Moeez: …with the last session of the day, and ultimately the last session of the performance boot camp of this year. We just don’t have another session, we have a very special session ahead of us, and it’s about CDNs. And the title of the session is the global website hack, how a CDN makes you local everywhere. And who better to call on stage to talk about CDNs than the people from Cloudflare. So I’m honored and excited to call upon stage the duo from Cloudflare. The first guest for this panel discussion is Than Oo, I would like to call him on stage now, and he is the senior manager of solutions engineering at Cloudflare, helping enterprise digital native companies tackle complex security and performance challenges. Thank you so much for being here on this panel. The other guest that we have, again from Cloudflare, is Trever Jackson, he is the senior systems engineer at Cloudflare, focused on building scalable, secure cloud and AI/ML platforms, with experience across Google Cloud, AWS and VMware. I can’t thank you both enough for being here on this stage and doing this panel discussion with us.

I will dive straight to the first question, and the first question is for Trever. For businesses expanding internationally, what’s the biggest performance gap that a CDN solves?

Trever Jackson: Yeah, first of all, thank you very much for having us here. When I think about CDN in 2026, it’s still fundamentally the same problem we’ve had, right? No matter what we do, you can’t beat the speed of light. So when you have distance between your end users, the browser, and the website they’re trying to reach, it’s going to cause problems, the further that they are, right? Latency is going to be a killer, GP could be a killer, there’s many things just in between the two. So CDNs are kind of a prerequisite for modern applications these days. The closer you can get the user into a PoP location and actually have an exchange where you’re doing a handshake or connection and then ride a CDN instead, you’re going to have a much better experience for your end user.

We don’t build out our environment like, you know, like the hyperscalers where we have these huge regions everywhere, right? We decided to build a massive network, and Cloudflare is incredibly, incredibly large, about 330 cities right now we have a presence in. So odds are we have a presence where your user is today. So getting that user onto that CDN is going to give you a much better experience and reduce all the speed issues that you’re seeing right now. It’s kind of a prerequisite honestly.

Moeez: Awesome. I want to talk about configuring the CDN, and my next question is from you. What are the most common mistakes companies make when configuring a CDN for global reach?

Trever Jackson: Yeah, definitely. One of the biggest mistakes, and I guess the most obvious and easiest one to do, is just set it and forget it, right? You should always fine-tune your CDN to provide the utmost importance and performance for, you know, where your user base is, what kind of content you’re delivering, right? Typically also out of the box, most CDNs aren’t going to cache HTML or things that aren’t cacheable or static out of the box. Obviously you shouldn’t cache anything behind admin and login pages, right, so be wary of that as well. But I mean honestly, you should be able to cache certain API responses, you should be able to cache dynamic content, deliver video files, things like that with chunking, right? So that’s one of the things I would say, optimize it, configure that to the best performance ability.

And then one other thing is, honestly, falling into, when I talk to customers a lot, a lot of the times we fall into the global sort of like average trap, right? The average might not really matter, right, because if the performance you’re trying to optimize is specifically like 95% of your users, or specifically in regions, when you look at a global average, a lot of that might be skewed. So what I recommend customers do is break down by regional focus. And in addition, I would say you should really push for percentile-based monitoring. P50, the median, is the experience of your typical user, right, I would really go in to measure that, then you can hit P95 and P99. But yeah, it’s a common trap to look at global averages as well. And the last piece is DNS, right? DNS is a foundational piece on the internet for a lot of these applications, websites, no one’s memorizing IP addresses, and, you know, try to use a performant DNS provider, because if you fall behind and use a slow DNS provider, that’s going to add additional latency for that request from end to end, right?

Moeez: Awesome. Before I move on to the next question, I have a few announcements to make. One of them is that if you have any questions from our guest speakers over here, feel free to post those questions on the chat and we will try to answer those questions by the end of the session. Another announcement is that we have an official subreddit on Reddit that you can check out any point in time and leave your feedback around this event on that subreddit, so our team will make sure to drop that subreddit link over here. And the final announcement is that, especially for the people residing in Mumbai, we are hosting a Cloudways developers and agency meetup in Mumbai on March 28th. So if you live in Mumbai and you’re watching this session, I would highly advise you to register for that event, and our team will make sure to drop the link of that event in the comment section, so you can go and register for that event and attend that event as well. So yeah, that was the announcements.

Trever, I would like to talk about some dynamic and static content now. How do you decide which assets to cache at the edge and which to serve dynamically?

Trever Jackson: Absolutely, so, great question. Actually a lot of the basics gets overlooked, so when we talk to customers they’ll ask those questions, what’s the best approach, strategy, right, static versus dynamic. Typically it was always the same answer, right, you would have your images, your CSS, your JavaScript, fonts, things like that always cached, that’s still pretty much best practices today. But again, these days the edge is becoming much more intelligent, what you can actually cache and how can get a lot more granular. So almost like the edge computing kind of paradigm, right? Instead, you still don’t want to necessarily cache, you know, somebody’s going to a login box or something like that, but we can actually dynamically bypass that now as it’s coming through, and then determine which one should be cached and then which one should go around it.

Another approach as well is, maybe you have users that are coming to your website that are anonymous, right, maybe we want to cache that information first, but then somebody issues a cookie or something like that, all right, then we’re going to bypass it instead or something like that. What we can do is actually start using cache rules that are part of, even inside of like Cloudflare itself. So if we cache something that’s an image, maybe your hero image or something, and it’s not in one of our PoPs, we actually have cache tiering, so if it’s not there, it may actually be in a regional cache location as well. So we can get a lot more intelligent about how we actually cache content and what it is, but cache as much as humanly possible if you can, there’s really no reason not to. It’s going to make everything faster, it’s going to make your vitals like you spoke about earlier have higher scores, which leads to better results and more people visiting your site.

Moeez: Awesome. My next question is from you, Than. How do CDNs help protect against DDoS attacks, bot traffic, or sudden traffic spikes?

Than Oo: Yeah, definitely, that’s a great question. I would say in this day and age, for sure you have to consider both security and performance in tandem, right? I truly feel like if you’re not considering security, your performance eventually is going to hurt, and, you know, your servers get overloaded, etc, things like that. So when you look for a provider, think of a provider who can do both typically, and who can do things in line, where that reverse proxy or CDN provider can also provide the security functions, right? Because otherwise, if everyone is requesting your assets from a CDN, one, you’re going to see a super high bill if you’re really not, you know, filtering out DoS attacks, bot attacks, and things like that. And two, you want to make sure, like at the end of the day, even with using a CDN, that your content and your assets are going to a valuable resource. Hopefully it’s a human eyeball, or if you have, you know, a handshake that you’re doing with agents, totally okay as well, as long as assets are getting delivered to the right people, right? So again, really important to pair CDN, which is a very heavy performance functionality, with security, you really, I would say don’t consider that trade-off.

Moeez: Awesome. Trever, we talked a lot about, you know, Core Web Vitals today in different sessions, and even yesterday as well, and I want to make sure that we touch upon that in this session as well. How does CDN deployment affect Core Web Vitals and perceived site speed for end users?

Trever Jackson: Yeah, absolutely. Core Web Vitals, when used in conjunction with the CDN, the CDN almost becomes really like a cheat sheet, right? As we’ve heard from a couple of other people before already, having that content be accessible and fast to the user when tools like Google Analytics are going to go and rate your site, things like that. You’re going to want to try and get at least, you know, I think the base we talked about was about two and a half seconds, right, to get your hero image and things like that, otherwise your LCP score’s going to way down. Having your images, your hero image, most of your site actually cached in a CDN, right, it’s going to be closer, Google is going to be able to actually look at those things and give you a better rating. Imagine if they were doing it and your website was, you know, 4,000 miles away, you have a CDN, what’s your performance going to be, right? That paint is going to be awfully slow.

From a user perception as well, websites these days are loaded with integrations, there’s scripts, there’s, you know, CRM tools that are integrating all kinds of things. You really want that to be as fast as humanly possible. So offloading as much of that from your actual site or browser if you can, into the edge where it’s already distributed, is going to give you a massive improvement in metrics and both the perceived performance from a user as well. So that instantaneous feel, right? Even the CDN itself will try and use something like Early Hints, okay, that will tell a browser, like, I think that this image or the CSS file is probably going to be loaded next, and go ahead and just prefetch that and load it ahead of time as well. So kind of like when a browser is anticipating what you’re going to click on next, we’re going to do the same thing, just to get that content delivered as fast as humanly possible. So it all works in conjunction with each other.

Moeez: Awesome. Than, you want to chime in on this as well, because I understand that, you know, Core Web Vitals is an important topic, and a lot of the audience on this event wants to know more about what you think, what Cloudflare thinks about Core Web Vitals. So if you want to chime in on this question as well.

Than Oo: You’re always monitoring it, like, Trever already touched on a lot of pieces as well, but, you know, when you’re monitoring Core Web Vitals, another thing that could be important is also considering, like, again, speed is everything on that, right? So protocol efficiency as well. Think about switching to HTTP/3, a little bit different from what we’ve been talking about, but, you know, from legacy TCP plus TLS, switching to HTTP/3 will reduce that round trip from three round trips to one, or sometimes zero, right? So another thing that you should consider about Core Web Vitals, because at the end of the day that’s just measuring your base performance.

Moeez: Awesome. I want to touch upon DoS and bot traffic as well. So how do CDNs help protect against DoS attacks, bot traffic, or sudden traffic spikes?

Than Oo: I think we did that one last, which I’ll answer. The question was about, I think, Core Web Vitals, if I’m not wrong, and sequencing my questions, we should be on number seven now, Moeez, I believe, because before we went to Core Web Vitals, I was the one that talked about DDoS and combining CDN with a security provider as well.

Moeez: All right. So for SMBs versus enterprise clients, is the CDN always necessary, or are there alternative approaches? Trever, my question is from you.

Trever Jackson: Yeah, of course. If you’re an SMB, I would say these days a CDN is almost like your insurance policy, it’s not a luxury item anymore, right? A small business is not going to have like a dedicated DevOps team to go and handle a sudden traffic spike or even a small DDoS attack, right? The CDN almost acts as, as Than was alluding to earlier, that set it and forget it kind of protection, right? If something does happen, maybe you invent the next greatest thing and all of a sudden you get a huge, you know, wave of people coming in, are you prepared to handle that, right? It seems kind of simplistic in nature, but these things happen all the time, we’ve seen websites get overwhelmed and come turn on CDN support, be able to grow their website for an SMB.

Anyway, on an enterprise, right, it’s more about like efficiency, compliance, things like that. CDNs aren’t just caching devices anymore, right, they have load balancing, they’ve got firewalling, they’ve got localized like residency, where is this data actually needing to be, things like that. Maybe I have to host things inside of the EU. Otherwise, what’s the alternative, right? These days, you’re going to go buy a data center full of, you know, hardware, have a bunch of boxes for independent tasks and things like that, we just kind of don’t live in that world anymore, it’s not very practical these days. You just need features, you need to be able to build and go fast. In the age of AI, you need to be able to go up as fast as possible, because somebody else is probably iterating a little bit quicker than you as well. Any benefit you can get.

Moeez: Awesome. I want to talk about latency reduction now. Than, how much latency reduction can businesses realistically expect from deploying a global CDN?

Than Oo: Yeah. Typically the reduction is from hundreds to tens of milliseconds, especially once things are optimized, fully cached at the edge, right? If it’s unoptimized, I mean, it could even be longer than that, right, and at the end of the day, we’re fighting for attention, if your website does not load, you know, in a very performant manner, most likely folks are going to go elsewhere, so very very mission-critical. Yeah, typically the key is from hundreds, from 300s plus, to 50 milliseconds or so. And on Cloudflare in particular, it still depends as well, there are customers who focus only on performance, and there is ability to get that down even lower if you wanted to deactivate some of our security functions, don’t recommend it though, however. But again, if you’re very very latency sensitive, you can, I guess back to protocol efficiency, definitely that will help cut those milliseconds down as well, because we’re reducing round-trip time.

Another thing that we also offer, and also other folks should just think about, is different sort of regional cache tiering. So, you know, if your customers and assets that you believe from a particular region are going to request more, store your static assets closer to them ahead of time. And there is cache expiry and things like that, right, but, you know, push cache on that particular PoP that you expect things on. So regional sort of layering, your tiered cache approach, will also be helpful in avoiding kind of round trips back to your origin.

Moeez: Awesome. I want to talk about agencies now, because a lot of the people in our audience are agencies, and they want to know about CDNs and if the investment is actually worth it. So how should agencies advise clients who are weighing CDN investment against other performance optimizations? And this question is from you both, by the way, so I would like Than to start first and then Trever to follow up.

Than Oo: Sure. With other performance optimizations. I mean, egress is another thing. I think egress is one of the top-of-mind things you should for sure think about between CDN and other optimization, right? Like, of course in CDN delivery, it’s a cost you’re taking on, but you’re going to be saving in the long run egress cost that’s hitting your EC2s, your storage buckets, things like that. So that’s definitely something that you should be thinking about. And actually I want Trever to dive deeper on this afterwards, because he’s actually one of our developer and AI experts, but edge computing is a huge thing you should consider as well. And I don’t want to think about edge computing as a diversion from your CDN investment, it should actually be complementary, and it should actually be part of your performance and CDN offerings, right? With edge computing you could really deliver regional focus, A/B testing for your customers, things like that, and make decisions on the fly. A lot of customers here that we see deliver ads real time and make decisions based on which region, which device, what kind of habits in a particular region that they’re taking on. So combining that next level edge computing and delivery, with traditional performance factors like CDN, is really like the combination in my opinion you should be looking for.

Moeez: Awesome. Trever, you want to chime in?

Trever Jackson: Yeah, I’ll just add on top of that a little bit, because he’s right on. If you are not considering what the next couple of years look like for your applications, you’re doing yourself a disservice. We’re in a time right now where we’ve never experienced this type of growth. Like, what does the application delivery mechanism look like for the next couple of years, right? Gone are the days where it’s okay to just have your application served out of, you know, one region. A lot of our customers only operate out of one zone, one region, and that’s kind of what they do. But think about that from your competitive standpoint now, right? If you’re going to try and run inference, because AI is being integrated into literally everything at this point, are you just going to run a model that’s across the country, across the world, wherever you can get capacity, right? The cloud providers only have a finite amount of GPUs. Have you ever had an experience trying to, you know, access a resource and you just get a 429 all over and over and over again? It happens.

Instead, think about edge computing in the concept of, what if we did inference at the edge itself, right, where your applications are as well. Your data is already there, if we’re going to ask a question of the data, doesn’t it make sense to do it there too? Lower latency, speed is definitely going to be faster because you’re right there, and the availability of it because it’s all over the place and closer to your user, they’re going to get a better performance and a better experience overall. Anyway, so if you’re not thinking about it now, highly look into it, see what that actually looks like, activate your AI at the edge and see what you can get for performance. But happy to work with people on that as well.

Moeez: Trever, on that one last note, would you mind explaining to the folks about what Markdown is, and in the new world of agentic AI, like how would folks even request this versus traditional HTML?

Trever Jackson: Yeah, yeah. For those of you who have ever tried to train a model, all right, the models themselves see the vast majority of the data that exists on the internet, but the truth of it is the data on the internet is messy and it has a lot of noise in it. So instead, if you can actually feed the model clean data or structured data, then you’re going to get better results, right? There’s less tokens you’re going to burn, because the world is definitely moving towards a token-first cost optimization strategy. So we actually have a new service, you can check it out on our blog as well, where you can actually request a website to give you a version of the site presented in only Markdown as well. We’ve got a couple of versions of this actually, something hit the blog today as well, where we can actually have a remote browser instance visit the website on your behalf and then give you just clean HTML afterwards. So definitely check those things out if you are building anything with agentic AI and you need to feed a model, or you’re doing any type of RAG environment, anything like that at all, or you’re trying to build out data, or do, maybe you need a crawler, things like that. But 100% for agentic, when you’re trying to have agents do things on your behalf, which seems to be the pattern that is emerging for this year, highly highly recommended to check that out.

Moeez: Awesome. I would like to divert our attention to some of the questions in the Q&A section from our audience. And I want to take a question that I just saw, seemed very interesting, and it’s related to e-commerce, and it says, how should a WooCommerce store automate cache purging when updating products or promotions?

Than Oo: Yeah, it really depends, I would say, on your preference. A lot of folks prefer a cache key. Cache tags is another piece, right, like, for products, promotions, you could, when you introduce an intermediary like a CDN in the middle, as long as your origin or wherever you have readable cache tags, cache keys like that, that CDN should typically respect that, right? And based off of that, you don’t have to purge by URL, you don’t have to purge by absolute whole domains, you could just easily do it by those tags and keys, which is a lot more efficient. But again, it depends at the end of the day how your store and website is structured as well. If you are already structuring that on a path where it’s easy to purge that way, sure, totally works, but most of the time, from what we’ve seen, folks will typically have those cache tags and keys that you can use on a CDN to just quickly purge.

Moeez: Awesome. Trever, you want to add your thoughts into this?

Trever Jackson: Just one thought, again, it goes back to the basics. There’s so many conversations we have with customers where they don’t explicitly like create their tags, or they don’t have a strategy, you know, ahead of that when they start their site. So just consider, like, what is the CDN strategy as you build a new app, as part of the design or the PRD as you kind of go forward, rather than like a reactive state. It’ll save you so much time.

Moeez: Awesome. Let’s take a few more questions from the chat here. Is there a point where adding a CDN alone isn’t enough, and deeper architectural changes like multi-region hosting or edge computing become necessary?

Trever Jackson: Yeah, I can take that. You’re hitting the nail on the head there, right? I would say the point is already here, right? We’ve had the cloud era already, right, we tried to do the centralized model for a long time. What’s the performance like, are you still having issues with speed, are users still saying it’s not, you know, fast enough, what’s the mobile environment look like? Plus the new environment, right, is your site, or is your more aptly called your data, accessible to humans but also agents at the same time, right? Having it in one strategic location, probably not the best, you know, idea at this point. If you can have something more spread out, more distributed, say something happens, right, maybe there’s an issue in a zone, a region, or a geographic part of the country, power outages, I mean all kinds of things can happen. If you can have that data available everywhere and do your caching, your compute, your hosting, serving, everything from the absolute closest place to a user, their experience is just going to be better, which means your business is going to grow faster, you’ll get more conversions, things like that. So it’s always a win-win. So I would say look at it, start looking at it now if you aren’t already.

Moeez: All right, Than, you want to add some thoughts?

Than Oo: Yeah. Aside from the sort of edge computing use cases that I mentioned, with, you know, local delivery, A/B testing, etc, things like that, I think, yeah, definitely those are all the things that you should be thinking about, and security-wise as well, right? A lot of regions have regional compliance requirements, you know, so in addition to CDN, make sure that the proxy or the provider you go with can meet those local requirements without causing performance issues, right? GDPR is huge, for example, in order to do business in Europe there’s each regional focus and compliance requirements as well. So understand, you know, how your data is proxied, how your data is encrypted, decrypted, things like that, for the compliance reasons as well.

Moeez: Awesome. I think these are all the questions that we can take in the time frame, so this is it for this session, folks. I would like to thank Trever and Than for being here and being part of this panel and enlightening us with their CDN knowledge, so thank you so much. Thank you Trever for being here. As for you folks, this was the last session of the performance boot camp, but we are not done yet, because right after this we will be announcing the winners of the activities that we did today and yesterday, and also the winners of the leaderboard, in just a couple of minutes.