price, like there could be a federal tax rebate for an EV, you might want to put a home charger at your house to make it easier to charge it, the cost for fuel, the cost for maintenance can be higher with one versus the other. So when you’re looking at different technologies, that total cost of ownership analysis is really going to help you make that best business decision.

Camberley Bates: And I go back to a partner of mine who used to say, “If you’re going to buy by gigabyte at a price, you might as well buy tape.” And it’s like, yes, we’ll give it a chuckle, but the things that go into TCO are, yes, there is the capital cost of purchase, but there’s the maintenance and there’s the operational cost. And also the other big area that we’re now looking at is the energy cost of what that costs. But there’s also a design and how that’s implemented, and I think that’s what we’re really talking about here is the implementation here. So I’m going to go to our second question, which is really looking at the components that go into the total cost of ownership in the storage analysis that has been done. And I want to go to Mitch. Mitch, maybe you could give us a little bit about your experience in working with TCO models. I know you’ve been doing this for many, many years in a lot of different environments.

Mitch Lewis: Yeah. So we’ve done a lot of this work looking at different storage solutions and what is the total cost and how can we model that? So going back to what Tom said, obviously the solutions have a price and that’s a starting point. So there’s going to be a hardware cost and a software cost and support costs. And depending on the solution, those are either bundled together or maybe they’re separated out. But then to get to a TCO, there’s other things that you need to add into that, so operational costs, what kind of maintenance or management these solutions need, who is doing that, how much time does that take? And then you also need to look at the specific features and attributes of the solution. Those can have an impact as well. Things like how they scale, how much capacity do they have, what features might make management simpler or data reduction? Those things can all have an impact.

And then the other things that get layered on top of this really are time and scalability. So these things change over time. How much are you scaling? Are you scaling up? Are you scaling out? How granularly can you do that? And then when you’re looking at time, when are you in current costs? Those can have an impact, as well as what is the overall life cycle here? So this is technology. Things do eventually need to be replaced. So in storage we typically look at a five-year life cycle and then we’re looking at replacing the solution. So it’s taking all those things and you need to understand how they interact with each other, how those things all change the cost and you eventually get to something that can be TCO over a set amount of time.

Camberley Bates: And as a second follow-up, we saw some really big significant numbers on the analysis we did, and we’re going to be putting a link on to the report that they have done. I’m going to read off a couple of the items here, so if you mind if I look down. 30% cost savings that you found by looking at what is going on with vSAN, with VMware versus some other options that we analyzed in there? Some of that big stuff came in administrative costs as well as savings in hard costs, which would be hardware and support, which a lot of people, those are the primary things that they look at. When you did the analysis, where did you find what I call the pressure points in this analysis? That is, what are the things that really made the bigger difference between one concept versus another concept, approach?

Mitch Lewis: Yeah. So our modeling was looking at VMware, vSAN and specifically compared to traditional Fibre Channel SAN array. So like you said, we found around a 30% cost advantage overall. And then as you pointed out, the hardware and the support costs were a big factor in that. One explanation there is vSANs ability to leverage lower cost commodity hardware versus these more proprietary solutions where typically, the support costs are really high. Another thing that was pretty interesting was we took a pretty in-depth look at the operational costs and we actually found around a 76% cost advantage there. So what we did was we looked at who needs to be involved, what the tasks were to maintain these systems over time, how frequently they need to be done. So for vSAN, we’re really looking at a VM admin and the solution had lots of different management simplicity built into it that could reduce the time that was required overall.

On the other hand, in the Fibre Channel SAN environment, it required both the VM admin and an additional storage admin. So you’re getting a requirement of more resources, more time. That all adds up into more operational costs. And as I was talking about before, this operational cost is something that adds onto itself over time and accumulates. And then one last piece that’s interesting is after we modeled these costs, we looked at how they’re distributed over time. So we found that a lot of the cost advantage was coming in year one and year two, really because the same environment required really high upfront costs with the hardware and the support and all that. So that gives us two advantages to vSAN. It provides a little bit more financial flexibility over this stretch of time, and then two, if you imagined the future of year six, when you do inevitably need this refresh, you have this really high cost once again right up front. Whereas in vSAN, you can add much more granularly and you don’t incur this high upfront cost.

Camberley Bates: Yeah. And if I recall correctly, we did internally on the administrative piece of it, is the Signal65 lab set those systems up in our lab room, setting them and running through the parameters or the operational environments as we’ve seen people having to do. So that’s something that it’s not just taking information from other people, but it’s actually our hands-on work. So John, I’m going to skip over to you. The last question we have is how can solutions really lower capital and operational costs? You are down in the trenches I know with the technology, et cetera. So what are the things that have gone on that you guys have done to really lower those areas, drop the cost?

John Nicholson: So vSAN itself has a number of technologies that help you lower the capital and the operational costs. Some of it are things like just data efficiencies, compression, thin provisioning, how we lay out the data. Some of them also, when you look at capital, there’s something I like to talk about called server drive economics to where when you go and buy drives and a server and you go and buy the same drive next year, it’ll probably get cheaper. And drives get bigger, they get denser. And I was actually just looking at drives today and looking at how far the prices are continuing to fall, looking at just regular NVMe server class drives. We have different options of them going from the mixed use to the read intensive, some different options in terms of performance tiering and lower costs. So maybe going from the 33 cents per gigabyte enterprise-class, mixed use down to a data center class, read intensive drive that’s near half that cost. We’re seeing just the commodity, silicon costs go down.

And the other thing also from an operational cost, it’s very easy to train someone to become a vSAN administrator who’s already a VMware administrator. They’re familiar with the existing tooling, the vCenter Server capabilities, the performance services, they’re already rich in there. And you end up comparing this against an external array, which I was a storage admin, I deploy to manage them for a living. And you often end up with a weird bubble scale cost problem there to where it’s small scale, those activities, you do them so rarely that even if you try to dump them onto one person, there’s a high cost of entering and going in and out of all those different interfaces very intermittently, or at scale you now have to have dedicated teams, dedicated skill set, dedicated SAN management tooling, and just from not only lowering those costs but a cost transparency and understanding those costs.

If I’m buying hyperconverged infrastructure or I’m buying a storage only cluster with vSAN here, I’m buying servers, I can bid those out, they’re fairly easy to understand the component costs. I’m buying the software, I understand that. I can buy those together as a common unit, have a common refresh cycle in all my hardware. I’m done, pretty easy. If I’m going with an external SAN with a dedicated storage fabric, I have to account for the life cycle of my server, which I might run anywhere from four to seven years. I have that storage platform that I might run from three to five years and then I have the fabric where its own life cycle and refresh. Next year is actually, 2025, mark it on your calendar’s everyone, is a supercycle and Fibre Channel refreshes. Most of the Gen 5 systems out there are going into life and to support, you’re going to have to go buy new.

And so trying to track this Gantt chart of these multiple different systems that each have their own software and hardware refresh cycles and handle those, not only is it lower cost to do HCI, but it’s just simpler. I’m not having to track all these extra line items and predict what their costs and what each interval are, and if I want to get off that roller coaster, now I have to get out of this entire ecosystem versus with vSAN and hyperconverged infrastructure and storage only clusters here, I’m just buying regular servers. They’re pretty predictable. Intel is going to come out with a new CPU every 18 months. The drives will just keep getting faster and better. It’s a good life.

Camberley Bates: I love your busy head, John. Well, that’s about as much time as we have. There is a link to the report. Please go download it because it’s got lots of details and some summary papers so you can do a quick scan. But if you want to go into how all the gory details got put together, there is all of that in the appendix as well to go plow through. So thank you very much John and Tom and Mitch and we’ll see you again on The Six Five On The Road sometime.

Other Categories