Wednesday, August 24, 2011

Asynchronous Programming In .NET Is No Longer 'The Big Bad Ugly'

Let's face it, the word 'Asynchronous' still conjures up thoughts of low-level multithreading challenges and words like 'mutex' and 'deadlocked' for those of us that have been developing long before the world of .NET and even early on in the day's of .NETs inception. However, no longer should this fear or concern be so prevalent like it was in years past.

Why? Well the good folks in Redmond have added so many layers of abstractions atop of the System.Threading namespace and on Asynchronous programming techniques in general, that the developer no longer needs to know how every gear under the hood works anymore. These abstractions of Asynchronous processing have evolved in many different forms including but not limited to 'Asynchronous Delegates' (Framework 1.1), 'Background Workers'
(Framework 2.0), 'Asynchronous Lambdas' (Framework 3.0 C#), 'PLINQ' (Framework 3.5), 'TPL' (Framework 3.5), and now the 'Async Framework' (CTP). All of these abstractions have a similar theme: allow the developer to quickly and efficiently create processes that execute concurrently with typically less code and a smaller chance for failure caused by improper coding that could occur when manually spawning threads attempting to achieve the same outcome. What's the result? The same one that a race car driver has when he steps into a car he didn't build himself: he doesn't need to know every detail of how the engine or car was built, but just how to drive it and finish 1st! Now it's not to say that basic or even mid-level knowledge of how these asynchronous processing abstractions (as I like to call them) are not required. In fact, if you go beyond just scratching the surface of simple asynchronous programming, the knowledge is important on how to proceed to more advanced topics and methods available. However, this still is far from the requirement of understanding everything about spawning and managing your own system threads.

One of the main draws to harnessing the potential of asynchronous processing now more than ever is the advancement in hardware that has occurred in the last 5 years. You might still have a dinosaur PC at home with a single core CPU, but odds are you have 2,4,6, or even 8 cores on your newest machine (like the 8 I have on mine, thank you Intel I7 -> read here:
My New Computer: A Developers Dream) and have bandwidth to spread processing out among available cores. This basic knowledge of the number of cores or threads available in your environment where the application will be run, can help you decide on which technique to use or if the processing time will actually be reduced. However it is most likely that your environment has at least a dual-core CPU and bandwidth available to run some processing asynchronously. The end result will be you can look like a hero in a few lines of code by running long running or redundant tasks concurrently with a relatively small understanding of all of the true complexities involved in multithreading and asynchronous processing.

So for now, I will leave this discussion here and going forward will have several posts aimed to introduce or familiarize you with using some of these asynchronous programming techniques available.

No comments:

Post a Comment