4

I have a Windows Service which executes large number of tasks in parallel using the Task Parallel Library (TPL). This is about to be extended to handle tasks which interact with an SQL Server on an external server.

TPL is supposed to be good at measuring load and assigning the right number of parallel threads to the tasks. Is there a way to make it aware of load to the external SQL Server instance? The actual code to run for each task on the local server is quite small, but the calls to the database can be quite heavy.

Am I not likely to end up with my service bogging down the database with request because TPL sees that the local server has loads of free resources, or is there a known way to handle this?

Improve this question
3
  • If there is a way to extract the database interaction from the parallel tasks, that would be your best pick. Commented Mar 10, 2011 at 12:14
  • The database interaction is what I'd like to do in parallel, so extracting that into non-parallel execution would make no sense... Commented Mar 10, 2011 at 13:10
  • This doesn't mean you will have to so it completely sequential, but when separating it, you will have more control over it. Commented Mar 10, 2011 at 13:17

3 Answers 3

Reset to default
3

There's is nothing native to TPL that will help you with this. TPL is about managing/maximizing the CPU load of your local application. It has no idea about SQL load, let alone on another machine.

That said, if you wanted to get crazy, there is an extensibility point called the TaskScheduler. You could theoretically implement a custom TaskScheduler that can watch the load on the SQL server and only schedule tasks to execute if that load is at some defined threshold.

Honestly though, I don't think it's the right solution to the problem. Managing load against a shared resource like a SQL server is a completely different beast than what TPL is designed to solve. You'd be much better off just making sure you design your application such that it doesn't abuse the SQL server in its own right by load testing, finding a sweet spot and configuring your application not go out outside those bounds. From there it would be up to your DBA to determine the right solution for the SQL server infrastructure itself to manage that application's needs along with any other external load.

Improve this answer
Sign up to request clarification or add additional context in comments.

Comments

0

If you parallelise the data access functionality in your client application, you will find that the next bottleneck is the SQL Server connection pool.

Improve this answer

1 Comment

I dont think that will be a problem, depending on the hardware configuration. Yet its completely normal ofcourse to run multiple SQL batches at the same time. Also, the question seems to imply that there is already a service doing heavy work in parallel not nessesarily bound to SQL
-1

TPL is good at partitioning your data, as for measuring load, that task is for your OS to determine (and in fact its pretty good at it). Therefore this is more of a configuration question then a development question. (Does your SQL Server instance have a higher priority then your service?).

Improve this answer

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.