Recently, CDNify, an upstart CDN reseller based in London decided to abandon its CDN partnership with OnApp. OnApp is a software developer offering CDN-in-a-box, cloud storage and cloud compute. The OnApp solution is ideal for hosting companies and some Telco’s who are interested in getting into the CDN business. OnApp glues together excess capacity from hundreds of hosting companies, and offers that as a CDN. However, as CDNify found out, OnApp wasn’t the ideal solution for them. They weren’t happy with the overall quality, so they decided to build out their own CDN platform, and become a true CDN startup, not just a reseller.
Gigaom published an article on the situation, and assertions are being thrown around by OnApp and CDNify. Gigaom, OnApp and CDNify didn’t really get it right. Some of the assertions, comments and expectations are way off, and we’ll give our reasons why. Let start off by saying that the most important element of building a CDN is setting proper expectations. Second, hiring someone from within the CDN industry is a must if a company plans on starting a CDN. With that being said, OnApp is never going to work for a pure-play CDN, both from a business perspective and technical. Next, the CDN-in-a-box solution is never ever going to compete with Akamai or Limelight Networks on performance or feature set.
Some Comments made by Gigaom, CDNify and OnApp
- Gigaom: Federation is a smart way of gluing together a network to rival Akamai and Limelight
- CDNify: With OnApp, some of the POPs ran slow, and others didn’t respond well
- OnApp: Different types of CDN infrastructure have different price points, you can choose infrastructure at different price points for higher quality and lower quality
- OnApp: You may have some online businesses that don’t require 5 nines of uptime
- OnApp: We put star ratings for each POP with higher rated POPs offering better performance
- OnApp: Customers need to choose the right POP mix for their business
- CDNify: It’s tough to differentiate from other virtual CDN operators
- CDNify: The CDN feature bundles are all the same across all CDNs and its hard to differentiate between CDNs
- CDNify: Big selling point is free custom SSL, whereas Amazon charges $600 per month
 CDN-in-a-Box vs Akamai
Akamai and Limelight have invested hundreds of millions of dollars in their CDN infrastructure over the last decade. In addition, both have a decade of experience operating, optimizing, tuning, tweaking, and perfecting their CDN platform. Akamai probably has more engineers than OnApp has employees, who are dedicated to optimizing each layer of the CDN stack, from the OS, to the routing, caching, storage, and so on. There is never going to be a CDN-in-a-box solution that offers the same level of performance and feature set.
OnApp’s Comments
OnApp stated that there are different price points for higher quality CDN infrastructure vs lower quality infrastructure. They also commented that 5 nines of uptime is not always required. Next, OnApp said it’s up to the customer to select the right POP mix for delivery. Finally, some POPs are rated higher than others, with the higher rated POPs offering better performance.
Response: In the US CDN Startup Ecosystem, not including Akamai, all CDNs offer similar pricing for http delivery, streaming, token authentication, and so on. If they didn’t, they would go out of business fast. In addition, the POPs for all US based CDN offer identical quality, if they didn’t they wouldn’t get deployed. So why doesn’t OnApp offer identical pricing for all infrastructure? If they don’t, their customers are going to have an extremely difficult time competing with the pure-play CDNs.
Lastly, every major CDN offers 5 nines of uptime. Five nine’s of uptime shouldn’t even be an option, but a given. If a pure-play CDN offers anything less than 5 nines of uptime, it’s going to be at a severe disadvantage compared to other CDNs. There shouldn’t be different price points for different levels of uptime. Also, keep in mind that by default, a CDN always offers 5 nines of uptime. If one POP fails, another POP picks up the load and continues to deliver the content. Conclusion: OnApp needs to tweak it’s business model to work out the kinks, and offer pricing/quality on par with the pure-play CDNs. Continued in Part 2. Part 1, Akamai and Limelight vs OnApp