Posts Tagged GAE
Never mind IaaS or PaaS the first question I keep coming up against is should I take advantage of propriety features or design my app so I can lift & shift to an alternative platform. The Public cloud services arena is still relatively new and the reluctance to take the leap is understandable. Simon’s post Google App Engine Hello/Goodbye briefly lists a number of reasons outlining why the leap can be dangerous. For apps that have been designed well there really doesn’t need to be the fear of embracing propriety features after all using propriety features is done every day in more traditional deployments such as using BizTalk or even the propriety features of a traditional Relational solution( e.g SQL server reporting services) . So the lift & shift question is becoming harder to defend as a first port of call really. The initial emphasis should be on the SLA’s provided, the stability of the provider (Amazon, Google and Microsoft all pretty sound) and the projected costs over the life cycle of the project. It may be evident that hosting on the cloud initially will be cost effective but during the lifetime it makes sense to move it in house so in that case the portability aspect does become an issue but it should not be the first question you ask!
- The reliability of GAE does need to be fixed, regardless of designing for the cloud.
- People do not know what they are getting themselves into when ditching traditional (read SQL) data stores and don’t know how to engineer for eventually consistent, high latency data storage.
- The engineering cost of developing for the cloud has to be considered – it may be more expensive than traditional hosting.
- Startups may wish for massive growth when their idea goes viral, but wishful thinking has a cost. Building a massively scalable application (regardless of the platform) may be unnecessary if realistic projections are used.
I like GAE and it’s vision and I like that the platform forces developers to think through their engineering. But for some the pain of learning may be too much and should probably be avoided if there is a serious business to run that needs to make money and keep customers happy.