Why should(n’t) you rely on CDN’s for hosting libraries?

CDNs such as Google’s allow developers to link to resources on it at no cost. This article addresses the benefits, and explains any drawbacks of taking advantage of this seemingly generous offer!

So what are the advantages?

  1. The speed of the download should be quicker. CDN’s work by installing endpoints in locations closer to the user, both geographically and topologically. It’s also likely the CDN offered by the likes of Google is also more optimized and beefy than anything a developer would be able to provide themselves. All of these factors should result in lower latency and a faster download speed.
  2. The chances of a cache-hit are increased. For a file to be possibly cached, the user must have accessed it before. If multiple websites link to the same resource (like multiple websites use the resources on Google’s CDN), the chances of the user accessing the file is increased compared to you hosting the file yourself (and so only your website using the resource).
  3. It allows parallel downloads. Browsers are restricted by how many requests they can make concurrently to a domain/ IP address. The limit is browser version specific. By hosting resources on different domains/ IP addresses, you increase the amount of resources you can concurrently download.
  4. It frees up your servers resources. Hitting a CDN rather than your server means it has to deal with one less request. That’s less bandwidth, less connections and less load and less money spent!

Sounds good so far! Are there any disadvantages?

  1. Hosting files on an additional domain requires an additional DNS lookup to resolve it. This requires a request to a DNS server. It’s worth noting that DNS entries are cached, so this DNS lookup won’t hit you every time. Whether this additional delay is worth the speed increase of the quicker download remains to be seen.
  2. CDNs do go down. Two outages spring to mind that had a major affect on websites relying on that CDN. The MathJax CDN went down a few weeks ago, and the Google CDN was unavailable for some users a few months ago. If your website relies on the resources on that CDN, your website goes down with the CDN.

Anything I can do about those disadvantages?

It just so happens you can do something about the second disadvantage. It’s possible (and easy!) to detect whether the CDN you’re using is unavailable, and if it is, revert to your own copy of the resource. There’s no speed penalty using this approach if the CDN is working fine.

To do this, place an additional <script> tag after the <script> tag referencing the CDN. Inside that, check whether the resource was loaded (in the example below, we check that jQuery is available). If it was, the resource was loaded and you don’t need to do anything. If it wasn’t, simply load your own copy of the resource;

<script src="http://ajax.googleapis.com/ajax/libs/jquery/1.7.1/jquery.min.js"></script>
    if (typeof window.jQuery === "undefined") {
        document.write("<script src="/assets/js/jquery-1.7.1.min.js"></scr" + "ipt>");

For further reading, there are a number of Stack Overflow questions which address these points.

Leave a Reply

Your email address will not be published. Required fields are marked *