@mhassanraza:In your example, you're using a single very hot partition per hour (so, no concurrent distribution). Better, take the last 4 chars of a GUID (which could be your message id) and then you have an even distribution across 64K of possible partition key values.
I would like to know how the functions are loaded and what MS is doing to minimize cold start time. Considering that functions are no more than assemblies (most in the range of MBs if not KBs), all sharing a few framework and libraries (.NET, Json and libs for interop with the bindings), I think that they should be keep in a sort of in-memory storage (or at least SSDs), in order to avoid those 10 secs pause of the cold start.
Very interesting talk about the F# coding style and Microservices practices.
However the C# examples could been written with less lines (for instance suppressing curly brackets for single statements in "ifs" and "usings") and the "discriminated union" not necessarily "in 4 files" as suggested.
Plus, having Linq, C# can be very expressive and concise as that F# query.
I've been using VS 2010 + .NET 4.0 + C# 4.0 since they first Beta (almost a year before), and I'm very pleased that finally they're now officially released.
But what I like the most is that all these stuff is not anymore the "future" but the "present" technologies. So now we can advance to talk towards a new set of "future" technologies, tools and improvements.
In particular, I'm very eager to see Anders Hejlsbegs talking about the new C# 5.0 (or even a new language if the changes are too dramatic to be ensembled into our loved C#), considering its PDCs and TechTalks comments about Compiler Services, Metaprograming,