Werner Vogels, CTO of Amazon, explains how web applications will be built in the future. His point is twofold. The bad news is that expectations for good web applications are sky high. It has to have rich media, available on multiple devices, very scalable, social networking and that is just the beginning. The good news is that a lot of this can be done by services that are readily available on the web, with reasonable usage based pricing.
For example, at The Next Web's conference a service called ECWID was showcased that gives your site a shopping cart, with all payment features. It integrates in minutes. Countless other services exist.
As for scalable hosting, there is hardly a need for dedicated hosting these days. Dedicated hosting does not scale. You can use cloud based virtual hosting, such as Amazon's EC2, or any Virtual Private Hosting, but that is just a start. You should move towards content distribution networks and scalable programming platforms. Examples of these include Azure for Microsoft based software, Google App Engine, or Heroku (for Ruby on Rails applications).
So for your next (or even current) project, consider what you do not have to do yourselves.
By Peter HJ van Eijk, Cloud Computing Coach, Author and Speaker. Visit the blog maintained by Peter HJ van Eijk here.
Related topics: Cloud Computing, Web
To post comments, please login or create an account.
SecuritySponsored byVerisign | |
MobileSponsored bydotMobi | |
DNS SecuritySponsored byAfilias | |
IPv6Sponsored byNominum | |
DNSSponsored byNeustar UltraDNS | |
Top-Level DomainsSponsored byMinds + Machines |
I'm with you that PaaS platforms are where things are headed (e.g. Azure, App Engine, VMforce), but it's far too early for companies to be using those for any serious business. Both platforms are still very limited, extremely immature, and unpredictable. I've had good experiences with App Engine for my own development, but I keep running into walls that limit what I could do with the platform. In a couple years, I think they'll be ready and then the true value of the cloud will become clear.
Some of the limitations I’ve run into using App Engine:
1. Adding/deleting mass amounts of data is a huge pain in the butt, very laborious and manual
2. No process can run for over 30 seconds, so you have to use all kinds of tricks to do backend processing reliably
3. It isn’t using some of the newer versions of Django, which makes I hard to do some things that should be easy
4. Twitter’s API highly rate limits hits from App Engine, I assume because other apps are hammering their API from IP’s you share
5. Custom domain names are somewhat hacky to use, especially when using domain.com (versus www.domain.com).