I've got an asp.net web page that is making 7 async requests to a WCF service on another server. Both boxes are clean without anything else installed.
I've also increased maxconnections in web.config to 20.
I run a single call through the system and the page returns in 800ms. The long and short of it is I think that the threadpool is being being overwhelmed as, once placed underload I cannot get more that 8 requests per second, even though both quad core boxes are running at 20% CPU load and the sql server it's connected to is returning the querys in under 10ms per call.
I've changed the service behaviour to concurrency.multiple but that's not seeming to help.
Any ideas anyone.
There are many different factors that could be in play here. Taking a stab at the remark that changing your instancing model on the service had zero effect (big IF here) then its possible the 'bottleneck' is upstream from the service. Either at the web server, or the client load generator.
You've got several areas to review for tuning: client, web server, wcf service server - that's assuming there are no network devices in the middle. Pick an end and work towards the other end. Since I'm already making an assumption that its not the service, then I'd start at the client and work my way towards the wcf service.
Client
What machine is driving the load against the web server? A laptop? A desktop? A dedicated test agent, or a shared one? The client acting as the load generator for purposes of this test is also susceptible to maxConnections limitation as this is a client setting.
What is the CPU utilization of the client generating load? Could it be that the test driver is just unable to generate enough load to push these boxes? Can you add additional test clients to your test?
Web Server
What does the system.net/processModel element look like in machine.config on the ASP.NET web server? Try setting autoConfig = true. This will allow the configuration to auto size based on the 'size' of the machine its running on.
WCF Service
Review WCF service for any throttling defaults that might be in play and tweak appropriately. See ServiceThrottlingBehavior on MSDN.
Let us know any changes in behavior you might observe (if any) if you make any changes!
The real answer here that everyone missed is that you're using an ASP.NET web page. That means your client is some form of web browser. Modern web browsers have a limit of 2 concurrent async requests at any time. This means that 5 of your requests were queued up and waiting for the first two to finish. Once those first two, it served the next two, then the next two, then the last one.
All of these round-trips and handshakes simply take time. I'm guessing that your roundtrip time is around 200ms, unfortunately you have to do it 4 times.
I also really dislike the "max 2" browser limitation on making webservice calls.
Is this service hosted in IIS, WAS or a Windows Service?
You should try to set Windows to run services on a higher priority. Your WCF Service is probably creating the threads it needs but they should be running at a low priority.
Hope that helps.
Related
We have a core windows service hosting around 9 WCF service and acting as a client to another 3 WCF services. We have a front-end website that communicates with this windows service through WCF.
At somepoint, the windows services is executing some heavy operations which results in 100% CPU utilization, usually split 60-40 between the windows service and SQL server.
This is where the WCF connection/requests between the website times out, and this results in a very non responsive UI.
I am looking for a way to make sure any UI-related WCF calls gets executed anyway and takes the highest priority.
Our main problem is that we need to stick with this deployment scenario, where the windows service, the website and SQL server are all running on one machine. We are required to maintain a responsive UI even with a 100% CPU utilization. I am not sure where to start looking for a fix for that ...
It sounds like you should split your service endpoint onto two separate hosts, one for high volume, or process-intensive operations and one for low latency operations. The high-volume endpoint would process from a queue offline, and the low-latency endpoint would handle requests synchronously from the UI.
The kind of problems you are having are typical of when you try to balance the conflicting resource needs of high volume and low latency together in the same process.
If you cannot scale out in this way then I can't really suggest much you can do about it and must apologize for not answering your question directly.
Another thing you could look at is moving everything asynchronous and using a pattern such as CQRS to provide separation between your read and write requirements.
I'm having no luck trying to find out how channging the instance count for an ASP.Net web role affects requests currently being processed.
Heres the scenario:
An ASP.Net site is deployed with 6 instances
Via the console I reduce the instancecount to 4
Is azure smart enough to not remove instances from the pool if it is currently progressing requests or does it just kill them mid request?
I've been through the azure doco, goolge and a number of emails to MS tech support none of which were able to answer this seemingly simple question. I know about the events that get triggered by a shutdown etc but that doesnt really help in web site scenario with a live person waiting for a request to their response.
You cannot choose which instances to kill off. Primarily this is due to Windows Azure's instance allocation scheme, where your instances are split into different fault domains (meaning different areas of the data center - different rack, etc.). If you were to choose the instances to kill, this could leave you in a state where your remaining instances are in the same fault domain, which would void the SLA.
Having said that: You get an event when your role instance is shutting down (the OnStop() event). If you capture this event, you can do instance cleanup in preparation for VM shutdown. I can't recall if you're taken out of the load balancer at this point, but you could always force yourself out with a simple PowerShell command (Set-RoleInstanceStatus -Busy). This way your asp.net instance stops taking requests, and you can more easily shut down in a graceful manner.
EDIT: Sorry - didn't quite address all of your question. Since you get to capture OnStop(), you might have to implement a mechanism to make sure nothing's being processed in that instance. Since you're out of the load balancer, and assuming your requests are processed fairly quickly (2-5 seconds), you shouldn't have to wait long to clear out remaining requests. There's probably a performance counter to check, to see how many active requests are being handled.
Just to add to David's answer: the OnStop event happens when you are off the load balancer. For web apps, it is usually sufficient time to bleed out all requests after you are disconnected from the LB until the instance is shutdown. However, for long running or stateful connections (perhaps to a worker role), there would be an abrupt disconnect in some cases. While the OnStop method removes you from the LB, it does not terminate open connections. It simply prevents you from getting new connections. For web apps, this is usually enough time to complete the request (and you can delay the shutdown if necessary in the OnStop as well if you really want to).
I have a WCF service hosted in IIS which takes a long time (around 5 hours) to execute. the WCF service basically generates some reports using SSRS (SQL server reporting services) and saves them to a locaton on the server. this service was actually stopping after generating few reports, so I disabled the "recycling of worker process", "shut down worker processes after being idle" and "limit kernel request queue" in application pool and that fixed the issue and all the reports were generated regardless of the amount taken to generate them. but I am not sure if this is the right fix for this and I would like to know what is the impact of unchecking these settings in application pool for the WCF service in IIS? and is there any better way to get around this problem?
For any long running process it is much better to do it outside of IIS.
In this case I would have a regular windows service running that monitored a request queue. When a request comes in to generate a report, it would then spin off a thread to perform the generation.
The web service would be responsible for 3 things. First, adding an item to the queue to be handled. Second, checking status on the queue as to whether the report is ready. Third, sending the completed report back to the calling client.
This would allow the client to essentially do a fire and forget on the report request and call back later to check on it's status. Further it would mean that if IIS recycled for whatever reason you are still OK.
For bonus points I would add some error handling code that when the windows service restarted it could restart report jobs that were in the middle of execution. This would make it a bit more robust and allow you to reboot the server at any point.
I have disable also all the automatic shut down process from iis for my application with out any issue. I have monitor the memory limits and of course the program work smoothly with out any issues on memory.
I think that this automatic shut down triggers are designed mostly for process that keep too many web sites together and possible some of them have not good programming. But if you are the master of your iis, and you have check your program that have not memory issue, then is better to not shut it down, or at least control the shut down process with some way.
Ok is better to make long running process outside IIS, but is not so simple to develop it, not so simple to install it, not so simple to check it out.
I have a WCF service installed on IIS7. I noticed that the first call to my service is always very very slow. The subsequent calls are much faster & acceptable.
If there are no calls made to the service for some time, it again goes to sleep mode. After this the next call again takes a long long time.
Any remedies for this problem?
It is because of process management on IIS. When there are no calls for certain period of time IIS release the recourses and stops the process.
This is why you can notice that it is slow for first request and for requests after a long delay. Because while the first request or after long period of silence IIS loads everything from scratch. JIT complier runs and etc...
Also note :
When you are hosting WCF services on IIS, the WCF services enjoy all the features of ASP.NET applications. You have to be aware of these features because they can cause unexpected behavior in the services world. One of the major features is application recycling, including application domain recycling and process recycling. Through the IIS Management Console, you can configure different rules when you want the recycling to happen. You can set certain thresholds on memory, on time, and on the amount of processed requests. When IIS recycles a worker process, all the application domains within the worker process will be recycled as well
If you need automatic starting: The Windows Service Control Manager allows you to set the startup type to automatic, so that as soon as Windows starts, the service will be started, without an interactive logon on the machine. So you can use Windows service as a host.
More details you can check in Hosting and Consuming WCF Services.
There is another approach through which you can make it better. We have some kind of scehduled process which keeps hitting our server like every 5 mins with very light 'fetch' requests to keep all servers "hot" (by loading most of the required dlls etc) so that user experience is far better.
I agree it is not a fool proof way but still is something you can consider apart from increasing the recycling settings in IIS.
This should be a simple question but I haven't managed to find the answer on google.
I would like to know, in terms an idiot can understand, exactly what application lifetime means in ASP.NET (and therefore when you can expect application start and end events to run).
I assumed it would be when you run and stop the app in IIS, but I've read things that suggest it's related to number of requests.
By default the lifetime starts with the first request to the app. And it ends after an idle timeout.
But this is configurable based on various things (including request count) in IIS.
And IIS7.5 has the ability to start an application when IIS starts, rather than waiting for the first request.
You do have to consider how the Application Pool that your site is running in is configured. Applications can be dumped in a pool with other apps or it can have its own. The pool can be restarted based on memory usage beyond a certain point, by a timer so to speak (reset daily at 3am for example) and I believe by a number of requests beyond a certain configurable number. Not a super expert on IIS so verify before you buy ;-)