Earlier this week, I wrote an editorial on database scaling in Azure, using federations. Two readers left comments to point you to the fact that federations have been deprecated. I have not used federations for a couple of years, and they weren’t really that old to begin with. What’s more frustrating, I went to Google and did a search on SQL Azure Federations when I wrote the Editorial. It pointed me to an MSDN article on federations last updated in January 2015. In the MSDN documentation https://msdn.microsoft.com/en-us/library/azure/hh597452.aspx they have a small callout near the beginning, warning that federations are being deprecated. I missed that. If you go to the documentation, it points you to other links telling what your new options are.
So, lots of people are frustrated, because they put a lot of effort into setting up their federations, and getting their application to work with them. And, after, what I consider to be a short lifetime, Microsoft has pulled the plug.
In my opinion, if Microsoft wants people to use their software, the lifecycle has to be longer than 5 years. A major component of an application, such as sharding, cannot be replaced overnight. It is not a trivial process, even with the conversion tools they provide.
I like their new approach much more. But what about my project for which I expect to get a pretty good lifecycle, that was a significant investment? How many of you still have applications running on SQL Server 2005, 2008, 2008R2, or later? How would you respond if SQL Server 2005 was disabled, not deprecated, disabled, when SQL Server 2008R2 was released?
I think this is irresponsible behavior from Microsoft. This won’t endear people into moving to Azure. When they host the software, and control the duty cycle for any feature, they can specify what will exist and for how long. That is their right. Now we are warned. Beware of SAAS, if you have software that takes a long time to develop. The platform could reasonably change during the time it takes you to complete a large software application, and you are going to have to re-tool parts of it when the infrastructure is modified. Sure, you can pay a lot of money to have them keep it turned on for you. But, doesn’t that defeat the purpose?
Being deprecated doesn’t mean they are turning it off tomorrow. It will be going away when they discontinue the Web and Business Edition features in Azure.
Does this change how we look at Cloud services? They should be fine for existing 3rd party products, because it is in their best interest to keep those going. Tools like Oracle One World would be kept in synch with Azure capabilities, because there is little more than hosting and deploying. Dynamics will be kept working, because that is a Microsoft revenue stream. But, when building your own apps, are there limits to what we should consider hosting in Azure? Share your thoughts in comments, or drop an email to btaylor@sswug.org.
Cheers,
Ben