For a number of years now I’ve advocated the benefits of CDNs (Content Delivery Networks) and how they can be used for technical SEO – something we coined as edge SEO.
Using CDNs for the benefit of SEO isn’t something new, it’s something a lot have us have been doing for a number of years. At SALT, we wrote this post back in 2015 looking at potential SEO issues of using CDNs – serverless applications such as Cloudflare Workers and Akamai EdgeWorkers bring about new opportunities, and challenges to our internal processes.
I’ve also been fortunate to work with CDN providers.
I’ve also written on the SALT website about how CDNs can be used for SEO, as well as on other blogs, such as these articles on OnCrawl:
- Part 1: Resolving SEO issues on legacy (and modern) platforms through edge SEO
- Part 2: Creating & Managing Redirects on The Edge
I even wrote a piece on Cloudflare’s blog looking at the application of Workers and technical SEO.
Cloudflare, Amazon & Facebook Outages
However, stark reminders of how far technology has still to go are happening in the real world, and that is that no technology is infallible.
Whilst Cloudflare bore the brunt of angst on social media as at the peak of the outage around 10 percent of Cloudflare handled traffic was being directed over Verizon, the outage also affected Amazon and Facebook, after Verizon wrongly accepted a network misconfiguration from a small ISP in Pennsylvania, USA.
Major outage impacted all Cloudflare services globally. We saw a massive spike in CPU that caused primary and secondary systems to fall over.
For around 3-hours, traffic and users intending to go to some of the biggest websites online were accidentally being rerouted through Allegheny Technologies, a specialty metals company.
This is being known as BGP super-blunder.
Shopify & Google Cloud
Other recent examples also come to mind, such as the Shopify outage, which became known as Shopify Outage Sunday.
Shopify depends on the Google Cloud infrastructure, which when it went down, took down a large number of Shopify websites with it.
Due to terms of service, many e-commerce retailers who utilize the Shopify platform were left out of pocket through missed sales and traffic due to their sites being down.
As well as Shopify, the Google Cloud outage also affected YouTube, SnapChat, and GSuite.
This is why CDNs are game-changing and should be welcomed with open arms and if appropriate, utilized, however they shouldn’t become the game on their own.
CDNs and edge technology can be game-changing for businesses, but it alone should not be the game.
The Google Cloud outage highlighted a flaw in Shopify’s model, that there isn’t a failover stack. As technologies advance and new challenges to our ways of working arise, we need to have the responsible conversation around edge technology.
We can also go back to 2013 when the ASP.NET Content Delivery Network went down.
Should A CDN Ever Go Down?
Large CDN providers like Akamai, Amazon CloudFront, Azure CDN, Cloudflare have an infrastructure of thousands of servers and a really good load balancing and high availability strategy.
The short answer is no.
Theoretically, it’s hard for thousands of servers to go down at the same time and lead to an outage — but as June 24th has proven, it can happen.
It is important to note that all CDNs differ in capacity, overhead, time to compensate for failures, and a number of levels of load balancing.
That being said as Cloudflare’s outage has shown, the ecosystem of the internet has a number of reliances – and in the instance of June 24th, this is a down to a large ISP (Verizon) not filtering out a bad request.
What To Do When A CDN Goes Down
Unfortunately, there is no hard and fast solution for when a CDN like Cloudflare or Imperva serving your whole site fails, unless you have a very active development team and are able to move DNS away from the CDN temporarily.
The Verizon/BGP mess up was a rarity, and a group called the Mutually Agreed Norms for Routing Security (MANRS) does exist and has provided recommendations to prevent this from happening again in the future.
The two technical recommendations are filtering and anti-spoofing, which basically check announcements from other network operators to see if they are legitimate and remove any that aren’t – which is something that Verizon (who aren’t members of MANRS failed to do and instead accepted the announcement blindly).
The cultural recommendations are coordination and global validation. This is the encouragement of global, national and mid-scale operators to talk more to one another and work together to flag and remove any suspicious-looking border gateway protocol changes.
However, if your scripts and JS libraries go down – you can revert to serving them locally – as covered in this article here by Scott Hanselman.
Adding Additional Nameservers
Typically when a CDN does fail, it’s not actually the CDN component that fails – but more something else along the line, such as the DNS or BGP.
This is where adding a secondary DNS in case of primary DNS redundancy can be useful, however, if you’re on Cloudflare, this isn’t possible.
Cloudflare needs to be your one and only set of name servers.
Article updated July 2nd, 2019