a more rich user experience to manage Azure services

When deploying a new version of an app to staging or production, Azure should automatically snapshot the currently-running version and save it in blob storage, along with a version number and time stamp. Then, if things go horribly wrong, you could easily roll back to a previous known working snapshot.

This needs to be super-simple: A check-box option to decide whether we want snapshotting, some options for how many snapshots and how far back they should go.

Restoring a snapshot should be as simple as clicking on the version hyperlink.

Make possibile to update a deployment without change the current service configuration (like number of instance, etc..).

It would be great if you could setup custom values for the cscfg config for dev/staging/prod. For instance, when in DEV, i might want to use a different SQL Server connnection string than I would in Staging or in Production.

Right now, it's one click to stop an instance in production. This can accidentally happen with bad consequences.

We can't have the information on dashboard if an instance is small or medium.
And we need to have capability to up this size.

My problem is :
When i uploaded my application, it was a medium instance and when i want to migrate to billing model, it kept this setting (the vmsize wasn't specified in my new configuration settings, i was thinking im on small instance, and now i need to pay for compute medium without know...)

The application pool idle timeout defaults to 20 minutes. That results in a slow experience for users if the site isn't hit as often. There are hacks involving a startup task or running the site elevated to use the management API DLL, but neither are elegant. Ideally you should be able to set the app pool idle timeout from the Azure Portal, as well as other advanced recycling settings (recycle on a schedule, recycle every x minutes, etc).

Hi,

By using REST API, I want to get the list of service namespaces created for the Service Bus.

There may be a reason for this, but I find this very unfavorable. I need to have a peek into Windows Azure's future so that I can plan ahead. This includes knowing what *won't* be coming in the next year or so. Example: Secondary indexes has been announced at PDC 09. If it won't be coming in 2010, just say so. Don't leave us in the dark. Having to guess and wait is a bad setup for talks with customers. Even the roughest of timeframes provides for much better planning.

I want to be able to use my own API key's for Google, Yahoo, etc in Access Control service, so that when the Identity Provider asks the user if it trust my app, it will ask about mydomain.com and not mydomain.accesscontrol.windows.net. It will also allow my app to lookup more information about the user. Please implement something closer to Janrain engage.

Startup Task script via configuration in management portal (in addition to the existing method of creating a deployment package from Visual Studio) – one usecase is just needing to execute a script that simply loads assets from Blob storage to create and launch the local application environment

Please add the capability to add custom metadata to storage queue messages so that we can more easily implement additional communication features on top of them (correlation, header based routing, ...)

Improve development fabric so that it does not consume 100% CPU for a small number of inserts into table or blob storage. It is very difficult to insert half a million rows into development storage -- it can take hours and 100%.

I wish I could stay completely within the IDE (Visual Studio) to publish to a staging instance on Azure, and then do a VIP swap with the production instance. As it is right now, I have to open the Management Portal every time I want to do a VIP swap.