Friday, April 29, 2011

We Still Love Amazon

Everyone jumped on the recent Amazon outage to let us know how dangerous the cloud is. Service providers that rely on Amazon were understandably upset. Others (see the articles by CIO and Computerworld) jumped on the opportunity to question the reliability of the cloud. Apparently, people are asking themselves whether they are better off running their software on servers in their own data center or in data centers of traditional outsourcers like EDS and IBM.

As if nothing ever goes wrong in those data centers. People make mistakes and technology will fail from time to time. This can be very costly and we should learn from each failure to ensure that it can never happen again, but I fail to see how this says anything negative about cloud computing.

Like the other unsung heroes of the cloud at Rackspace and EngineYard, Amazon is providing an important service. They offer a flexible infrastructure that enables other service providers, like the ITRP Institute, to focus on the functionality of their applications, without continuously being distracted by the infrastructure that the applications are running on.

Yes, the ITRP service also runs on Amazon Web Services and this gives us an important competitive edge. If we had to acquire the physical equipment needed to run ITRP, as well as the know-how to maintain it, we would never have been able to get ITRP off the ground so quickly. In fact, ITRP would probably not even have been built. It would have been too expensive.

If you work for a more typical organization, just ask yourself this:
- How long did it take the last time you needed a new server?
- When you don't need a server any more, how complex is it to get rid of it?
- What did you get the last time you suffered an outage of the infrastructure on which you run your services? At least Amazon is giving its affected customers a service credit.

Cloud services are important. They allow us to scale up and down as we please at a reasonable price, without having to deal with capital investments. The breakthrough benefit that the cloud services offer, however, is that they make disaster recovery affordable even for the not-so-business-critical applications. This ensures that cloud computing can offer reliability at lower costs with much greater flexibility. Corporate data centers and traditional outsourcers are simply not able to offer continuity options at such low prices.

Clearly, this does not mean that things can't go wrong, even when you make use of the high-availability options that cloud service providers offer. But it also does not mean that things won't go wrong when you are running all your services in your own data center on equipment that you own and operate. If you think that gives you more control, just try and start your emergency power generator. When it doesn't - which often happens - how 'in control' does that make you feel? And how many other things can go wrong that will leave you at the mercy of your suppliers?

In the end, whether you rely on cloud service providers or other vendors, the trick is to find ones that you can depend on; true specialists who take pride in the quality of their products and services.

We are confident that Amazon is learning valuable lessons from this event. They even seem committed to improving their customer communications. In the long run this experience will make them an even better service provider. If not, other cloud service providers will be happy to look after Amazon's customers. The cloud's success will not be stopped by a major incident like the one Amazon just endured.