Afternoon report from the Windows Azure Platform Launch Day, Stuttgart Germany

This is the afternoon report from the Windows Azure Platform Launch Day, Stuttgart Germany.

We see a demo of creating a development project and deploying it to the cloud.

It does not make sense to repeat all the steps here. There are good tutorial videos on the Windows Azure Platform.

Interesting from my standpoint is, that one has to choose a datacenter for their application to run (why?) and their data to store (why?). I would have preferred just “in the cloud”. However I was told by a Microsoft professional that it might have legal implications where the data ist. ok.

An interesting part of the demo is what I never saw before, the usage of the SQL Data Service. First, an ADO.NET Object ist created, then in the local SQL server the schema is created. Then the implementation with ASP.NET MVC (business as usual) is done. The program can be tested against the local SQL server instance. Later you get the cloud SQL Data Service connection ID via the Azure Portal and you connect to the cloud SQL data service. That sounds to be quite easy. Interesting from my standpoint was, that the Database Size for the SQL Data Service is limited to 10 GB. If you need more, then you need to create more instances or you need to utilize Azure Tables, which have a very different programming models. Azure Tables is not a relational database, but an entitiy data model. There are also no Transactions across entity instances. It is much more scalable. The programming model is different.

Therefore if there is the question, how much effort is it to migrate my existing application to Windows Azure, then the answer is: it depends. If you have a Web application with SQL Server, it should be quite easy if the limitation fits. If you want to use Azure Tables, this – in my opinion – is a different programming model. Even if you can make use of ADO.NET, you have to think about consistency of the data storage in a different way. However in my personal opinion database consistency is overestimated anyway (if we are not speaking about FIN posts for example). There are a lot of applications that do not need a very high database consistency and if they use Azure Tables, they can save a lot of overhead, that is not necessary for the use case at all.

The next presentation is titled RIA apps. However the applications were about connecting a smartphone via the service bus to a desktop application. This is interesting in itself, but with RIA I expect something like silverlight. An interesting emphasis was, that it is possible to request the service bus my mere http – which should be clear in itself – but the usage is, that it can be used to connect any kind of devices to it – and not necessarily only by the .net library. This is in general the open architecture of Microsoft Azure, also for data store for example.

A complaint that I have is the missing of a search infrastructure for Azure Tables. There is nothing and it is not known inasmuch something will be offered, althouh – naturally – many customers are requesting it. However it will be built or offered sooner or later – just because people need it.

Closing this remarks I must say, that the market is moving. My personal observation is, that even today much more is already technically possible, than that which is planned to be exploited. There are exciting opportunities ahead and nobody can afford to just ignore it.