How the world's largest enterprises are approaching really-hybrid cloud
Transcript
>> Hi. I want us to take this opportunity to congratulate Keith Townsend in the CTO advisor team for creating this conference, in such a short period of time and given everything else going on. And thank them for the honor to participate in this event. I definitely want to thank each of you guys for attending as I know there are so many competing priorities going on today in all of our worlds. My name's Petrisa Pecnik and I am the VP of Customer Success or BlueCat Networks.
In case you're not sure who we are, we are the industry leader in the DNS networking space providing DNS, DHCP and IPAM solutions for some of the largest enterprise customers around the world. Over the last five years that I've been here at BlueCat, I've had the exceptional opportunity to work closely with some of the largest customers and really see, firsthand, how they've been managing their network complexity in general. The reality is the network of yesterday is not the network of tomorrow. And the move from on-prem to virtual infrastructure to now what we're here to talk about today, which is the move to cloud is ever changing.
And the one reality and constant in my world is that DNS is central to all of that. But the shift is happening no matter which way we look at it. The goal of this presentation today, and I know many of you have read the short blurb that we have posted, but the goal is effectively to talk about three key areas. One, to share what we've learned from large enterprises on how they've adopted the cloud shift in their environments. And those conversations and learnings have really come through both our customer base as well as conversations in the industry with leaders like yourselves.
We want to talk about, in this presentation, the DNS designs that we see and some of the architectures that include cloud, in various industries that we work with, we've taken three examples that we'll walk through. And then finally, the part that I'm most excited to share is really about the learnings that we've seen from our customers that we ourselves have learned as we've made the shift to cloud. And ultimately, what we've looked at as we've gone through this progress and this progression into cloud infrastructure, and moving out of our data centers into the cloud.
When we talk about the cloud environment in general, three to five years ago, we had seen a lot of customers who have made it part of their IT mission to move to the cloud. And when we talk about cloud, it is either private or public cloud. There seems to be a lot of discussion ongoing around the public providers who they are and what what they provide, as well as companies, we're seeing a shift where they don't want, necessarily, all their private data in the cloud.
And so we'll walk through the different examples that we've seen. In general, here at BlueCat, we define private versus public in really these key areas. From a private perspective, when we talk about private clouds, it's traditionally owned and managed by the enterprise. It usually combines some form of off-the-shelf technology, whether that's the the hardware that the platforms are sitting on, or the software that is used to manage the environments, but there's usually a proprietary component that we see our customers have. It tends to be behind some form of corporate security, either VPN access that restricts to only authorized users, or some sort of firewall or security provision that we as organizations put in place, but ultimately it has a management layer that is owned by the company and the organization, the enterprise themselves.
When we look at the public cloud just to level set when we talk about public in this presentation what that means it is usually the third party operated clouds. So it's usually one of the three major players that we hear with traditionally AWS, Azure, Google Cloud. There's other players in the market. Predominantly, these are the three that especially in the North America and EMEA markets we hear more of. And conversations circle around these three providers. These three providers are responsible for capacity planning, geographic diversity, redundancy.
I take some of the responsibility away from the organization, the enterprise, and that's what we pay for when we use these third party providers. And the model that that these individuals use or that we've traditionally seen is based on usage. There is a flat rate or a fee burn usage. In our case, when we use these providers, it's usually a usage model that can get extremely expensive. And so there's efficiencies that need to occur. The reality is, most companies don't fall into one of the two buckets.
They, including BlueCat, we're more hybrid cloud environment than anything else. We use all public clouds. We support all three of the public clouds as we have products in each. So we have to use each in our day-to-day activities. But we also have a private cloud that we manage ourselves. And so we like most of you and most organizations who are going through this cloud transition, have made the decision to have diversity in it and have a hybrid cloud model. This can make it exciting for our teams to use the different technologies available.
But it is not a strong suit for our CFO who would much prefer to get a single monthly bill versus multiple monthly bills that you can negotiate a discount on but the reality is No matter how you do hybrid cloud in some way, shape or form in your organizations, it's likely to be a combination. If we look at some of the trends that we're seeing, and we've partnered with Gartner and used some of the data here, but all of this data has been confirmed in almost every conversation that I have with any customer.
But there's really four main trends that we see from an adoption standpoint. And these trends are driving a slower adoption rate than what we had seen in the past. They really focus around cost control, insufficient skills, provider diversity and local access. From a cost control perspective. It is one of, by far, the most common conversations that I have with leaders. is managing the cost and driving cost optimizations. Because everybody would like to do more in the cloud, but it is, it has been cost prohibitive for a lot of organizations to date from an insufficient skills perspective that this is actually slowing down a lot of customers from an enterprise is from moving pure cloud forward.
And areas where we see it really coming into play is that the skills themselves tend to be niche. We tend to have organizations including our own, where we have specialists based on the provider. So we'll have an AWS certified specialist or an Azure certified specialist. Having cross functional generalists is not something that the industry as a collective whole has a lot or an over capacity amount of that skill set, yet it is coming but it isn't there today. The other challenge with this insufficient skill set is that there are a lot of legacy platforms and applications that organizations have that follow a lift and shift model.
And that can be challenging because legacy applications don't work the same in the cloud. And there are opportunities for refactoring, but you need the skills in order to refactor applications to make them work in the cloud. There's also cloud native functions that you need to optimize in order to have applications work efficiently in the cloud. And so all of these are being impacted by the newness of all of this. When we look at cloud provider diversity, we expect and Gartner expects that by 2024, most large enterprise customers are going to have a multi-provider strategy.
They're not going to be single threaded with a single provider. And in that model, the complexity of the network is going to increase. Lastly, the geographic diversity of most large enterprise organizations has become more and more relevant in all fathoms of a network architecture. Gone are the days of huge backbone pipes that would traverse back to the DC and that all traffic would go over. Many verticals that we engage with are having edge conversations where they would like the data to go out to the internet directly from the edge.
Low latency is more prevalent in terminology than we have ever seen before from a discussion standpoint. Having access to compute at the edge and in local regions is becoming more and more prevalent as we look at privacy factors that are now coming into play such as GDPR and EMEA. There are concerns from all aspects of an organization to have exit strategies to have traffic traverse via the internet securely. But to make it easier for the end clients to get what they need, without necessarily having to understand all of the intricacies and complexities of a network.
On the right hand side of the slide, you'll see a diagram, which is the hype cycle. And we've confirmed this with a lot of our customers. Over the last four years or so, there has been a significant ramp, which is the arch on the left hand side. And more and more companies have started talking about cloud. But they didn't exactly know where they were going with it and how they were going to start adopting it. And there may be some of you in this conference that are just starting the cycle.
And we have some customers and enterprises that we work with and that we talk to regularly who are very far on the right where they have have gone through their adoption cycle and have started really getting into what cloud can be for them in the future. But the reality is no matter which customer I talked to, whether it's at the beginning of the cycle or the end of the cycle and where they are in that medium, the realization is that they're still two to five years out from really leveraging all thing cloud.
And from really feeling secure that what they're doing in the cloud is having significant operational impact for the entire organization. There's pockets in every organization where it is being used at different varying levels, but as an entire organization, there isn't a customer that I've talked to that is completely moved to the cloud or is that even ready to do that? And I know that when we talk about things like two to five to 10 years out, it can be daunting, but with large IT scale projects, as all of you know, that timeframe is a blink.
And we will be past two years. As we look at the two pictures, or the two graphs here on the slide on the right hand side, there's a number of organizations that we see who were early adopters of the cloud, who are revisiting or reevaluating why they use the cloud and why they want to move to the cloud or expand their usage of the cloud. By far above and beyond reason to review what customers have done or what organizations have done is really to drive cost optimizations.
The existing initiatives and activities that customers have pushed out to the cloud have been changing for an organization. But they've also come at a significant cost. And so organizations are spending more time now looking at ways to expand what they're doing in the cloud and having a cloud first strategy, but they want to ultimately look at the cost of those initiatives. We see a lot of customers have a lot of conversations around moving towards a cloud first strategy. This is a significant shift.
Before in the last five years that I've spent time in this organization, conversations that we used to have with large enterprises was all about the API's and automations. And we're seeing a shift over the last year where cloud and its impact to the organization is becoming more prevalent. Organizations have started to figure out their API and automation strategy. And now they're looking at the next thing. And the next thing on the horizon is cloud. There is this ongoing desire to lessen the footprint of a data center and lower the costs of ownership for equipment that is constantly having to be refreshed or having to be cycled in data centers.
And for some verticals, they want to move to cloud. Some verticals that we deal with, financial, for example, is slow to move to the cloud. Without a doubt, they still are using the data center model in everything that they design and do. But even then, they are moving to the cloud and we are getting more and more questions on how do you support what is happening in this move to the cloud. If you look at the left hand graph, you'll see what we see in terms of the industry breakdown of who's using what.
Without a doubt, every one of us knows that AWS tends to be the market leader. We see it more in enterprises in the tech or it industries, as well as startups are heavily reliant on AWS. Google is very much the selection of choice in the retail environments for competitive reasons. They retail stores have told us numerous times and numerous retailers that they can't and will not use AWS strictly from a competitive standpoint, which I'm sure we can all appreciate. We see Azure consistently in conversations.
Those customers that are large enterprises that are more globally distributed tend to use Azure more. Also organizations that use more software-based type applications today, people who are comfortable with the global distribution and breakouts in sales platforms like Salesforce or Office 365, we do see an uptick in their use of Azure. And we're starting to see some new ones coming on to the list. I probably asked about once a month about Alibaba in the APAC regions, and we see that constantly coming up more and more as we go.
So there is somewhat of a shift, for sure, and AWS is losing some ground with the large enterprise companies. But we see it happening for different reasons based on the purchase model of the organizations. Given all of these kinds of things, when we talk about hybrid cloud and multicloud approaches, what we see is that it is a ever expanding situation for any organization. The reality is DNS is critical for all of these types of environments to not only work cohesively and in harmony, but it is a underlining protocol that can ultimately be used in a much more effective way, if there is visibility into the different types of clouds that are being used.
Many of the operational practices that exist today in enterprise environments, we see them expanding into the cloud. And they're not directly relational, there are adjustments that need to be made. And we see customers doing that regularly. But ultimately, what we see by far most common in the enterprise customers and in the enterprise discussions that we have is that cloud is being treated as an extension of the networking environment. It's not being treated as something different or something unique or something that sits out on the side.
It's really a capacity play or capability available to organizations so they don't have to continue to invest in data centers and they don't have to continue to invest in the old technology, but really being absorbed and adopted as a new technology front where applications and dev can occur with very little limitation. If we look at our customer segments. And we've just chosen a few segments that we can walk through quickly. We have verticals that have commonality associated with them. But the reality is, when we look at any of the enterprise customers, you're talking about large scale networks, there's millions of IPs that need to be managed thousands of DNS zones that are complicated, and span internal or external environments.
There's changes millions of changes that update an environment weekly, daily for some of our customers. A lot of that is driven by automation. But there's still a lot of hands on manual changes that occur in our customers, by thousands of users. Now that is in a traditional model, which was on-prem or virtual, with the focus and shift cloud, the reality is the immediacy of service and the need to have changes propagate within seconds is even more critical. And as enterprises scale into the cloud, millions of changes a week are going to expand to hundreds of millions of changes a week.
And enterprises need to be ready for that type of explosion, because it isn't something that we normally consider in day-to-day operations. When we look at retail industry, and this is a DNS visualization of one of our customers, who has a fairly standard deployment across their stores. Their setup that is based on one main configuration that is pushed consistently out to the stores. In this case for this customer, a lot of what the stores manages is the Wi-Fi and the IoT environments. There's a lot of data that needs to come back into the central organization of the data center.
But everything at the store level, they wanted it to be dispersed to the store. And when we look at the shift that we're seeing towards cloud, what we're seeing in the retail environments, is that each one of those stores, although they may go back and traverse to the data center, it's the data center that is effectively moving out to the cloud. And so when we actually look at the detailed architecture of a DNS environment for a retail store, as I mentioned earlier, it's probably dominantly Google that they're using as their cloud provider.
There is some private cloud but less than some of the other verticals that we work with. The bulk of the traffic patterns that we see with the retail environment today includes a shift where all of the data to an extent still makes its way to the data center. It's less about the MPLS routing to the data center. Customers or retail stores are being routed via the internet to the data center. Though we see that the cloud component in the retail environments is used in a couple of different ways.
One is really as an expansion of capacity for the data center. When we look at things like retail hot spots are hot dates, back to school, holiday season, a lot of our customers will spin up extra capacity into the cloud to allow for activities to happen. New applications or new purchasing applications or marketing campaigns or any activities that we see especially during those seasons are being housed in the cloud more than in the data center for easy spin up easy tear down. It's very logical for retail to use cloud in that capacity.
When we look at the IT industry or a tech company, it's more dispersed environment that we see with them. We see uniformity, for sure. And there is still some centralized management that exists. But you'll see that there is a secondary set starting to expand. And if you look at the right side of this visualization, you'll see that there are servers being left on the outside are much more close to the edge. And that is an example that we can see where cloud would fit into that area where there still is a connection back to the main infrastructure and network design.
But there's also visibility and transparency and an effective model for cloud out to the edge. When we look at the architecture itself, from a IT standpoint, we see much more Azure and AWS in the environments. There is usually some form of private cloud for corporate or sensitive applications or sensitive data where it needs to be productive. IT industry and software companies that we work with are by far the most cloud forward. They were the leaders when we moved from data center to virtualization.
And cloud is a much more natural step, next step for these types of organizations and their entire environment is more willing to take a risk on that. The third infrastructure type that we see traditionally in our large enterprise customers is large manufacturing. As you can imagine, the complexity increases as you develop size of the enterprise that you work with. And in this case, this company had multiple business lines, multiple resolution paths. There were servers, if you look at the bottom right, in the gray dots that were not connected to the main infrastructure, but were identified in their environment.
Ultimately, it's a complex design that you can see whereby the intricacies and complexities of the network are expansive. When we look at the architecture that we traditionally see in these types of companies, they tend to be a little bit slower to move to the cloud. However, they're also the ones that are most comfortable with direct connection between their offices or their factories or manufacturing facilities to the clouds themselves. And they're also where we see large enterprises focused on dropping MPLS costs and having availability at the edge for services that are reliable and scalable with no downtime.
But always maintaining security and cost as two parameters that they do not move on. All of these different, and we just looked at three complex designs that we see most common in enterprise customers. But all of these different complexities really focus on four areas with problems that we've seen with existing customers. Effectively, the lessons learned areas really focus around visibility into the cloud, provisioning of cloud services, securing the cloud and ultimately making sure that you have advanced resolution. This is not the type of DNS resolution that we saw in the past, there is more complexity as much as that last design was complex.
The cloud introduces way more complexity. From a visibility standpoint, without it, companies are flying blind, and large enterprises including even medium sized enterprises like ourselves like BlueCat, none of us tend to be comfortable with flying blind. You can't control who is doing what. You have a hard time controlling costs. That said, there isn't a need to necessarily go full Big Brother on every single activity in the environment. And organizations need to identify what the middle ground is. How much visibility do you need into the cloud?
How much control is necessary. Is overlapping IP space okay? Is everybody being able to change all records, DNS records a concern or are you comfortable with allowing a single developer to change their environment but nobody else's environment. Without having visibility into the cloud, you cannot make those types of decisions from a provisioning perspective, gone are the days of being comfortable with IT change management ITIL is still prevalent, but the idea that it takes two weeks to get a change propagated into environment is unheard of.
We have requests from large enterprise customers who want changes to be made in seconds. And they're measuring that. And that's one of their KPIs and their drivers. And so what you need to do and consider as you're looking to this transition to cloud, is from a provisioning perspective. You need it to be what I call stupid simple. It has to be simple enough for anybody to do it and automate it enough that you don't have to have your end client worry about what they need or don't need in order to spin up an environment where you still have the ability to control you still have the ability to see.
You still have the ability to secure. If you design a process like that, you limit the exposure that you have of shadow IT organizations being created, because the process was too difficult. Another area that you want to consider is security. When you design how you're going to approach the cloud. And this is something that we've learned that customers had not necessarily forgotten about, but they haven't applied the same rigor that they had with the data centers. You want to consider why you have certain measures in place in the environment that you have, because the application will be different in the cloud, but the why you have chosen to do or protect certain activities or data sets or applications will be irrelevant of where that application sits.
So I highly recommend and all of our enterprise customers have said that when they've defined what the why was for why they were doing security, it was much more beneficial as they designed their security up front for the cloud. Lastly, we talked about DNS resolution and it's more critical than ever with the move to the cloud, you want to ensure that your cloud is connected to your existing infrastructure. And the same way we talked about it in the beginning, you don't want to have this as something completely different than your existing infrastructure.
And needs to be thought of as an extension and not a replacement. At least as you design and as you move towards the cloud. The reality is, is that I'm not sure any large enterprise is going to be a 100% cloud. I have not talked to a large enterprise who has made that stance. The largest enterprise that I've talked to, has said that they're going to go 80% Cloud, but not 100%. Because there still are legacy environments that can't move.
And so because of that, you're going to want to consider your resolution paths. Throughout the change. As you look at things like the hybrid cloud solution and connecting it to your existing infrastructure. How is that going to work? So it's something that you need to consider and remember as you go through this Lastly, we've seen some tips that I can share with you that have tripped up our customers or large enterprises in general. Try to identify who's going to be responsible for why.
The reality is you want to think about this from a operations perspective, not just a design perspective. People often forget that there are a lot of people and teams who want to be involved in the cloud conversations. Or have skin in the cloud conversations. Whether it's the security team, the DevOps team, the networking team, they all have a desire to be part of this conversation, because they're all going to work in this environment. There is a lot of DevOps teams that are working faster than infrastructure and networking teams are.
And so we're finding that enterprises are being caught a little bit flat-footed, in terms of who owns the DNS for the organization and we're seeing more shadow IT organizations spin up but also more control over that is more challenging to get. And there hasn't been a standard across the enterprise platforms. Also, keep in mind that if you define the responsibility up front when something goes wrong, and I wish I could say that it never goes wrong, but trust me, it will go wrong at some point, something will go wrong.
You want to avoid finger pointing. Having a defined plan of who owns what really makes it easier for all of the teams to work together to get things resolved. You want to understand and again, you can't avoid something going wrong. But what you want to understand, have a plan for is who does what, when, if something goes wrong and when something goes wrong. You can't predict the outages but you can prep for them. We have a number of large enterprise customers who do regular disaster recovery, testing and redundancy testing of both cloud and on-prem solutions.
And they'll flip pure to the cloud application. They'll cut the cloud connectivity and see what happens and who it impacts. And so there's ways that you can work with your provider to do that. And there's things that you need to think of and work through in case something happens. What happens if AWS becomes unavailable for a period of time, everybody says organizations are too big to fail. But I can assure you that whether you're a Netflix user or a Facebook user or any of social media platforms, you know that there are times where connectivity or connections are problematic.
Finally, you really want to make sure that your costs don't spin out of control. A prime example is what BlueCat went through at the beginning as we rolled out costs. It's far too easy for people to not recognize where the costs come from. In BlueCat, we have both private cloud and public clouds. We use all three public clouds. And in the private cloud, our support organization, for example, and and some of the other teams that are internal customers of the platform are used to leaving things spun up and leaving things working.
Because we own the infrastructure and the hardware that they sit on, there's an inherent cost, but it's not an excessive cost. And so that bad habit of leaving the private cloud running created significant costs that we weren't focused on when you looked at the public clouds. And it wasn't until our CFO got the bills for the first few months where we have to change our operational procedures and what things we were doing in order to better fit for the public clouds. And you can't have your cost of clouds spin out of control because when you do that, it's much harder to have organizations adopt cloud because it sets the fear into the organization that's going to cost more.
Hopefully this presentation helps provide some insight on how and what large enterprise organizations have been doing as they make the move to the clouds and some of the lessons learned that we have seen through the working with these customers. I want to thank each of you guys for attending this presentation and working through it with me. I also want to thank Ekman Dana on the BlueCat team who helped tremendously to put this presentation together. Thank you very much.
