Episode 10: The Modern Day Smokestack? The Economics of Cloud Management

Steve Prentice (00:03):

With so many companies moving their workloads. What used to be in the data center onto public clouds, they are constantly looking to optimize their investments, but sometimes the steps they take end up being misaligned with their plans, their budgets, or their security strategies. And this of course creates an ocean of confusion about what is the right cloud path to take. And more importantly, how to know that you’re doing it right both now and into the future. Hello and welcome to the cloud tweaks podcast, where we look at developments in stories, dealing with cloud cybersecurity and other areas of business tech. I’m your host, Steve Prentice. On this episode, I am pleased to welcome our SIM resoc who is CEO of Yotascale. So he can share his experience and expertise on this constantly evolving sector of business technology. You are to scale helps companies like Zoom, Hulu, and Compass manage their public cloud resources by empowering engineering teams with accurate cost visibility, actionable recommendations, and collaborative workflows. I welcome to the show.

Asim Razzag (01:08):

Thank you, Steve. It’s my pleasure. I’m looking forward to it.

Steve Prentice (01:11):

Before we talk directly about the economics and benefits of effective cloud management. Overall, I want to start by asking you about a piece of cloud technology that often flies under the radar. And that’s the concept of its carbon footprint. Perhaps you could start by telling us a little about your scale and how important the carbon footprints consideration is in an overall cloud strategy.

Asim Razzag (01:33):

Yeah. So your a scale, uh, helps companies optimize their, public cloud, infrastructure. So think about the fact that a lot of companies are moving their workloads, but used to be in the data center. And there’s a lot of cloud first companies into public cloud, like AWS, GCP and Azure. And with that, we have entered this era of OPEX versus CapEx, right? So you, you rent this machine, you rented resources and there’s a lot of wastage, right? As a result of that, there is economic wastage. And we’re also very passionate about the carbon footprint, which is part of it. So we help companies run their workloads really efficiently on a public cloud. Uh, so the price performance ratio is the best that they can get, which, which goes back to their bottom line. So, margins profitability and, cost of goods sold. A lot of those business KPIs they’re able to achieve by leveraging a product like you would a skill,

Steve Prentice (02:32):

The carbon footprint concept. I’m going to hear a lot about carbon footprint, especially when it comes to data mining and crypto. But what are we looking at in comparison to let’s say other industries like diesel pollution near trucking pollution, so forth, what is the carbon footprint impact that you were saving by doing this

Asim Razzag (02:47):

Know in the early days of that? Because for the carbon footprint, for digital infrastructure, there still needs to be a lot more visibility. That is a vision and passion of your scale, like downloading our product that we want to, we want to do, but we still want to profess right? That we want to make people aware that the waste that they have is not just an economic wastage, right. There is also the sustainability part of the equation. And I’ll just give you an example, like a, one of the cloud providers, Google, they recently launched this initiative where now they’re surfacing the carbon free energy percentage for the different regions that they w the workloads could run into. So, so this gives the ability for companies of trade-offs to say, look, I can run it cleaner in this region or in this particular zone for Google cloud, as opposed to some of the some of the other zones.

Asim Razzag (03:34):

And the challenge is clearly that because it’s not visible, the carbon footprint, the digital infrastructure is like more than the smoke stack, if you will, right? It used to be the employee to server carbon footprint would be the same, like as much carbon for current employee is causing will be one server. Now it’s hundreds of thousands of servers to one employee, right? So you can imagine where the big challenge is this day and age. So I think we’re, we’re in the early days of this, when it comes to digital infrastructure, there is more consciousness around this. As I said, Google has done that. And the opportunity for a product like your scale is to, this is something, as I said, we want to do down the road is not just provide the economic trade off, but also the carbon footprint trade-off and incentivize people, because people might not sometimes care about dollar amounts, right. But they might care about the planet. And I think that’s the, that’s the goal. But in the meantime, we, we do want to profess to these companies that have this digital infrastructure and public cloud that do bring that aspect, that consciousness to your teams, the fact that wastage is not just about economics, but it was also about sustainable.

Steve Prentice (04:40):

Yeah. The world has, or is trying to turn a corner towards sustainability and more sort of global responsibilities. The spending on public cloud services is forecasted to grow substantially, of course, and this has led to costs getting out of control. But the good news is that controlling cloud cost as well within the capabilities of most organizations. So you are here to tell us about, you know, how a thoughtful cloud cost control approach can help boost profits and benefit the planet. So, first of all, why is cloud cost management so difficult?

Asim Razzag (05:08):

Yeah. So just a little more color on that, right? Is that now a public cloud spend item is becoming the second, most critical item after payroll for a lot of companies, right? So there is, there is real ramifications of not making it efficient. And then you quoted the figures through the pandemic. We’ve seen an amazing uptick in cloud adoption, right? So it’s only accelerated because people have had to deploy the digital strategies much faster than they thought, right. Just to stay competitive from a business point of view. So what makes it challenging is a number of things, right? So one is that we have this advent of new architectures, right? So we have Kubernetes containers. And the adoption of that was not as much, three or five years ago, but they are being adopted at a pretty fast rate. So these technologies allow you flexibility, right?

Asim Razzag (05:59):

You can deploy things faster. There’s a lot of shared infrastructure concepts. There, things like microservices, you can put out there, but all of this makes understanding where the money is going very, very hard and basically understanding who owns what portion of the money that’s being spent hard. And hence that accountability part is difficult, right? That’s number, uh, number two is that every company has to, and to actualize their cloud spend based on their own business. Right? So there are companies that think about product lines, like how do I connect the dots between the products that I have some companies think about features. Some companies think about cost of goods sold because if they can save the cost per customer, right. That goes back to the bottom line. So every company has a unique way of looking at this cost and contextualize it, the cost is hard, right?

Asim Razzag (06:47):

Cause if you think about product business line services, like how do you make sure you understand the cost of those units in this new world, where things are moving so fast and they’re so transient, right? And then a lot of the techniques that are out there today, which are mostly coming from generation one cost management products are very manual. So it’s your millions of resources, right? And I call it death by tagging, which is like you tag every single resource correctly and that you can generate pretty reports. The bottom line is that in any infrastructure that is moving at a certain pace, that’s very, very hard to do. Right. And nobody can get this right. So, you know, there are elements of making sure that this gets very automated. There’s the level of sophistication to help customers connect the resources back to their enterprise context, as I mentioned, and then there are various, uh, teamwork, right?

Asim Razzag (07:33):

So we, we feel that in, in some cases, this tends to be more of a finance oriented initiative. So the finance team is caring a lot more about cost. And they’re just looking at the engineering team to magically, you know, make things better or different, but there needs to be a communication between the key stakeholders, whether it’s finance, budget, owners, engineering, dev ops team, and, uh, that communication, a lot of times it’s happening out of bed, right? Pre pandemic, this would be in the hallways, right? Like, well, I told you somebody who wants an analytics service financing as well. You know, there’s a lot of recommendations. There’s a lot of waste based on the report that I ran from the genuine product that you could basically drive efficiencies, then they have to do hours of research, go back to them and say, look, part of this is my backup field or cluster, or there’s all these reasons, right?

Asim Razzag (08:19):

And those are all out of men like this emails or pre pandemic, hallway conversations, things of that nature. So there has to be a system that is agreed upon between the different parties to see what does efficiency mean and how much waste do we have agreed upon and what are the actions we’re going to take it, that’s the other aspect. And then finally, like a key ITOs for us is engineering empowerment. So a lot of times engineering teams and the feeling disempowered, because they’re being thrown these reports, right. That don’t have a level of context. And, you know, they have to deploy workloads. They have to worry about performance, that we worry about security. They have to worry about reliability and all this is new entrance, which has cost. So like they have to go understand all of these resources, how do they work and understand exactly what portion of it is that they can effect.

Asim Razzag (09:06):

Right? So if not, if they’re not getting the right data on what they own and what they should be accountable for, they write off a lot of these initiatives. Right? And they’re like, look, whatever you’re telling me, doesn’t make sense. Right? Because you’re asking me to make changes in my system that are not going to lead to better performance or better reliability. Right. And I don’t in a lot of cases, I don’t even own most of the costs you are attributing to me. So I think those are some of the key challenges that makes us complex, like in anything there’s people, technology and process pieces, right.

Steve Prentice (09:39):

Play hosting for the clouds week’s podcast is brought to you by ISC squared, the world’s largest nonprofit association of certified cybersecurity professionals, ISC squared membership of more than 150,000 is made up of certified cyber information, software and infrastructure security professionals who are making a difference and helping to advance the industry. ISC squared offers a portfolio of credentials that are part of a holistic programmatic approach to cybersecurity visit I S C two.org to continue your journey and get certified today. It sounds like an increasing degree of literacy and sophistication in the management of data. And that’s a very interesting concept that identifies why it has been so difficult in the past and some of your very proactive initiatives in, in making it more efficient and more understandable and very interesting sort of side comment about the lack of hallway interaction. You know, as we’re seeing as a side concept here, some of the impacts of people having to work, you know, from their homes in this pandemic situation, it’s interesting to observe that kind of creativity or opportunity or the opportunity cost that this has caused in, in not being physically available to discuss these things.

Steve Prentice (10:56):

So can we add to that? Any additional challenges you have in achieving cloud cost optimization?

Asim Razzag (11:03):

Yeah. So it’s a one is culture, right? So you have to have a culture. We believe that the culture has to be culture of empowerment, right? So engineers, you have to treat causes first-class citizen. So you treat performance that way you treat security that way you treated reliability that way. Right? So what we’ve observed is that company cultures, where it’s a lot more command and control, which is the central group, that is the one that makes all the decisions and basically takes a stick approach to engineers and beats them up when they’re out of budget and all that. And there are, there are companies like that, right? It’s not, it’s not a very, it’s not very conducive to driving change and efficiencies. And, and we believe in a culture of, again, empowerment that equip people with the right data, right at the right time and the right actions that they can take that are implementable, right.

Asim Razzag (11:52):

Don’t insult their intelligence by stuff that is not possible to do or is not relevant to their world. Right. That’s another one. And then it’s about trade-offs right? So everything in life ends up being about trade offs. So it’s not like you can cut your cost down to zero a shutting down on the servers, right. Nothing’s going to be able to do that. So you have to make a trade off between performance, reliability, cost. And I think trusting engineers and putting at the fingertips in the context of the services and applications to do these trade offs is important. And if it’s that’s missing and it was like, look, you can save a bunch of money on your Memcached server as an example, right? Because the CPU utilization is low. Well, you know what, in America, silver is going to be very memory bound. So I don’t really care about six years realization.

Asim Razzag (12:38):

So what is the trade off you’re asking me to make? And there are other trade offs where a lot of times engineering and dev ops teams are worried about just quite frankly, they’ll get into a lot of trouble. If the system goes down, they’ll get into some trouble. If a system is too expensive, right. That’s kind of the bottom line. So how do you give them the optionality to say, here are two or three different actions. You can take recommendations. If you will. In one case you have a lot of headroom, you have a lot of buffer for performance and you can save less money. But in another case you have less headroom and you can save more money, but you make the trade off because you have the most amount of context we can equip you with two or three options. I think those are the ones.

Asim Razzag (13:16):

And then finally, as I mentioned, attribution is important because a lot of times there is this ecosystem of data that you have to bring into the mix. And what I mean by that is that there are surrounding tools like your APM systems that have performance and memory data, right? There are logging tools. There is directory services like service now, and others that have hierarchy and ownership data. So the list goes on and on, which helps with respect to what the context is. So a lot of these simple minded tools show hell, they just don’t look at it in the context of all of this CSED tools like Spinnaker and others, that divide things by sort of deployment and workloads. Some of that stuff has to be integrated and brought into the mix to have the right context of, you know, okay, this workload makes sense to optimize. And the attribution is correct. So I think those are some additional challenges that, that make it increasingly difficult for teams to do a good job at it.

Steve Prentice (14:13):

Well, in the introduction you were saying about so much of this is not only about improving processes basically, but also having this global responsibility as well. So what principles can you identify the cost optimization that will help boost profits and benefit the planet at the same time?

Asim Razzag (14:30):

Yeah. So you can’t, you can’t improve anything you can’t measure, right. It starts with visibility. So it is back to creating the visibility where we have had customers that do believe in this, uh, concept of engineering empowerment, right. They want to make sure that it’s self-service accountable. So if a team owns a set of services or a product, they understand what the cost is and then they can take action. Right? So in, in these scenarios, when you provide these teams, the visibility, they become very conscious of it. And if they believe and trust in the data, then they will take action. So in one case, we’ve, we’ve saved a company like tens of millions of dollars because they’re spending, they have a very high spend in AWS as an example, just by providing the visibility and accuracy of data and enabling and empowering these teams.

Asim Razzag (15:19):

And I kind of mentioned this already, that you have to treat cost as a, as an important KPI at the end of the day, by which I mean, and this as an economic piece to it. And there is a, as I said, a carbon footprint piece to it and what you will see in the future more and more again, we want to help innovate on this front is, you know, what is that, what is that footprint like? Is it three trees or, you know, because everybody can relate to that on that emission. And then I would say, it’s connecting the dots for these teams, right? So I believe I come from an engineering background, right? And it’s a creative endeavor and engineers at the end of the day, you know, we’ve read Daniel Pink’s drive, right? It’s about autonomy purpose mastery, right? So you have to connect it back to the higher purpose instead of saying, Hey, you’re over your budget, cut down your costs.

Asim Razzag (16:10):

If you don’t, you know, ramifications. But if you understand what are the KPIs you want to drive as an example, if I’m a business and every dollar that I can save in terms of a deployment for a customer, I can invest back into my business so I can save jobs. If I explain that to my engineering team and help them connect the dots back to how they’re helping, um, reduce the cost of goods sold, then they will be loved more incentivized, right? Because that’s a higher level purpose. And also giving them autonomy to say, Hey, here are the parameters within which you need to operate in your services, a range of budget that you should operate within. And how you do that. It’s up to you. I’m not going to micromanage you, right? And the third one is mastery, right? That every engineer wants to get better at their craft and helping engineers understand that if they can build systems more economically and sustainably efficient, they’re actually building rare and valuable skills that are going to make them a lot more relevant in the industry.

Asim Razzag (17:11):

And so incentivize them to learn that craft. I mean, I did a webinar a little while ago with some of the engineering leaders. And we talked about how you could even have that in the interview process, right? Talk to people when they build systems in the cloud, how did they, how did they make it efficient? And I think that that, that connects all three key pillars of autonomy, mastery and purpose. And that creates a culture where you don’t have to micromanage. You don’t have to use a stick all the time. Right. Everybody knows what their work is contributing towards in terms of efficiency and a better planet.

Steve Prentice (17:46):

So here’s the question about cost versus free. Should I pay for a cloud governance tool when public cloud providers have alternatives that are free?

Asim Razzag (17:54):

Yeah. That’s a good question. And I think all a cloud providers will always give you some basic capabilities, right? So I think if you have a simple deployment, right, you have, you’re spending a few thousand dollars per month. You don’t have complexity of teams. Yeah. You don’t have a very complex infrastructure. Like you don’t have cloud native and Kubernetes and microservices. Right. And all you’re looking for is basic understanding of your bill. Like just slicing and dicing it. That’s the level that we’re at at which you are by all means, you know, use, use these tools. I would even say, save some money and don’t buy a generation one product that kind of ends up doing something very similar. Right. But if your, if your needs start becoming beyond that, right. Which is you have complexity of infrastructure, your scale is a lot larger.

Asim Razzag (18:41):

You have workloads that are running on Kubernetes and containers. You have microservices. And ultimately multi-cloud also becomes one entrance into this, right? So you have workloads in Azure, AWS, GCP, and others, and you have complexity of teams, right? So it’s not like there’s one single team that owns one workload. And if you slice and dice that you’re good to go. Right? You could have dozens of teams that are working in many different parts of the product. And then finally, how do you connect the dots back to your enterprise? How do you connect the dots about how you think about cost? And that’s where you have to integrate a lot of subsystems that are doing, where you can get system metrics, where you can get hierarchy and ownership metrics, right? And you can, overlie a lot of that. So if, if it is about really driving the health of your business, right, focusing on their profitability, empowering your engineering teams, driving a culture of self service accountability, then a Euro skill can be, can be a really good fit at that scale and complexity. But it’s, it’s very, very basic. The in fact, we can sometimes walk away from a customer it’s very, very early, right? And they don’t need a lot of slicing and dicing the very early in the journey in cloud. And those tools have a place for that. I mean, at the end of the day, you get what you pay for. It’s something I say all the time.

Steve Prentice (19:53):

Absolutely. Are there industry specific challenges to cloud cost management? We should talk about,

Asim Razzag (20:00):

Right. So there are, if you are a media service, like, um, a video streaming service, like a Hulu or zoom or somebody like that, then the network costs is going to be very important to you. So you have to focus a lot in terms of the IO cost and network IO cost. In additionally, there are specific metrics, right? So for a video streaming provider, it’s more about the cost of every frame delivered, right? That’s the bottom line or for like a zoom. It could be the cost of the recording feature, things of that nature. Some of the other could be around file sharing. So if you’re a Dropbox or box or somebody like that, try to just storage costs will be a pretty big portion of your total cloud costs. So you have to think about storage costs and how do you tier data a lot, right?

Asim Razzag (20:48):

That becomes increasingly bored. I’m not saying stories costs are not going to be important in other domains, but I’m saying there are some domains that some of these things are more important than others. If you are a very machine learning AI ML type company, then compute because you’re running these models, right. And they are very compute intensive. So a lot of times it can be very memory intensive. So I think that might be an important function of, um, the biggest signal in terms of what your cloud cost is. So, so definitely depending on the kind of industry you’re in the kind of product that you are building, you might be more sensitive to certain types of resources that are being leveraged, as opposed to others.

Steve Prentice (21:26):

It’s a great summary of some very timely wisdom. Um, and I’ve really enjoyed listening to what you are describing. So is there anything else you’d like to include in the podcast and the other message or concept for our listeners?

Asim Razzag (21:39):

Yeah, I think I’ll just reiterate the two points. One is back to empower your engineering teams, enable them and make this a collaborative effort and have a culture of cloud cost. I mean, you’re a scale, obviously it’s, self-serving, it’s, it’s a really good product and tool. It can help, but it’s, you can’t do everything. And if you don’t have the right culture, then you know, the, the product can only do so much. And the second one is, as I said, increasingly, even though we are early in terms of providing the visibility and the carbon footprint, when it comes to digital infrastructure, be cognizant of that, right. Leave the planet a better place. So think about not just the dollar amount, right, but also think about as you save this money, their incentives are aligned, right. A lot, sometimes sustainability might go against the economic part of things, right. But in this case, if you make things more efficient, you save money, but you also help make the planet a better place. And I think it’s incumbent upon us to ensure that that’s also front and center, just not the economic benefit.

Steve Prentice (22:39):

So as expected a treasure trove of wisdom, not only just about the economics of cloud, but the vital elements of global ecological responsibility, workplace culture, and of course, sustainability and growth, truly some great lessons from an exemplary industry leader. And of course you can find out more about Yotascale, including a demo and great case studies at yodascale.com. Thank you so much for joining us here today on the cloud tweaks podcast.

Asim Razzag (23:15):

Thank you, Steve. And thank you for having me. And it’s been a lot of fun.

Steve Prentice (23:19):

And as for us, you can check us out at cloudtweaks.com and follow us on Twitter at @CloudTweaks. If your company is looking for some great exposure to thousands of decision makers in the it cloud and related industries worldwide, please get in touch. We can craft a campaign that will get you noticed through our website, social media and newsletter channels, all of which enjoy substantial readership. So if you like what you hear, please consider leaving a review wherever you get your podcasts and tell just one more person about us until next time. I’m Steve Prentice stay safe and thanks for listening.