Steve Tuohy:

Welcome, everyone. My name is Steve Tuohy, Director of Product Marketing at Aerospike. I'm going to be your moderator today for our webinar, App Modernization, Navigating the cloud and your database. Very excited to have my colleague, Matt Bushell, as well as an exciting strong guest from Forrester, Lee Sustar.

Matt serves as head of product marketing for Aerospike. He focuses on the usual product marketing domains, product messaging, positioning, contents, go-to-market and launch. He's played similar roles at a variety of enterprise software companies. He spent years at IBM, was part of launching Db2 10, as well as managing their SMB middleware portfolio. Matt has three degrees from Northwestern.

And Lee, we're going to let him introduce himself. He's got Chicago roots as well as do I. That's a common theme here. So, we've got the Chicago crew together. Lee has extensive experience both as a practitioner and now with expertise in application modernization, IT modernization broadly and the cloud at Forrester. So, we should have a rich conversation with that background.

And with no further ado, well actually, so Q&A, we're going to hold till the end. So, contribute via the tool and we will have Q&A for interaction after some set presentations. The first one will be Lee. So, Lee, take it away.

Lee Sustar:

Thanks a lot, Steve. I'm going to be talking about application modernization trends. It's a big picture, big picture look at it. It encompasses many things including data, which is often brought in secondarily, often to the detriment of the overall effort.

But before we get into this, I'll say a little bit about me. I come to Forrester from financial services. I was involved in cloud and oversight role at the Federal Home Loan Bank of Chicago as the independent security officer there. Prior to that, I was at Discover Financial Services, where I worked on both the infrastructure and cybersecurity architecture teams. And I've been active in the industry for a number of years in terms of trade journalism and technical writing and so forth. So, I've seen it from a lot of different angles. And for today, I want to focus a lot on what we're seeing on the analyst front.

There's a lot of focus quite rightly in the industry about modernizing applications. We've heard a lot about the modernization of infrastructure with cloud native and Kubernetes taking what the cloud providers have been doing for a number of years. Of course now, the applications that we've all been running on have to adapt to that. So, we'll be talking about the why of all that.

And we're going to make the case that it's really not an option these days. I mean, you have to, if you want to be responsive to in private industry in the market or if you're in a government organization, have better citizen or user experience or greater operational efficiency. And you've got to face up to the challenges of data, which I suspect a lot of our conversation today will revolve around in the Q&A.

So, first, in terms of why applications are modernizing, as I mentioned, we've seen the Kubernetes explosion and it's hard to believe now, but it's just only been about eight years since Kubernetes was brought into the open source world and it's transformed infrastructure as we've known it. We've moved away from virtual machines, which were software representations of the kind of technology that's been on our desktops and our data centers for 40 years now.

And the virtual machine technology while created a huge advance in the late '90s and early 2000s was still subject to those old kinds of constraints. So, people began to look at other ways of doing things. Can we use containers, kind of a limited subset of the Linux operating system and create some consistency for developers in these discrete software packages? Containers could be built on a laptop and basically be deployed anywhere. So, you start to see the move towards microservices, breaking down some of the big monolithic applications that had dominated enterprise class IT for some time.

Serverless with the cloud, you have a direct interaction at the level of code in a given application to make use of cloud scale services from any of the big hyperscalers or the smaller cloud players as well for that matter. And that's set the stage for modernization between the cloud scale and the speed to market and so forth. And containerization on the other hand, they were set up to accelerate something that had already been taking place elsewhere, which is a move towards DevOps, a software pipeline, continuous integration, continuous deployment infrastructure as code.

Those efforts really would date back to the '90s, but they start to intersect with containerization from the standpoint of application development brought out at cloud scale, where you can do things faster and bigger like the cloud providers. So then, the question of application modernization flow much more quickly along with that. So, here we have cloud scale. We have cloud speed. We've got these new automated ways of developing. We can limit the dependencies the applications have on underlying infrastructure through containerization that set the stage for application modernization.

Now, if you're building net new, that's relatively easy. We've got containers with everything going forward. We'll build on a containerized platform. We've got serverless we can make use of in our code. What about what you've already have? What can you do to move into the modernization of your existing applications, which are still often the bread and butter?

So, obviously you don't want to just modernize for the sake of modernizing. You might get some efficiencies and so forth, but that can be very vague. But there are very specific cases for modernization, which I'll go through. First of all, the question of automation. I talked about the CICD pipeline, the development tool chain, modern ways of developing application.

They rely a lot on automation. You get a lot more of that if you can modernize the apps themselves. So, the developers need that. And if you can start to move functionality from these highly customized monoliths that are often found in the financial service industry, which makes upgrades and so on difficult, you can start to move functionality from some of those monoliths into more modernized function components that can then work. And that really facilitates the automation at scale.

And this is true for business to consumer and business to business. Now, for the big web apps that came along with the cloud, customer facing, think retail, that was relatively straightforward. There's a clear line to that. Business to business focus for application modernization is a bit more recent because it's more challenging to have a generalized pattern for business to business because it varies considerably by industry.

But the high point of the pandemic crisis and the disruptions of software supply chain really shown a spotlight on what was happening in business to business type of application. So, you saw a lot of effort in just the last two or three years to try to address that.

Software stability. The more automated, the more you need predictability. And when you're talking about cloud scale, that stability is very important because it's no longer a question of a limited impact. If there's some kind of outage or issue or some kind of patching that needs to happen, you need more stability.

So, the more modernized the app can become, the more containerized its functions or broken down into microservices it can be, the easier it is to ensure stability of the kind that people are getting more and more used to in fact in their personal lives with how they interact with the web in many different ways, people need that in their work lives as well.

So, in the business context, software stability need a rising curve of expectations from users of all kind. And my colleagues at Forrester talk a lot about user experience and customer experience being very closely linked in fact today.

Multiple software deployment options, you need modernization because not everything can go to the cloud. There are sometimes latency concerns, things can be on-premises. There's plenty of use cases in which on-premises will be important for data sovereignty and latency in particular. So, you need modernization of applications so that they can actually be deployed in different ways that make sense if you're developing internally for your own organization.

And of course, innovation's only going to further accelerate. Today, we're talking about Kubernetes, but on the horizon we already see WebAssembly, which is using bytecode as computers and the orchestration side of that is still to be developed. But WebAssembly is essentially a way of taking browser-based technology and abstracting application modernization even further from infrastructure. So, that's going to move things along even further.

So, you want to have a modern application development set of patterns and practices in order to be able to take advantage of those changes as they come, even if they're far in the horizon. If anything we've learned over the past several years stands out is that prepare for more change and modernization as a practice as well as an end state is really important in that regard.

We see five patterns to modernizing apps with the cloud. You can replace it with software as a service. We see that in some cases where it makes sense for organizations not to have to have. So, there're traditional core applications like customer resource management or HR types of functions. There's rebuilding. Let's just build anew what we have on microservices in a new context. But there's still going to be a lot of work where you're going to need to make use of what you have and modernize over a period of time.

Sometimes we see the replatform, where you're simply lifting and shifting to the data center. That's less of a pattern today because people are more sophisticated about having several years of experience and can learn from what's happened before. So, typically, the lift and shift using the cloud as a second data center or an alternative data center, that still happens when people have a need to get out of their existing infrastructure by a certain deadline to wind down operating costs and so forth.

But really what we see is a question of modernization. Sometimes people can modernize on-premises and move towards containerized applications. Sometimes they find that easier to do in the cloud. But essentially the modernization, whether it takes place prior to the move to the cloud or after is the dominant pattern that we're seeing today.

And some of the numbers show that people are doing some of all of this because in some cases, it does make sense to replace with SaaS and that's the first call. The next time you might want to look at that, you might say, "Well, we're going to rebuild this net view." But typically we're looking at a mix of moving and modernizing as a way to go and replatform, which is less radical than rebuilding where you can take some of the existing core functions that you have, containerize those applications and there's new technologies to be able to do that.

So, the stage is really set for multiple ways of modernizing and there's a lot of organizations, the cloud providers themselves and other players, third party systems integrators who've gotten a lot of expertise in doing that. So, really what's left for the user is to decide what strategy makes the most sense for them. They have to have a set of priorities and as we'll get into, they have to keep a close eye on what's happening with their data.

And this is a very interesting diagram that we often use to try to get people to break down how they think about their data in their core systems because those core systems can have a lot of data gravity. They typically can go to the core of how a business organizes internally and also how it interacts with its customers, its partners, suppliers of all kind.

So, the systems of engagement on the web-facing side of things, that's more relatively straightforward. It's newer. You might still have to find a way to interact with backend data. There might be mainframe in there someplace. Sometimes these core systems is where a lot of the challenges are that we see. It might be around managing business records of various sorts, which can be for all kinds of reasons, compliance reasons, operational priorities that have to be there. The shared services that are core to any organization are often of course very data-intensive necessarily.

And then there's a lot of automation in terms of business operations that we see where people are looking at whether it's a core banking function in that context, whether it's merchandising, patient journeys and healthcare context, engineering of all kind. Each one of those has its own set of data considerations, data gravity to contend with.

So, when you're modernizing apps and moving things into containers, you still may have to have some reference back to some core data that might be in a legacy system and mainframe. But you're also going to be wanting to look at new types of data that have come with the cloud data at scale, more real-time oriented, what my colleagues will refer to as translytical databases, where you have aspects of both transactional data and analytics.

Those sorts of capabilities really can close the gap between the wrestling with some of the data gravity in these core business functions and with the cloud-scale speed of development and modernization that people are looking for today.

And that's I think sets us up for having a discussion about how cloud and data modernization can move into the new era of cloud on the one hand, scale speed and app modernization on the other and interacting with some of the new database capabilities that are available across the market today.

Steve Tuohy:

Fantastic. Thank you, Lee. All right, so that's a good overview. We're going to send it over to Matt in a second. So, we'll let him pull up some continued slides. That's a lot for an enterprise to navigate. Great broad overview. We at Aerospike have sort of this bias towards the database, so Matt's going to go a little deeper in there, but obviously we see organizations modernizing across their landscape and we see some pitfalls in terms of them incorporating the database lens. So, that's what Matt's going to go deeper on and the floor is yours.

Matt Bushell:

Yeah, thanks, Steve, and thanks, Lee. A lot of what you said, Lee, certainly resonated. And during the introduction in terms of the function of product marketers, one of the things that I enjoy quite a bit is working with customers. And so, I really wanted to bring to life a lot of the things that, Lee, you were talking about and the things that Aerospike sees on a day-to-day basis across its customer base.

But application modernization is part and parcel very core to a lot of our customers. So, I really want to give two industry examples, different industries and explain their journey and what they were doing in terms of microservices, being in the cloud, working with mainframes, their journey, the phases that they did and really contextualize it for you because I think, yes, there are definitely trends. But when you experience it as closely as Aerospike does, we live with our customers, it really gives a strong sense of satisfaction for what we are able to do with and for them.

If you think, if you remind yourselves, Aerospike is a real-time, multimodal NoSQL database, very highly distributed, has a very strong technology stack which we'll cover in some of the benefits and why Aerospike is able to do certain things. But if you really think about what the key is for enterprises when they consider application modernization, they really want to launch new applications and they want to do it at internet speed. So, this is a lot of the agility that you were speaking to.

And no matter the size of the organization, they want to think like a startup. They want to roll these out, yes, in a very DevOps manner. And they do this because of the competitive pressures that they experience. So, even if their infrastructure got them to a certain point, they realized that looking forward, it's a journey forward. And that's why there are certain phases and we certainly covered them and I'll contextualize what those phases are.

And so, the customers, our customers, the Aerospike customers that we work with, they survive and they grow when they delight their end customers. And that's really the bottom line. There's no other metric. There's no other measurable. If you think of Amazon and their wild success, they are completely customer-centric, not even competitive-centric. They really focus on the customers.

But when you're undertaking these prospectively very large undertakings, you really need to manage the risk. You cannot necessarily rip and replace, therefore things may need to be gradual or at least low risk in terms of rolling out. So, again, the DevOps approach and also the infrastructure, the brownfield approach, not always greenfield.

One of the things that modernization should do is it should reduce complexity. And the worst thing you can do is throw more hardware, throw more infrastructure at to solve a problem because that's just going to give you problems going forward. Because of the nature of data and the nature of the world we live in is that there's always more growth. A lot of our customers, 30% data growth is standard in Aerospike customer base. And so, they're very forward-looking and that's why they really scrutinize a data platform such as Aerospike when they do deploy us.

And given the total current economic environment, the whole notion is you can modernize and you can reduce costs at the same time. It's not always adding a lot more coal to the fire and just burning more and more. You actually can have the best of both worlds if the technology underneath will enable that. And so, we'll get into some of that.

So, some of the trends Lee covered, I think perhaps it's perhaps obvious that things are trending towards real-time. Now, every industry has a different definition of real-time. Every application has a different definition of real-time. And I'll cover what some of those service level agreements, those SLAs, what those time windows are because it matters and it depends.

And depending on how fast or the latency and even the reliability, it's not just the average speed, but at the 99th percentile, how fast can your system operate? And that's part of that reliability that I know Lee was referring to before. So, it's can you really execute all the time? Do you have the headroom in your system or is your system that reliable?

One of the other things that being a database we see is people are looking to have different programming models on the same database. They don't want bespoke systems. And that is one of the core tenets of a system like Aerospike. But is it a key value, which is very good for user profile lookups. It's very good for recognizing an identity, but also graphs. Can you program a graph to say, "Is this an individual? Is this their device? Is this a location?" And so, the graphs can pre-map that information and then determine for fraud purposes much faster.

Do you have document or object-based management in your models? This is the lingua franca of a lot of web applications. And so, organizations are looking for different models in the same database to say, "Hey, I want to use you for this application and I want to use you for this application." And the shared service that Lee referred to before is you can spool out the data for different applications, but the applications might be programmed in different data models. And so, this is another trend that we see.

In terms of cloud, Aerospike has customers that are multiple clouds at the same time on the same database within the single cluster. And so, this trend, customers don't want to be locked in. A lot of the major cloud service providers have their own databases, but that does obviously create lock-in. And so, Aerospike is seeing customers wanting to move. We've had customers move from AWS back to on-premises actually and certainly vice versa. And so, we see it's all of the above is a trend that we observe.

One of the other items that we see is that people do start small. But when they get to a certain size and they want to stay in the cloud, they're like manage it for us. So, managed services at a certain size makes a lot of sense. It does free up the individuals for higher strategic operational things than operating or managing a database. So, we do see some of that as well.

And I already talked about low latency of being really ... Aerospike has customers that have been running literally for years on end without interruption. But the secret is can you do it at a meaningful scale? It might be simple if it's a couple of gigabytes, maybe a single terabyte, but we have customers that have grown with us from one million users for an online gaming platform to 40 million users to 100 million users. And they've not had to replatform. They've stayed on Aerospike the whole time.

So, with that preamble, the idea behind it is that we have observed for a lot of modernization there is a phased approach. And so, I'll cover a customer example that went through this has been with Aerospike a long, long time customer, but we do see these phases if we put on screen kind phases of the moon if you will, although these phases certainly don't need to move as quickly as the moon phases.

In terms of a cache phase and initial phase, the idea behind this is really to increase the ability to handle more reads and to access that data in current back office systems. So, if there's higher demand for visibility into say customer data or say product-related information, the back office systems can push to a cache system for very high read levels.

Now, at the end of the day, you can put it in a cache in the first step and then if there's insufficient workloads for scale, they can handle online transactions. And so, it really frees up the existing systems for those transactions. But the read information can be handled upfront and there's high value in that phase alone.

The augmentation phase is if you're building a new real-time system, it could be at the edge and you want to augment an existing say mainframe. And the idea is that the systems do need to sync with each other. So, if you have a customer-facing application that needs to now execute writes to the data system, they can be handled by this new application deployed on a system such as Aerospike.

Now, the back office systems records can be updated now asynchronously with this model. And the back office systems can be, if they fall offline, it doesn't matter because the front end system is still online. And so, this is really getting the best of both worlds in this augmentation phase. And yes, the two systems still synchronize.

And then the ultimate is really a new system of record. And the idea is that if you have new workloads and you want to modern real-time system, the idea is that you can capture transactions that can have extreme workloads. We're talking not just tens of terabytes but up to hundreds of terabytes and beyond. And so, when you have a system of record that is as agile and as fast, you can do a lot more with it. And so, this is both for transactions for reads at extreme scale as well as writes.

And the net new real-time system can be developed with what we describe as the non-linearity of the internet. What's the best problem to have is people all of a sudden want your service and your system can handle the spike in load. So, scale can be not just size of data, but also can be concurrency. And so, this is the approach or a pattern that Aerospike has seen.

So, the first customer example I'd like to step us through is a top three global brokerage. And what they're looking to do is really to do the scaling to scale without barriers. And they, like an online brokerage, want to provide a superlative customer experience, no surprise. And so, they were really seeing that they wanted to have engaging new mobile applications, but they realized that their infrastructure was a bit dated. They had relational databases. They had mainframes.

And so, they saw that their existing cache-based solution with their intraday system of record would require 150 servers, but then they forecast moving to a thousand servers and they thought that was not very practical in terms of time, labor and associated costs with infrastructure. And so, they thought that more important for the business was relying on nightly back processing from the intraday system to their master Db2 or their book of record system. But they found that that was expensive, cumbersome, not user-friendly. And so, they're having some issues.

So, the company made a decision to leverage its mainframe system, which was their compliance system being financially oriented and they want to expose 10 million customer records. And so, the requirement was process 250 million transactions with over 200 million updates a day in positions. And they wanted the ability to update stock prices and show balances on those 300 million positions, but they wanted to do that in near real-time. And so, they needed really to create additional compute capacity to eliminate the data inconsistencies.

And so, with their current systems, yes, they were experiencing frequent crashes. The RAM-based cache systems were really getting overloaded and they took on the order of an hour just to reboot those systems. So, reliability was an issue for them. And again, they were expecting 1,000% data growth. And so, we'll cover some of the use cases and the phases, but that's a little bit of the background. And you see some of the metrics here, the before and after what they're able to handle on a modernized system in this case with Aerospike.

So, in their first phase, the company very specifically, they realized that the legacy system was handling reads and writes, which was a problem because the typical workload was 80% reads of the interactions. And so, that meant that that higher ratio was causing a lot of stress on that legacy system. And so, the solution was to set up Aerospike as really an integration layer and that's what you see in the middle of the slide here.

And so, all the data, the system of record data, you could see all the information is flowing in one direction from the back office into Aerospike. And so, now they're like, "Hey, we could perform over 13 billion queries a day and we can do this incredibly reliably." So, the numbers that they gave Aerospike was 99.9% of the requests were returned in submillisecond speed, pretty strong, pretty strong capability. And so, from this first phase alone, they were able to keep all their intraday event data available, which is something that they were not experiencing before with the crashes.

And so, the data was able to be very consistent across all the clusters and that would help them manage their risk management system. And so, they're able to service again their 14 million user accounts in real-time. So, that's the first phase.

So, the next phase was when they wanted to move to more higher complex as they characterize it right in trading order applications. And they wanted to move these onto modern layers such as Aerospike. And so, when they publish an event to legacy system, they can do so asynchronously. And so the right queue process will update the source system. And so now, you see on the screen here, you see bidirectional arrows between Aerospike and the back office systems.

And so, now Aerospike is really a true operational data store during the day and now, the brokerage is really benefiting from a subsystem when it's this available. And so, again, if there's an outage, the customer experience is not impacted because the layer with Aerospike is always persistent.

So, if you ever wonder, if you think about online brokerages, they offer free trades, if you ever wonder how they're able to do that is because they're able to do margin lending. And this is the exact application that this system is able to execute now. So, instead of once a day doing the synchronization from the mainframe and the Db2 system to the operational layer Aerospike once a day, those calculations are run overnight, all those 300 million positions and the market positions and the history, and they run the risk calculations, but that's based on once a day information.

This system, because it's now bidirectional in very performant, they can calculate risk every three minutes. And what this organization was able to do is they were able to triple from $20 billion a year in margin loans to 60 billion in margin loans. And this is very highly profitable for the company because they make interest on that loan amount.

So, this is a mainframe augmentation phase for the company. But they realized that they weren't done yet and that they wanted to continue to modernize the applications at the edge. So, you'll see they have performance reporting all the way up front now as a real-time application. And when you're on a common platform that can execute speed at the edge and can scale to such as a system of record, you now really can I say uncork even more applications.

And so, the use case that the company wanted was again to retain historical data with over 500 terabytes that is readily accessible. And so, this is a net new application for the brokerage. And so, they're really able to deploy Aerospike globally and they're able to really derive a lot of financial benefit, but it's also operational benefit because of the reliability of the Aerospike system.

So, the other use case I wanted to show and talk about is in a totally different industry. So, this customer is Globe Telecom. Now, Globe Telecom is a major telecommunications provider in the Philippines. They have over 85 million subscribers and they really wanted to modernize their whole business. And it sounds very, very dramatic. But when you think about a lot of telecoms, they've been in place for a while. They are in legacy systems and parlance. They are a communications service provider. It's a bit hard to differentiate on just the communications aspect.

And so, how they fancy and describe themselves today is as a digital services provider. And these services, some of which I'm showing up on the screen, enable them to drive a lot higher revenue or a lot higher value from their customer data. So, if you know any telco, they have operational systems and they have business systems. And so, this is kind of the yin and yang. Historically, a lot of this information is very siloed and the idea is that you want to get it together into one system because it enables you to do multiple things.

Yes, in different parts of the world, a little bit differently, say from the United States, but a lot of these organizations, a lot of these telcos have really their mobile network operators, they have applications and they provide services on those applications so they could make offers for customers if they're in a specific location. They can offer rewards for certain loyalty and so forth.

Now, one of the beautiful things that you see here is the geospatial tracking. The Philippines, if you're not aware, is really an island nation. There's several islands. And so, the offers need to be very specific because a lot of the landmass is small and surrounded by water. And so, if an individual is in a very, very specific location, they can identify what businesses are local on the Aerospike system because it's of our native geospatial capabilities. So, this is part of the multimodel aspect to Aerospike.

So, in terms of what their infrastructure looks like, they would describe their platform as having three main components. They have a streaming transaction event processing system, which makes sense given the live dynamic nature of mobile telephony. And then they have what they describe as a central hub for collecting all this insightful information on customer experiences and then they call it a unified marketing tool to really make decisions on that pool of information for certain things.

So, they describe this as an open architecture and they have multiple areas. They have ingestion. They have streaming. They have event orchestration, real-time reporting analytics and so forth. And so, at the end of the day, they're able to synchronize to data lakes, you can see on the screen there. So, this is a microservices-based architecture. They were able to process over one million data transactions per second. Before, they were under 400,000 and they have over 2,000 pods or microservices operating.

And so, these microservices are very part and parcel to their industry. So, they do as the screen is indicating, call drop detection. If an individual, a lot of times they will do what's called a top-up. So, every month, they may add more money to their account. And so, given that, that might be indication to give them an offer to say, "Thank you for topping up for the fifth month in a row. Would you like a free offer to what have you?" It could be an online subscription or service.

And so, this has really made them much more agile and much more scalable. And so, if you think about the data flow in this case versus say a brokerage, their data sources, what are they? They're things like interactive voice response, IVR systems, as well as web and mobile. And then they draw this information in that data preparation layer. They draw it in, excuse me, a raw format to Kafka, to different Kafka topics and then it's treated. The data is standardized using Flink and then it pushes back to Kafka and then for up into the cloud and then they can operate or make decisions in real-time on that data.

So, again, they really became a microservices architecture on Aerospike. And so, that real-time reporting and dashboarding on different systems really enables them to also monitor how business is going. So, if you were to hear the replay of this customer, they do talk about the reliability of the Aerospike system. They're like, "With Aerospike, I sleep at night." One of the common quotes that we hear is, "Aerospike is the forgotten system," because it doesn't crash all the time. There's no wakeup calls to the operations team or to really say, "Hey, system crashed. We need to reset up the cluster."

And so, that's part of the beauty of the Aerospike infrastructure. It is very robust. So, these are really the two use cases that wanted to relay to bring some color to application modernization.

I think these customers are really building. They really derived a lot of value today, but they see the growth potential with a system such as Aerospike. They have this mindset that they can do more with the company given all the applications that they're now adding in a more conservative industry as well as one that's looking to gain a lot of users by the tens of millions.

And so, yeah, cloud is a part of it. It's not always the central premise, but they also save a lot of money with Aerospike. One of the things that I perhaps neglected to mention is in that brokerage, they were saving over $10,000 a day because they were saving on MIPS because they were just offloading all that read workload from the mainframe. And so, I excited that the server reduction was going down from 150 servers to 10, but they were forecasting on the old system going out to 1,000 servers. And so, that six X server would've been just from 10 to 60 with Aerospike.

So, total cost of ownership, you can get a lot of the best of both worlds with a good system if you really plan it out. And yeah, I think that's really what I'd like to share upfront, Steve.

Steve Tuohy:

All right. Well, thank you both. A lot of great content there. I'll boil it down. We will incorporate some conversation and questions at this point. But for me, Lee gave a great wide overview that's applicable to really any large enterprise with legacy applications and is going through this journey. Matt hit some specific examples that he's seen at Aerospike really with that database lens. And so, I think I'm guessing a lot of where the questions are coming from and we're going to hit discussion is bridging that gap and sharing perspective.

So, the first one that I've got here, it's to both of you, but we'll start with Lee. So, Matt talked about some examples with this three phase, the cache, all the way to system of record for a database, and you touched on it, Lee, but you also talked about prioritizing workloads, et cetera, over the course of this journey. So, A, does Matt's example resonate, and B, what other ways are organizations prioritizing?

Lee Sustar:

Yeah, I think the example of the brokerage firm was very interesting because it got to a lot of different questions and it was about how you can modernize with mainframe. We talk a lot about modernizing mainframe and sometimes, people take that to mean that you're going take things out of COBOL and you're going to put them somewhere else. And there's a place for that sometimes that might need to be done, but you have critical systems and critical data that's running on these systems today and the question is not it's performant. It's doing what it was designed to do. The question is how do you make it available in changing conditions and dynamism?

So, I thought that was very interesting in the way that you could have an intermediary type of layer that would provide the real-time and the focus for sophisticated needs that by anybody's definition would be modern and yet you would still have a mainframe that was playing still a foundational role there. So, I think that's a good example of how modernization has to have that broad sweep where you can have ... Doesn't mean replacing everything you have. It means making the most of what you have with the tools and technologies that are becoming available today in ways that deliver what your organization needs and what customers or partners are expecting.

Steve Tuohy:

Thank you. Okay, next set of questions. I'm going to combine a few here, but I think everyone out there listening to this is they're on a journey in some way and you hear COBOL and mainframe and you're like, "Well, that's not me," or "This takes a really long time." So, I guess to both of you, how would you characterize how long these journeys take right at the long end and the short end, just for people that are ... No one's sitting there, "We haven't started to modernize yet and we're waiting for this webinar to hit the go button." But just some perspectives on whether this is never ending and continuous or whether people can put realistic timelines behind it.

Matt Bushell:

Yeah, I'll go first. Obviously, the short answer is it depends. Aerospike, one of the other things that came to mind was there's other ways that people attempt to do what was described in the brokerage use case. And so, in some regards, Aerospike is able to consolidate servers because of its technology when other people try caching with in-memory systems that are not as reliable, that require a lot more infrastructure.

We have customers that have said, "Yeah, in three months, we're able to knock out 450 servers and go with 60 with Aerospike." And it took three months and they were up and running, up and ready. We have another customer in Europe that is looking to shutter two of its five data centers and it's an order of magnitude greater. So, that is 6,000 servers of a in-memory system to one-tenth of that on Aerospike. And so, that does need to go in phases and in steps. And so, it can depend and it can be very fast.

What I will say is that the changeover that the Globe Telecom experience, they said it, they were surprised how fast they were able to make that switch. So, once they did their QA and testing, the loads, they just shifted them over. They saw that it was very stable very quickly, so they were a bit surprised in that shift over. I can't say that's always the case, but you've got from a very significant deployment taking only three months, bigger ones can take more time and other ones, once the application is ported, moving or migrating the data is not always that difficult at all. So, it can be just a matter of weeks.

Steve Tuohy:

Lee, want to chime in beyond that?

Lee Sustar:

Yeah, I think this question of phasing is important and also prioritization. And sometimes, the things that you would really like to modernize the most are less amenable to it. So, you might want to find something that's slightly down the chain in which you can actually model success. It's still important. It's not the least important, maybe not be the most critical, because there's some operational risk that might be associated with that one and in any case, so let's break this down element by element.

And some applications do lend themselves to containerization. And then the question is how are they going to interact with the data? Is that going to remain consistent in ways that people have expected before or in the new ways that we're looking to modernize that? So, it's very much of a strong scrum master good execution, iterative approach that we see as successful. The tools are there. There are more people out there, more vendors out there of various products and services who are poised to help do this.

But in the end, you have to be able to set those priorities yourself and approach those on an incremental basis and have a clear idea of what your goals are. And if that's done, if you can model success with some early patterns and show that meaningful difference, then the organizational barriers to some of these things could be ameliorated.

And when I say organizational barriers, I think sometimes there's a stereotype that people are against modernization because it's new and they're setting their ways and so on. I look at it differently. I think people are tasked in any element of information technology, of keeping things going like let's make sure things are up and running. Let's get the software development out the door.

And so, that's their day-to-day priority. If there's something that they perceive as disruptive to that, that can lead to some questions and concerns. So, breaking this down into an iterative process and actually having some constituencies around that to deliver really makes a lot of difference in terms of success.

Steve Tuohy:

Thanks. Okay. Again, I'm going to weave a couple different questions that came in together, so maybe you can choose the one that you want to tackle. So, I think this probably makes sense for Lee to take first. Matt talked a fair amount about real-time use cases and cutting-edge ones, frankly, largest brokerage, one of the largest telcos in the world. What about the organizations out there that says, "Well, we're not real-time. We can take our time with this."

So, I guess the question is, as it came in, are organizations with real-time use cases just inherently apt to take a different approach to modernizing. Maybe in terms of the options you laid out in the survey work you've done, Lee?

Lee Sustar:

Yeah, I'll answer that-

Steve Tuohy:

There's one-

Lee Sustar:

Sorry.

Steve Tuohy:

And I'll tee up the strangler pattern, too. That was something someone asked in this as well.

Lee Sustar:

Right. Well, I'll start with the strangler pattern first, then come back to the bigger picture. The strangler pattern is a little bit related to what I said before about setting priorities. So, sometimes the easiest way to start is to say, "We're not going to develop on this platform any further. We're going with Kubernetes and cloud native. We're not going to invest anymore in virtual machine technology. We're not getting rid of it. We're not looking at rip and replace, but we're going to actually move some of this functionality."

Or if it's a big monolithic app, which often characterized financial services, "We're going to move some of these customer-facing elements into containerized applications that can interact with that data. And we're going to do this over a period of time, because that monolith is still performing. It's still doing what we need. If there's a mainframe in there as we saw from the brokerage example, that can work as well."

So, that strangler pattern is one we see that's often. And sometimes, that's driven almost by, it can be driven pragmatically only, but I think having a more strategic approach to that about not deciding where you want to stop, but actually where you want to develop new capabilities and bring those two things together.

On the question of real-time, I'm going to put even a wider lens on that, maybe get back towards real-time. What we've seen with the generalization of Kubernetes and we're at late stage adoption now perhaps, where it's become the de facto standard. Even if the majority of your workloads are not yet in cloud native and Kubernetes, it's still the reference point now for where you want to be, hence the strangler pattern and so forth.

So, what that's doing is putting into the hands of people who don't think of themselves as especially cutting edge. All kinds of power that was just a short while ago available only to the large enterprises, to people with huge IT budgets and so forth, that's now available to organizations that are using a lot of off-the-shelf software and focus and so on. They can make use of that today.

And I would think real-time data is part of that. It might've seemed exotic some time ago, but in fact these systems are available now and actually using database as a way of integrating some of the key systems as opposed to try to wrestle with a lot of the multi-cloud integrations, which is still challenging. If you can get your data where you want it, when you want it in a performant way in terms of access and database capabilities generally, then that's something that's going to be attractive and accessible to organizations I would think of the broad swath of enterprise class IT.

Matt Bushell:

Yeah, I mean, we saw Db2 up on the screen and the brokerage and these systems exist. And what Aerospike has found is that customers realize what they can do. It's almost like the art of the possible. Once they say, "Oh, my gosh, I can get 10x the data and actually more reliably in a smaller time window," and so it actually uncorks new use cases.

We have a customer that actually charges, they have an identity management solution and they actually charge more. They have two tiers of latency because they service financial services organizations. They actually charge more. They're able to have a premium service now because of how reliable and how fast Aerospike is for their service. It's something that they wouldn't have foreseen before.

The other example also in the fraud arena is everything runs on, we hear about ChatGPT, which is AI, but really machine learning is part and parcel almost everywhere. The recommendation engines and those offers are all spooled based on machine learning. And those algorithms aren't always super duper complex. They're tested with a lot of data.

And so, when you have something like fraud, if you can levy 10 times the amount of data in the same time window, you can be way more accurate. And that's the time windowing, if you will. That was the brokerage example where they said, "We can compute risk now in three-minute increments versus in daily increments." It gets better business outcomes. And so, it's not so much that there's no need for real-time or people don't think about real-time. I think consumers, they see something in one industry and they expect it in another industry. And that's what Aerospike sees.

Steve Tuohy:

Great. Okay. I'm just looking at the clock here. Got a couple more minutes. I think I'll just moderate one more question, but open the floor to you all to respond to each other if there are things we think we missed. And I'm going through my list of what questions have come in and what you covered. Apologies if you hit this, but Matt, the two examples, what role did cloud play there? So, maybe I'll lead that. And Lee, obviously you talk cloud is just embedded into this modernization, but why don't I steer the wrapup towards cloud and I guess the cost implications of cloud, if you want to double click that direction. Thanks.

Matt Bushell:

Yeah, I mean the short answer for Globe is that they run on AWS. All those 2,000 microservices, those pods all on AWS, so they're very happy with Amazon Web Services.

Steve Tuohy:

All right. Lee, I know you talked a fair amount about cloud. Any final thoughts or areas you think we didn't touch on in the final two minutes?

Lee Sustar:

Well, I'm going to pick up on something that Matt said about how you see something in one industry and you expect it in another. That's a big part of what a lot of my colleagues at Forrester deal with every day because this constant effort to try to for digitalization and user experience and customer experience, it's becoming very interactive and mutually reinforcing.

So, meeting those, a rising curve of expectations is really on top of every organization's agenda today. Even organizations that you wouldn't think necessarily would face that, they're more business to business, they're not consumer facing. Nevertheless, the expectations that people have about what's going to be available to them, how to interact with their data, how apps work, all those expectations are being raised. And usually that does come down to scale and speed and access to data. And that's where cloud comes in.

Now, again, I'm a cloud analyst. That's what I do all the time. Cloud is not the answer for everything. There are industries. There are trading firms which have mapped out things in the last millisecond. They're more comfortable doing it themselves and putting in the cloud. And there's data localization and sovereignty concerns. But in terms of scale and access to data and so forth, this is where database technology and cloud have had a very strong mutually reinforcing dynamic.

And that is, AI is obviously at the top of everybody's talking points today, but behind that is all the data that people are making use of in maybe less flashy ways, but are increasingly the core of how businesses and consumer organizations operate today.

Matt Bushell:

Yeah, I mean, over 40% of Aerospike's customers run on AWS alone, let alone the other two service providers. In Aerospike, we have customers that are amongst the largest of all of Google cloud's customers are actually Aerospike customers. And so, it's part of the strategy. It's for modernization in some regards. Selfishly, you could say anytime you're looking to deploy Aerospike or system that competes with Aerospike, it is a form of modernization because of what you're looking to do with data. And so, the cloud is without a doubt a strong part of it.

Steve Tuohy:

All right, well, we're going to have to make that the final word. Thanks so much to Lee for joining Aerospike folks here, and Matt, thank you. Great work. This will be recorded and available. Want to wish everyone a great day.

Lee Sustar:

Thanks very much.

Matt Bushell:

Thank you.