Virtualization Technology News and Information
VMblog Expert Interview: Catchpoint - To Unlock Tomorrow's Dynamic Digital Applications, We Need CDNs More than Ever


Content delivery networks (CDNs) are one of those technologies that fly under the radar with consumers but play a huge role in shaping the online experience. Now, as more applications and intelligence move out to the network edge, CDNs will become even more important. But are today's CDNs up to the challenge? How will they hold up under the more dynamic content on the horizon? And what should enterprises know to prepare for what's coming? VMblog spoke with Mehdi Daoudi, CEO and co-founder of Catchpoint, to find out.

VMblog:  CDNs have been around for decades now, and they seem to have done a good job keeping up with evolving digital content.  What's happening now to change that equation?

Daoudi: Several things. First, web pages have just gotten so bloated in the sheer number of requests and images they contain. That trajectory is not new, but a typical web page today can have hundreds of requests, thousands of images. At the same time, we're beginning to ask CDNs to serve different kinds of content.

A typical CDN still mostly moves relatively dumb files-static images, maybe some Javascript. Now though, we're starting to see full applications move out closer to end-users to meet stricter latency requirements. Think about connected cars, telemedicine, remote learning. This new generation of applications requires very low latency, as much as 50 times lower than what most applications need today. So, it's a huge gap. If I'm in L.A. having a telesurgery procedure, I can't be relying on compute that's hundreds of miles away in some data center in Sacramento, it's not going to work.

That's why people are talking so much about the edge. CDNs are positioned to play a critical role here. They have the presence and investments at the edge to lead this next evolution of the Internet. But do they have the right architectures to make the transition? It's an open question.

VMblog:  Won't the new HTTP/3 protocol solve many of those performance problems?

Daoudi: Yes and no. In theory, HTTP/3 should make a big difference. It uses QUIC, a UDP-based protocol that's much more lightweight and efficient than TCP. So, it should allow for a much quicker, lower-latency Internet. But, that assumes that HTTP/3 will be adopted consistently everywhere, and that is not a given. There is still no single, official version of the protocol that every browser and CDN is adopting.

On top of that, like anything else in a web page, you are only as strong as your weakest link. You could have every other element of your page doing HTTP/3, but if there's one domain in there that hasn't adopted it, the whole thing breaks.

I would love to see the CDNs take a much more prominent role in helping the industry get HTTP/3 rolled out properly, so we can hopefully avoid the mistakes of HTTP2. But at the moment, it's a bit of a mess.

VMblog:  You've mentioned some of the technical challenges that CDNs need to solve.  Are there outstanding business challenges as well?

Daoudi: Yes, definitely. The biggest being that, for a business with a global online presence, using one CDN just isn't good enough. Because there is no one CDN that delivers excellent performance everywhere in the world. It doesn't exist. If you have users in China, for example, no U.S.-based CDN can give you good performance there. You need to deal with a Chinese CDN. At Catchpoint, we just added our third CDN provider for that reason. But the concept of a multi-CDN approach is still not fully democratized yet. It's still very complex and very expensive for the average company.

Look at it this way. No matter what kind of car you drive, you can pull up to any gas station, hook up the pump, and fill up your car. It's universal. You don't have to drive around with a bag of adapters and dongles like with your phone charger, right? In the CDN space right now, that kind of uniformity doesn't exist. From Akamai to Cloudflare to StackPath to whoever, there is no common configuration. They all speak different languages. So, you need a very sophisticated traffic engineering team that can maintain some kind of dictionary to say, "This configuration in Akamai means this in Cloudflare." It's not simple.

We're seeing similar problems get solved in the cloud with Kubernetes. If you have a container, you should be able to run it on AWS or Google or wherever, it shouldn't matter. We need that kind of concept for CDNs as well. Today though, only a subset of companies have the experience and the money to do multi-CDN. It's not for everybody.

VMblog:  How urgently does the industry need to solve these problems?  Or, put another way, how important will CDNs be to the future of the Internet?

Daoudi: One of the first lessons I learned when I was with Doubleclick back in the late 90s was that the closer you are to end-users, the better. By using early CDNs to serve ad images closer to users, we were able to get more clickthroughs, more impressions, better viewability. Getting closer to users translated directly to better business outcomes. That's just as true today.

But I think the pandemic has been a huge wakeup call for the industry. If there was any doubt before, it's crystal clear now: we're digital forever. There is no going back. So, if we know that business outcomes depend on performance, then we have to solve for that over the last mile. There's no other choice. We don't all live in a 30-thousand-story apartment building; we're spread out everywhere and anywhere. If we're going to take the next step with this new generation of applications, we have to be able to deliver great performance everywhere. These are not just IT metrics, this is revenues, brand, operational excellence. This stuff matters, and we have to get it right.


Published Thursday, May 13, 2021 7:33 AM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<May 2021>