TECH-IT-OUT: Cloud Optimisation - Maximising Cloud Efficiency

In this episode of Krome Cast, Tech-IT-Out we delve into the world of Cloud Optimisation, focusing on maximising efficiency and reducing costs through the correct sizing and management of your Cloud environment.

If implemented incorrectly, improper Cloud utilisation can lead to inefficiency, unforeseen expenses, compromised security, and can cause unnecessary environmental impact.

Krome's CTO Rupert Mills and Commercial Director Sam Mager, explore the intricacies of Cloud optimisation, from both a cost optimisation and technical optimisation perspective, and discuss the importance of using the right tools and management strategies for monitoring and adjusting resources to achieve efficiency and cost savings. 

Various Cloud Optimisation tools such as from the large hyperscalers, such as AWS Compute Optimizer, and Microsoft Cost Management are explored, as well as third-party vendors like VMware CloudHealth (Aria) and IBM Turbonomics.

► ABOUT KROME: Krome Technologies is a technically strong, people-centric technology consultancy, focused on delivering end-to-end infrastructure and security solutions that solve business challenges and protect critical data. We work collaboratively with clients, forming long-term business partnerships, applying knowledge, experience and the resources our clients need to solve problems, design solutions and co-create agile, efficient and scalable IT services.

► KROME WEBSITE: https://www.krome.co.uk/

► SOCIAL MEDIA
• YouTube: https://www.youtube.com/@krometechnologies
• Linkedin: https://www.linkedin.com/company/krome-technologies-ltd
• Instagram: https://www.instagram.com/krometechnologies/
• Twitter: https://twitter.com/KromeTech
• Facebook: https://www.facebook.com/KromeTechnologies/

► CONTACT
• Telephone: 01932 232345
• Email: info@krome.co.uk

Transcript

SPEAKERS

Sam Mager, Rupert Mills

Sam Mager: Welcome to Krome Cast, Tech IT Out. I'm Sam Mager and I'm back once again, with my business partner Rupert Mills.  And we're here to talk about Cloud Optimisation. 

Rupert Mills: Hey Sam.  

Sam Mager: Exciting stuff, well it can be, albeit it can be very unexciting, should you get it badly wrong, which is something we've also,

Rupert Mills: Very painful should you go about it in the wrong way. 

Sam Mager: We've seen that on a few occasions. So I'm going to kind of chuck it your way to give us the three things we're going to work through, just you and I today.

Rupert Mills: Sure. Yeah, I think we're going to talk about: What is it? Why is it important? And how do you do it?

Sam Mager: Cool. All right. Let's jump in. Number one.

Rupert Mills: What is it? Yeah, so. So cloud optimisation is it's kind of two things. So it's using cloud in the best way possible, obviously, the optimisation piece, but the thing that most people talk about when they talk about and optimisation is cost optimisation, they are two different things.

Sam Mager: They are, but they're kind of hand in glove though right? If you if you put it together badly, it will cost more. 

Rupert Mills: Yeah, absolutely. Not even necessarily badly. If you put it together, without paying attention to the small print, basically, it's the same thing as, it's nothing new under the sun, historic contracts would have been okay, here's the small print, if you don't read it properly, you're gonna pay this much per minute for your calls to Afghanistan, or whatever it might be. 

And you suddenly get stung with a huge phone bill, clouds got kind of the same thing, it's there's some great services out there does a lot of good stuff. But actually, if you don't read what you're subscribing to, when you subscribe to it, and you just sign up and put everything in there, then you can end up with some challenges. So yeah, the cost the cost piece is, is the piece that most people refer to when they talk about cloud optimisation. But actually, the technical piece is, okay, how do you pick the right service? What do you do? And that's the, that's the bit that then leads to the cost optimisation. 

Sam Mager: Yeah, and I guess there's been, as we've seen, and this is not new, we've talked about many times over, the big push to the cloud and Gartner out there still throwing numbers around about how much we've done in cloud. This that and the other, it's, as we've seen, the bit I'm most conscious of when we've had, and are having conversations with people that have kicked off their cloud journey, either recently, or sometimes quite awhile ago. It's the biggest gotcha I've seen, the first thing I ask is, did you just lift and shift it? Because we often see that yeah, we took it exactly as it was when we put it in the cloud. And guess what? It's not very efficient!

Rupert Mills: Yeah yeah, absolutely. If you don't, if you it's that same thing again, if you, don't do properly, just do a quick, off you go. And it's all there, then you can suddenly find yourself lumbered with a very large bill. So yeah, I mean, when you get on to the why is it important, it's that bill piece, it's the environmental aspects of it. 

Sam Mager: Of course yeah.

Rupert Mills: So same thing is if you're running a data center in your building, and you've got or, or in a data center for that matter, and you're looking at the efficiency of it, the people like Microsoft, Amazon, etc, Google, they have hugely, hugely, hugely efficient data centers, like I mean, they're achieving efficiencies that just wouldn't be dreamed of in lots of other data centers, but it's still cost around a machine, you turn the machine on, it's still generating heat, it's still consuming power, it's still doing all of that stuff that consumes resources, that ultimately, that optimisation is for there for environmental reasons. If you run it and you use it, it's going to cost you money. That sort of stuff, basically, is the okay, we need to need to care for that as much as, if you put it in an on-prem data centre.

Sam Mager: If you had it on-premises, same thing, right, 

Rupert Mills: Same principle. 

Sam Mager: And we are, obviously a lot more now than then historically, it used to be put them into a server cab either on-premise or at a colo. People didn't think about that you spent your money on it. And that was kind of it was done. 

Rupert Mills: Yeah. 

Sam Mager: And if it created a lot of heat, well you put more air con, you turn the air conditioning up, or that sort of stuff, people obviously a lot more cognisant of the impact and you are right, you know, you put it into a very efficient data hall, whoever owns it. But you are right, if you're not using, you spin up 100 VMs let's say, and use 50 of them. That's massively inefficient. And there is, whilst you've lost, as you said you've put it somewhere that is more efficient, there's still an impact, it still creates heat and all the rest of it uses power that's just not being used. So it's looking at the the efficiency and the impact in that model as well. Even though you've you've not made it someone else's problem is still your problem, our problem, everyone's problem, but 

Rupert Mills: It's out of sight out of mind.  

Sam Mager: Out of sight out of mind, that's exactly it.

Rupert Mills: That's the issue is that people go okay, I've put it all in the cloud. It's not my problem anymore. And until the bill comes,

Sam Mager: I think that's driven partly then by the how do you monitor that? And that's, not a dark art, because tools are available. I'm sure we can talk about that. But you know, it's you put it in there and it works. You've done, you've ticked some boxes, we went cloud-first, right? We've hit part of a strategy, it's working. That's great. And that's if it isn't too cost-inefficient. If it's set within our budget, well then it is now out of sight out of mind. We're not necessarily looking any further and we know that if you do look further, you could be more efficient from both an environmental and a cost.

Rupert Mills: Yeah, absolutely. And then we've got the security angle, obviously. So you have virtual machines running in your data center, they're an area for compromise or attack, potentially, if you forget them or forget to patch them, you're the same thing going on with cloud, you've got that out of sight, out of mind thing going on where you built a load of machines, and even with shadow IT where someone else's below the machines, and you're not even realising they're there, you've got that, that area where people can pick on. And if you're not maintaining not looking after them, not only are they costing you money, but they're an attack surface. 

Sam Mager: A threat vector. 

Rupert Mills: Yeah, it's a threat vector. So you shut that sort of stuff down on it, you say, Okay, let's get all that shut down. Not only is it then not a threat vector, but it's also saving you money. And it's, there's the consume or Pay-As-You-Go consumption albeit that you have, again, discounts for commitment and all the rest of it. But effectively Pay-As-You-Go mentality of the cloud, means that you can turn around and say where historically you'd have invested in that VMware estate and said, or that that VMs running 16 Gigabyte of RAM, it only needs 8, but it doesn't matter. We've got plenty of headroom in the system, you're actually trying to get rid of headroom in the system altogether. When you're optimising, because you're just trying to say right, the more headroom I have, the more I'm paying for, even if I'm not using it. And the ability to move that dynamically is a big thing these days as well. So I mean, that's, that's coming on to the tool thing as well,

Sam Mager: Maybe we should go into that, because that's where, excuse me, I've looked recently, I know you've been looking at the team and looking at, different tools to kind of do the right sizing bit, but automatically, because otherwise someone is having to look at that and you're adjusting thresholds and this and the other and then then it's inefficient because you're taking up a member of staff just to do the management there obviously. If we can actually use the correct tool set, and a lot of thought has to go into what is the correct tool set for your particular use case. It's not, there is no panacea, as far as I'm concerned, at the moment, I'm sure there'll be a market leader at some point that will potentially outstrip them all. But right now, there's a good choice for different, depending on what hyperscaler you use and how many you might consume to the different tool sets you can use to, to do as you said there, to do the monitoring piece, and then adjust it as we need, as we burst let's use more. But as we come out of that one, downsize it again, because even if you if you burst up, let's say it's down to an individual to, to move that. And then it's up to an individual to let's let's resize it again, if you miss it by hours or days, then the impact depending on the size operation could be enormous.

Rupert Mills: Yeah. Absolutely, I mean, if you get into tools, you've got the big hyperscalers, sort of AWS, Microsoft, they've got tools in there, Microsoft's got Cost Management, that will do Azure, it will also look at AWS, AWS, has the Compute Optimizer, that will allow you to look at what you're doing, and try and work out where to, how to right size, your resources, mostly within the AWS estate. But you look at those sorts of tools. And then you've got third-party vendors, people that you've known for years, like VMware or IBM, you've got what was CloudHealth, it's now Aria, and you've got IBM is one you were talking about the other day.  Yeah. Turbonomics. So and those tools are doing it from the perspective of having no vested interest. But there's an underlying theme there, that it's okay, how do we move these things around and make some revenue for a historic, or traditional, then historic vendor, that's the wrong way of putting it, but a traditional vendor that's done more of a, an on-prem environment, that now you're moving into, right, they can work in the cloud and help you optimise how you're spending your money with the big boys in the cloud space. But they have different levels of automation, and different levels of manageability. So it's really picking the right one for the client and saying Okay, what do you use? Do you use 17 different clouds, you get things like, like VMware's CloudHealth, or Aria, and we'll look down at things like Alibaba Cloud. Now, they're fairly sort of down the list of things you use, but depends on where your business is based. If your business has got a large footprint out in China, or something, you might be making big use of that. So it really depends on what you're using, which is to decide which is the right tool? And then how well do you implement that tool. And the tools will help you look at the concurrent spend of okay, if you take a compute resource, so say, an Azure VM, or an EC2 instance, in AWS, and that's a VM that's this big, and it only needs to be this big, or it's this big, and it's should be that big, and moving it around to right size that it, doesn't look at right-sizing between services. So if you take, as a really simple example, you could take Active Directory, and so historically, we've run Active Directory. So I'm going to go and build some domain controllers in Azure as compute instances. But actually, should I be using Azure AD to do that? Or, okay, historically, I've built a SQL Server and I've run a SQL server cluster, which will be two virtual machines or two SQL Server licenses. And they're clustered together, or however you do that. And then you've got clustered instances within Azure or in EC2 or whatever, that or actually within AWS, that will give you the ability to say we're not actually using the underlying software that we use to use at all we're getting the same end result, the same service end result. And it's picking that service end result correctly. That sometimes says you don't run the virtual machines at all and you do it completely differently. But that is a whole nother level of optimisation as well, that doesn't tend to get picked up by the tools or tools are generally comparing within a bracket. And then you've got to have somebody who knows what they're talking about to compare outside of that bracket, and really look at true optimisation, to then give the business the ability to charge it back. Pick different organisational units. 

Sam Mager: Cross charging etc, yeah. 

Rupert Mills: Yeah, and those, again, different tools will allow you to build in that cross charge will allow you to build in different departments, different labeling for systems, or tagging for systems to allow you to work out where it goes. But all of that should be in your in your cloud optimisation strategy. Yeah.

Sam Mager: Looking at, I guess just the way I think about things, you've got the cloud optimisation, we're talking very much about hyperscalars. But can we pull that toolset into, because most operations are hybrid? And whilst there is still a lot of noise around everyone will be, its cloud-first, I think, we don't disagree, that most enterprises are, and will probably remain hybrid. 

Rupert Mills: Yep.

Sam Mager: And you'll see a lot of the tooling though, that we can use to manage hyperscalers coming down on edge sorry, not on-prem. But we are, you see that, like I forget the, what's the Dell Edge tool, HCI platform? 

Rupert Mills: You mean VxRail?

Sam Mager: VxRail, but it uses.. No, Azure HCI. 

Rupert Mills: Ah right yeah, Azure HCI. 

Sam Mager: So we're now seeing, kind of the way we manage Azure, but we're seeing that, on the edge or on-premise, or whatever you want to call it, so now, you know, that same toolset and optimisation piece actually goes across both elements, right?

Rupert Mills: Yeah, absolutely. And you're back into the, the thing we've said multiple times, sort of predictable workload, cheaper to run there, 

Sam Mager: Yeah.

Rupert Mills: Burstable workload cheaper to run there. And then let's look at what 

Sam Mager: And now manage that correctly.  

Rupert Mills: Very true. 

Sam Mager: So again, it comes back to, it sounds like we beat the same old drum, but it's, look at the data, audit first.

Rupert Mills: But if you use the right tool, and it's doing the automate the automation, so some of the third party tools have really high levels of automation, some of the, the built-in tools have sort of scripted automations, or webhooks, or things like that you can connect to at the back end. But if you're using that you can move those resources around completely. And realistically, if you buy a fixed asset, as a piece of hardware, that's in a datacenter. And then you're using Cloud to supplement that, you should be filling that fixed asset as full as possible, first, and then bursting. And that's how you get the maximum efficiency out of that. But again, it's very much looking at it on a customer-by-customer basis, or client-by-client basis, and making sure that you say, what's the right thing for this business? And how do we make that work for them, whilst helping them? It's just a simple, simple thing. So do you want that a one year or three-year commitment? That's a massive saving from a cost perspective. But realistically, if you do that, before you optimise, you could end up really stuck because you've just committed to three years, and suddenly realise you only need a quarter of it. 

Sam Mager: Yeah. Kind of, you know, measure twice, cut once

Rupert Mills: Due care and attention. 

Sam Mager: Yeah, absolutely. 

Sam Mager: So just understanding what the what is the business trying to achieve, what is the best tool set, or product to invest in? And why as long as you've got that mapped out prior to any commitment, then yeah, we should be good.

Rupert Mills: Yeah, pretty much. Yeah. But certainly look at what you're doing, because it's amazing how much all of a sudden, you can go, I can make that a heck of a lot more opt, more optimised. Or efficient, that's the word I was looking for.

Sam Mager:  We'll finish on that word efficient. Thank you, Rupert. Cheers. 

And thank you for joining us on this edition of Krome Cast, Tech IT Out, and remember to like, subscribe, comment and share. And join us again next time on Krome Cast, Tech IT Out.