Cloudflare belatedly copying Fastly?

OK, I admit that’s a bit of a clickbait subject title, but I can’t help wondering what’s really going on.

First, in case you didn’t know, Fastly built their own network over half a decade ago. In 2015 they decided to start to tell the world about it here:… and maybe click through their slide deck here:…

Fastly’s network is built leveraging Arista’s routerless technology (and, of course, Software Defined Networking). As they said: We’re unique amongst our peers in that from inception, we’ve always viewed networking as an integral part of our product rather than a cost center. As I talked about months ago, Riot Games used Fastly’s blog postings as an inspiration for them to create their own dedicated high-speed gaming network.

So, when Cloudflare’s CEO goes on CNBC and twice states “we literally building our own network.” I have to wonder if they’re playing catch-up to Fastly in terms of their CDN performance, or whether it’s something in which they think they can have higher security.

Anyone understand the reasoning behind what Cloudflare is doing, and whether it’s a big deal or just CEO to Wall St. hype?


No they are not copying or care to copy Fastly.

Cloudflare its an older company with thr moto of Making The Interner better. They see their network as the operating system of the internet. Cloudflare network is bigger, better and offer more features than Fastly, their architecture has been designed to scale from day 1.


They see their network as the operating system of the internet.

Can you explain that? I don’t understand how a separate network can be an “operating system” for the standard internet.

Cloudflare network is bigger, better and offer more features than Fastly

What data do you have to compare Cloudflare’s network against Fastly’s? Is Cloudflares’s complete, or, as the “ing” on the verb implies, are they are still building it out? How complete is it? What features does it have that Fastly’s doesn’t? On what technology is Cloudflare’s network based?

BTW, I own both companies, so I’m just as happy if Cloudflare has advantages over Fastly as I am if Fastly has advantages over Cloudflare.


I’m with Smorgasbord1 on this one.

Why didn’t Fastly just use V8 like everybody else? If you didn’t know V8, is a runtime environment that was built into chrome to separate out tabs, so that if one tab crashes it didn’t affect the other tabs. V8 was built for browsers and wasn’t designed for the edge.

So Fastly skipped those off-the-shelf conventional serverless solutions on the market, like containers or the V8 Engine and chose the harder path to build their own runtime on top of WebAssembly.

Fastly open sourced Lucet, its native WebAssembly compiler and runtime designed to take WebAssembly beyond the browser, and build a platform for faster, safer execution on Fastly’s edge cloud.

This is core to their vision and what they see as building blocks for future Edge apps.

And like Smorg mentioned, Fastly took a similar path very early on to build their own software routing infrastructure from scratch to offer a lot of flexibility in load balancing and routing, instead of relying on expensive routers from Cisco and Juniper etc.

There’s a lot to like about FSLY’s technology. Having written a lot on NET and FSLY on this board, I’m still biased towards FSLY when it comes to the technology. The last time I checked, FSLY’s Compute@Edge was oversubscribed. So hopefully there’s good demand and things will go well and pay back in the future.

However, let’s keep in mind that WebAssembly is relatively new ( became a W3C recommendation on 5/12/2019 and alongside HTML, CSS, and JavaScript, is the fourth language to run natively in browsers. ) And the fact that the first WebAsssembly conference was held in February this year just before the pandemic hit hard. So a lot of this is pretty new stuff and it does look like there’s work to do before the full power of WebAssembly can be used to empower uses cases like…better performance of apps on low-end mobile phones etc.




I had missed to mention this in the previous post…

Content hasn’t changed BUT Technology Has!

When I was working on a site featuring content to end users, our team worked very hard to show what’s relevant to the end user to increase the time they spent on the site (their engagement). This study shows how important personalization of content is for any website/service :

Ask this to Amazon or Shopify and they’ll tell you how important this is.

So, progress in CDN technology is making us re-think about the boundaries of static vs dynamic content. Traditional CDNs are poor ( or almost fail) to cache dynamic content, e.g. like inventory changes, account updates ( I had mentioned some of these use cases a long time back when first evaluating NET vs FSLY on this board). Traditional CDNs have to make more round trips to the origin to fetch that data. This means more cost and latency. For those of you who are new to CDNs, “origin” means the the server/s which contain the original/up-to-date version of your website vs the CDN edge servers which caches a lot of that original content/data.

Additionally, this lag in “state information” between the edges and the origin can also cause visibility problems when you’re trying to build metrics using logging data from edges and origins. They also cause data consistency issues, for example, if a piece of information about a product changes, related items like images may change which usually causes a lot of purges (cleaning up and refreshing of caches). Certain cache-unfriendly technologies like GraphQL can aggravate this problem by mixing up static and dynamic content ( like the items in a shopping cart with their related pictures etc).

Thinking about Fastly, their high performance POPs can alleviate a lot of those issues faced by traditional CDNs by making fewer round trips to the origin ( you should know what origin means by now :)). One way to achieve that is via instant Purging and Selective Purging. Also realtime log streaming is another import feature that helps developers monitor API performance. The apps of the future will be build on the Edge ( as I mentioned in the previous post). It will be a world where developers should not worry about static vs dynamic, where round trips to the origin are not going to be the common case.

Compute@Edge from Fastly is attempting to solve a lot of these problems. Somehow, I feel it won’t be too long when FSLY is going to be in the CDN space what CRWD is in the Security space as devs will be building more and more apps on the Edge ( without worrying about static vs dynamic). That’s just how the dev in me feels :). The reason I feel that way ( and perhaps many in the dev community) is that when I look at FSLY’s tech, I’m always impressed and happen to learn something new and that seldom happens when I look at NET’s tech ( Though I own both at this time being a prudent investor; since Mr.Market doesn’t distinguish between the best and good tech).




P.S. Since I’m new on twitter there’s a lot of stuff I don’t know there except a few folks I’ve been following since a long time. I wasn’t happy with my twitter handle so I changed it to @cloudandstocks ( this one also sucks but most of the ones I liked have been taken for now).