Google Cloud Partner AI Series | John Maddison, F5, & Prerak Mehta, Google Cloud

Clips
More from Google Cloud Partner AI Series

John Maddison

Chief Product and Corporate Marketing Officer

F5

Prerak Mehta

Director of Customer Engineering, Strategic AI & ISV, North America

Google Cloud

play_circle_outline Overview of the partnership evolution between F5 and Google Cloud since 2018.

play_circle_outline Google's advancements in security and its importance for multi-cloud environments.

play_circle_outline Contribution of F5's orchestration to secure multi-cloud traffic and optimize performance.

play_circle_outline Exploring Future Opportunities in AI Security: F5 and Google Unite to Combat Evolving Threats with Innovative Collaborative Solutions

Info
Transcript

John Maddison, F5, & Prerak Mehta, Google Cloud

John Maddison

Chief Product and Corporate Marketing Officer F5

Prerak Mehta

Director of Customer Engineering, Strategic AI & ISV, North America Google Cloud

In this Google Cloud Partner AI Series interview, F5’s John Maddison, chief product officer & corporate marketing officer, and Google Cloud’s Prerak Mehta, director of customer engineering for Strategic AI & ISV North America, join theCUBE’s Rebecca Knight to unpack how the F5–Google Cloud alliance (dating back to 2018) is evolving for the AI and agentic era. Maddison outlines F5’s journey from application delivery and load balancing to expanded capabilities across WAF, cloud services and NGINX – now moving toward cloud-native offerings within Google Cloud’s ... Read more

explore Keep Exploring

What has been the evolution of the partnership between F5 and Google since its inception in 2018? add

What strategies do organizations use to enhance their security and manage risks in multi-cloud environments? add

What are the key advantages of using F5 for web and API security in multi-cloud deployments? add

What are your thoughts on the latest acquisitions and SaaS offerings related to AI security and partnerships in that space? add

bolt Powered by CUBE AI

John Maddison, F5, & Prerak Mehta, Google Cloud

search

Rebecca Knight

>> Hello everyone and welcome to the CUBE Studios in Palo Alto and the Google Cloud Partner AI Series. I'm your host, Rebecca Knight. We've got two terrific guests for this next segment. We've got John Madison, Chief Product Officer and Corporate Marketing Officer at F5. Welcome.

John Maddison

>> Good to be here.

Rebecca Knight

>> And Prerak Mehta, Director of Customer Engineering, Strategic AI and ISV North America at Google Cloud. Welcome, Prerak.

Prerak Mehta

>> Thank you for having me.

Rebecca Knight

>> So I'm going to start with you, John. For the viewers who are unfamiliar with F5, why don't you tell them what you're all about and what you do?

John Maddison

>> Yeah, F5 is a company that's almost 30 years old. And like all companies, they started in the garage because all companies start in a garage.

Rebecca Knight

>> All good companies.

John Maddison

>> Yeah, all good companies. And a couple of students decided that the applications weren't scaling and so they built a system to make that happen. It's called load balancing, sometimes it's called application delivery controllers. They came up with a name, which in my badge there, F5, because F5 I think is the highest tornado state. It's the most fierce tornado in protection-wise. And so that's why they came up with the name. We're headquartered in Seattle. Last year, around $3 billion, about 7,000 employees. We've expanded from our traditional load balancing into WAP and cloud and NGINX for modern applications.

Rebecca Knight

>> Excellent. Now I want to put you in a bit of a reflective mood because the partnership between F5 and Google dates back to 2018. How has that partnership evolved into the deep alliance that it is today?

John Maddison

>> Well, obviously Google Cloud is a very important company out there and many of our customers are joint and they want to see us working together on new capabilities. I think obviously we're in the marketplace, we have about 50 different products that work there. But also some of our services, like our bot services, for example, work in Google Cloud. And then we're starting to work on what I call new cloud native services as we go forward. So it's marketplace. It's putting some of our services inside Google Cloud and it's working on new capabilities as well. So it's a very, very strong partnership.

Rebecca Knight

>> Prerak, similar question for you. From your perspective, what makes this F5 partnership so critical for Google Cloud customers, especially as they move into the AI and agentic AI era?

Prerak Mehta

>> Great question. So Google just celebrated its 27th birthday.

John Maddison

>> Oh, we're older than them.

Rebecca Knight

>> Baby.

Prerak Mehta

>> Yeah, we're older. I'm older than Google. We all are, I guess. But it is fascinating to see the leaps and bounds of progress Google has made over the years in various technology fronts. One of the things that near and dear to me is how it's built. Google Cloud initially, the infrastructure below it was built for this global search and it was globally scalable. The way it was like cannot fail. It had multiple fail-safe processes. But when they decided to go in public cloud area, one of the things that they had to consider is how the enterprise customers run the application. Most of them tend to have a fallback policy, which in today's date and age is called multi-cloud, or they diversify their risk not to be in one place. One of the things Google does really well is how we have built our security foundations. We made a lot of advancements. There's a lot of patents in the industry. We have a big slew of researchers that dedicatedly work on this. We have a team called Project Zero that's dedicated to it. But that is all within Google to secure Google traffic and protect Google from nation state active threats and what have you. When I look at the partnership with F5 and similar partners, we think about our enterprise customers and what they want. And oftentimes where they shine is in the area that is outside of the Google perimeter. We can secure very well within the perimeter. The value of this partnership really shines when we are looking at multi-cloud environment, when we're looking at on-prem and Google hybrid installations and what have you.

Rebecca Knight

>> So, John, I want to ask you about that too. In your conversations with leaders, what are some of the biggest security and performance challenges that they're facing right now, especially as it relates to the changing attack surface?

John Maddison

>> I think there's two topics that come up a lot. One, as he just said is hybrid and multi-cloud, multi-cloud and hybrid. The second one obviously is gen AI. So on the first one, because of regulation, because of data, because of multi-cloud, that's here to stay, I think. And I think companies or enterprises we speak to have that hybrid and multi-cloud and they're going to have it for a long time. So you have to be able to deploy in SaaS and hardware and software and native, which is why NGINX as a service is going to be really important to us. And the next one is AI. We split it up into four conversations. One is around delivery. There's just a huge amount of data moving around. And so we use our low-balancing technology in front of data stores or RAG or whatever to make sure that data moves around. There's what I call security for AI, and we did a recent acquisition of Calypso, which protects models for example. There's AI for security. That's putting security like AI assistance and technology inside your products. And so AI is not just one thing, it's three or four things which impact both our products, the way our customers deploy, and the way they want to make sure that their models and their AI infrastructure are protected.

Rebecca Knight

>> Because the bad guys are using AI.

John Maddison

>> And the third one, sorry, the fourth one I didn't bring up is AI increases the attack surface. And so not only now have you got models, but eventually agentic and agents out there. So the attack surface itself has increased. And then the types of attacks like prompt injections for example, will be different. So yes, there's a lot of conversation around AI, but there's a lot of securing of AI that needs to happen.

Rebecca Knight

>> Prerak, how do you see the multi-cloud operational complexity evolving? And why is it so hard for customers to solve it on their own?

Prerak Mehta

>> Think about, I think John said the magic word AI. You don't live in 2025, you don't think about AI. One of the key priorities for Google this year is investing. I think in other words, we decided the folks who will potentially win the AI race are the ones who kind of own the models, own the chips. So when you think about the very foundational level, we made significant strides in both of those areas. This is not new for us. We've been using models internally for more than 10 years. This is great within the Google Cloud landscape. But as we think about, as you mentioned, multi-cloud, one of the key things is securing traffic between if you're thinking about a different cloud in Google Cloud or an on-prem data center Google Cloud, corporate offices. Securing is one thing. You'd also think about performance, right? Because I mean security and encryption can take a little bit of an overhead. I personally have been an F5 user. I was a consultant prior joining Google, and I used F5 in all of my e-commerce customers. And the one thing that they stand out for, and that's still true today, is they can provide very differentiated web and API security when thinking about multi-cloud deployments. Especially in today's date and age, think about a large bank. Banks have significant amount of data. They give access to the analysts for multiple data sources. And a lot of this data is confidential in nature, highly sensitive, a lot of API information. As you mentioned, the biggest threat we have today securing AI models is from other AI. And I think when you are interacting with multiple entities on-prem, various clouds, I think you need a security layer that is looking from a layer above end-to-end between endpoints. And I think that is something they do really, really well along with load balancing. So I mean, three things I'll highlight is F5 load balancing, firewalls, and the performance that they have on the securities in parallel.

John Maddison

>> Yeah. One interesting thing is what I call the F5 data plane, which is a layer seven data plane reverse proxy. Unlike a packet network, can look at layer seven and the layer seven language for AI models or tokens. And so we see a future where there's a token engine, not a packet engine that's making some decisions around routing or security going forward.

Rebecca Knight

>> And that could really change things for customers.

John Maddison

>> Yes.

Rebecca Knight

>> Yes.

Prerak Mehta

>> We partnered with F5 and NetApp to create a solution to secure all the LLM traffic between different endpoints, including APIs to provide a customer's secure and performance solution. So I mean, I think the unique value add, as I mentioned before, is when you have a vast amount of data and disparate data sources, it is very hard to secure from one particular place. And solutions like F5 kind of standout.

Rebecca Knight

>> And that applies to so many different industries-

Prerak Mehta

>> It does....

Rebecca Knight

>> and enterprises that have data in disparate places.

Prerak Mehta

>> I talked about financial services, but we see F5 in retail quite a bit as well, in media space too.

Rebecca Knight

>> So, John, one of the things that Prerak said is such a strength of F5 is this load balancing across AI models, which sounds exceedingly technical. But it's quickly becoming a business critical issue for enterprises adopting generative AI. Can you talk about that a little bit?

John Maddison

>> Yeah, because you just mentioned the data, and so just think of the amount of data that's coming out of some of these storage systems. It's enormous. And so the ability to do AI data delivery, which is another form of load balancing, but just another name for it, is becoming critical. And make sure that not just in front of data stores, but in training models and inference models, for example, in RAG, the data's becoming critical. And moving it around at high speed with security is going to be very important.

Rebecca Knight

>> Prerak, what is the role of Google's Gemini models, TPUs, and Vertex in the story? And how does F5's orchestration help them work together and become more powerful and also cost-effective for customers?

Prerak Mehta

>> Great question. So if you think about Gemini models, Gemini is large language model, we call them LLMs. Companies focus right now is to create diffusion models that is basically focusing on audio, video, text to speech, what have you, that kind of stuff. And then you have reasoning models that are more like doing mathematical computations with hedge funds, like high speed transactions and/or scientific models like molecular biology, like working on drug compounds and stuff like that. So I think the near term investment is to refine Gemini model and focusing on upcoming models, getting better in that space on those three areas. The thing about TPU, I think one of the biggest constraints today in chips is power. So the advancement in the chip with the TPU V-7 Ironwood coming out, I think the focus is to shrink the footprint of power and physical footprint of the chip itself while creating 48X performance to the prior version. So I think we are making significant strides in that chip space and in the LLM space. But as you think about our partnership with F5, and from my end customer standpoint, they will use multiple models at any given point of time to do different things. While we have great models, we also have the notion of Fortex AI, which is an AI platform. We have multiple models in it. We have a model garden. We partner really well with our competition like Anthropic. We have some open source models. And customers use all kinds of models to do all kinds of things. They might use Gemini, get an output and send the summarization to a different open source model that's running on a GKE cluster. That might be running somewhere else. To secure that traffic, we rely on the likes of F5 to provide performant, bidirectional, like high-speed data transfer between models and also secure end-to-end customer data.

John Maddison

>> And I think Google Cloud has realized, as with hybrid multi-cloud that you need different models in there. So you're opening it up to different partners. And for us, we did a recent acquisition of a company called Calypso AI. It's actually based in Ireland in Dublin. And they'd really focused on protecting models, not just a specific model, but up to 12 different models out there. And in fact, they run about 10,000 attacks each month against the different models and give them a ranking and then put it into the product. So we see long-term. That's already working inside Google Cloud, but some of those other models that the companies are using, then maybe that's a product offering we can provide.

Rebecca Knight

>> So, John, NGINX as a service is about to launch for Google Cloud. What makes this offering different? And why should customers be paying attention?

John Maddison

>> Because I think you can put all your products in the marketplace and load them up and that's fine. But what's so powerful about Google Cloud and these applications is it's cloud-native. It works natively with everything inside there. However, sometimes they maybe need a different application or different features and different functions. So NGINX as a service is a very well-known load balancing application as well as web server. And so yes, it's available on NGINX on the marketplace, but now we've integrated as a cloud-native SaaS application right inside Google, works alongside their engineers. So that the customer can now decide, yes, they can still use the Google load balancing, but if they want something that works maybe across multiple clouds or some of the features of NGINX, which they're used to on-premise or whatever, they can now spin it up as a cloud-native just as easily as the native load balancing itself.

Rebecca Knight

>> And does it help them deliver value more quickly, would you say? Or?

John Maddison

>> Yeah, I think it's just different applications. I mean, again, I think for customers, they have a choice. There's the native tools they have and they're great, but sometimes they want certain features that another application has. But they don't like the ability, the mess of spinning up hardware and managing themselves and patching themselves. They want it to be cloud-native, they want it to be SaaS. And that will be the power of the new offering we've got together.

Rebecca Knight

>> So as we wrap up here, whether it's NGINX as a service, which is soon to launch, or Calypso AI, what do you both see as the most exciting opportunities for this partnership to expand in the next 12 to 18 months? How about you? Why don't you start, Prerak?

Prerak Mehta

>> I think John mentioned it. I'm excited about their latest acquisitions and the latest suite of SaaS offerings. I know we are also working on some of this in this space. Like Model Armor is one of those things that's coming out. But I think it's an exciting space and I think there's a lot of work still needs to be done to secure AI. For me that that's the biggest area of partnership with F5 is as we can explore and come up with new models. As agent space and agents kind of mature and start doing more than basic things, I think we'll need these tools to be very smart and very sophisticated to prevent bad actors from penetrating into our systems. So I think F5 and Google will play a critical role there in creating joint solutions.

Rebecca Knight

>> And fighting our adversaries.

Prerak Mehta

>> And fighting our adversaries.

Rebecca Knight

>> Yes, John, how about you?

John Maddison

>> Yeah, that's a great one as others as well. But I just like the attitude of the Google Cloud teams. It's like, "Hey, yes, I can do everything myself I want, but they're offering a customer choice." And so we like working with them because they're open about that. Yes, they have their own products that sometimes we compete with. But they're open in saying that if the customer wants this, then I need to make sure it's inside my cloud. And that's why we like working with Google Cloud.

Rebecca Knight

>> Good attitude. All right, I like it. John and Prerak, thank you both so much for coming on the show, a really fascinating conversation.

John Maddison

>> Thank you.

Prerak Mehta

>> Thank you.

Rebecca Knight

>> And thank you for tuning into this edition of the Google Cloud Partner AI Series. We hope you'll stay tuned for more.