BLOG: How Cloudflare and Afterburst saved Pale Moon

[Moonchild]

This is the first of a few blog posts I will be posting on the forum, since the blogging site has been taken down (all it did was give me a lot of extra work dealing with harvesting and spamming bots, even with host blocking in place) and I don't want my previous write-ups to be completely lost.

Two companies have been pivotal in the success and availability of Pale Moon to the public, namely Afterburst (previously called FanaticalDev) and CloudFlare. Also, not wanting to sound like a promotional blurb for this new startup that has taken the 'net as a whole by storm, I did want to write up how CloudFlare has made a difference for me personally and for Pale Moon as a whole.

The problems

Increasing popularity

As Pale Moon became more well-known over the years since its release in late 2009, I could no longer serve all the requests for the start page and home page comfortably, even with the gracious donated bandwidth of my Dutch hosting benefactor Sity.nl - getting up to 100,000 page requests per day is a lot to handle, even if the page is a simple HTML page with a few icons.

Additional bandwidth requirements were imposed as Pale Moon grew in popularity to be able to offer the automatic updates/internal updates. Serving a large number of people with software updates is (as you can imagine) something that requires a large amount of bandwidth.

Location

With (initially) a single server to serve the website and start page, location became an issue as well - especially very frequently requested pages like the default home page should be fast, regardless of location of the user, and hosting either side of the Atlantic would invariably bias for local access, with how regularly the Trans-Atlantic pipe is clogged these days. This asked for a CDN (Content Delivery Network) that would bring the site "closer to home" for people spread out around the globe.

Budget

CDNs are generally expensive. No surprise, since they are used mainly for corporate use and require dedicated datacenters. Pale Moon being a FOSS (Free and Open Source Software) product sees very little on the "in" side of the equal sign, and most CDNs would be out of my budget range, not to mention the fact that I had to think about having several servers available for different parts of making Pale Moon available to the public (website, start page, forum, update server, etc.).

The solutions

First things first: more servers

Before thinking about how to deliver all this data to people around the globe, I had to think about where to serve it from. A single server was obviously not going to cut it, and getting a VPS (Virtual Private Server) had been a nightmare: VPSes are regularly severely bandwidth limited, suffer from "bad neighbors" in the IP range or on the host itself like DDoS attackers or people abusing resources on host machines impacting others having VPSes there, not to mention providers overselling their hardware by trying to squeeze way too many VPSes on a single host.

After a few miserable failures trying to find a good provider, I spent a few days straight reading reviews, comparing specs and prices (finding the "Goldilocks" range) and eventually signed up with Afterburst as my provider of choice. Sporting good price/value, and more importantly not having bandwidth restrictions (unmetered, and no that is not with quotes like so many "unlimited" plans you see advertised). I couldn't have been happier with them from the moment I started using them.

Then: A CDN

This was a tougher nut to crack, but it didn''t take long for me to end up with CloudFlare. Why? Because they have a free plan, which allowed me to test drive their service without too much of an issue. Focusing on web page content delivery, they immediately tackled my main concern: the start page with many repeat hits every second of the day. Also, I saw the potential of using CloudFlare for distribution of the software updates, but that would be beyond the scope of what CloudFlare is set up for. Working with the people at CloudFlare, though, I managed to get a deal that would allow me to pay a relatively small amount each month and reap the benefits of a fully distributed network to deliver the software updates.

Thirdly: Security

As any site owner may be aware: security is an issue for any owner with a project that grows in popularity, ranging from script kiddies who try to disrupt services "for the lolz", to people who are "fanbois" of "rivaling" pieces of software, to people who do things large scale for blackmail/extortion or other criminal reasons.

Here, too, both Afterburst and CloudFlare came to the rescue. Afterburst by providing a stable, fast platform with full server configurability in multiple geographical locations at a budget price, CloudFlare by the very nature of their anycast network, and their advanced security and DDoS countermeasures.

Conclusion

If it weren''t for the two companies saving me a lot of headaches and allowing my limited budget to stretch to a world-class service delivered to my users, I may have given up on Pale Moon. So, in short, they did save my project from going under - or at the very least never getting beyond the "hobby" stage that it started out as.

[Night Wing]

That was a very nice informative post detailing your trials and tribulations concerning the 5 years of Pale Moon's existence and your dedication to take Pale Moon beyond a hobbyist stage.

Thanks for sharing this amongst the members on this site.

[dark_moon]

Agree. Nice to read this and good what you find a solution for all.
I remember the server problems.

Just a question: When you start the pale moon project, did you expect so much interest in your project or just having fun with optimize the browser ?

[Moonchild]

Just a question: When you start the pale moon project, did you expect so much interest in your project or just having fun with optimize the browser ?

When I just started, before it was made public (Firefox 1.5/2.0) it was just tinkering with it, and sharing with friends.
When I finally made it public, in Oct 2009, I did expect interest in my project (at the time it was just an optimized re-build of Firefox, unlike what it has grown into now) but not this much. It's been growing steadily though, so it helped me take my time and get past those hurdles without too many issues for the actual users.