After deploying my Web Api to Azure, I noticed that I have a very high response latency. I stopwatched a method that awaits a Http request to a controller that just returns a "Hello" string. The times I measure are also not consistent, but most of the time I get something around 0.9 seconds. The problem is, that my database queries take forever, even the least fancy ones take something around two seconds (and when my UI updates multiple elements, it takes up to 4 seconds until the whole thing is loaded).
I have really no idea where to start diagnosing this issue, so any help (even the most basic) would be highly appreciated!
I found my problem, I used Database.EnsureCreated() on every service request, since it was called in the constructor of the DbContext I inject on Startup. Seems pretty unefficient. Whoopsie!
In our application we need to make around 10k REST api calls to 10k different endpoints URLs per minute. Earlier I was using synchronous model, but quickly realized that I can not scale beyond ~2k+ limit, so I am working on switching to an async model. Using HttpCore-NIO lib I could scale upto 5k or so, but beyond that I randomly get an error 'I/O reactor has been shut down' and the entire app basically stops processing requests. Don't see any stack trace either, which makes it extremely hard to debug.
So I am trying to evaluate what could be the best strategy/library to achieve this scale with Java as the programming language. Any suggestions on which libraries out there should I look into ?
If your machine itself is not stressed out on CPU, Network etc, then perhaps one thing that can help is horizontal scaling - if you can achieve 5K with 1 process, try firing requests from two processes; see what you get; how far you can horizontally scale. The extra coding to achieve horizontal scale should not be that much.
OR try trapping exceptions and ignoring them, to prevent I/O reactor shutdown. Experiment with what can be ignored. See example here https://hc.apache.org/httpcomponents-core-4.4.x/tutorial/html/nio.html#d5e477, search for "I/O reactor exception handling"
Experiment with thread count - setIoThreadCount
Make sure you are not doing too much processing with the HTTP responses; that would hold up the reactor.
Benefits of reducing threadcount
It is fashionable these days to attempt to improve the scalability of web-servers by reducing the number of threads handling requests, in an effort to prevent unnecessary context switching from having tens or hundreds of threads. This is particularly important when waiting on some kind of application event for a 'long polling' (aka COMET) request - it simply is rude to consume all the resources of an entire thread and do nothing with it.
Node.ns as an extreme example
Node.ns is a rather extreme example of reducing thread-count, requiring handlers to give up the single thread handling all HTTP requests when waiting on any event (even I/O from a local disk!). In this way, a CPU core is kept busy doing useful computational work, instead of 'wasting' instructions pausing and resuming many different threads.
But API support is needed
However, arranging for resumption of the handler once the waited-for event has completed typically requires explicit support in the handler API. For example, there is a part of the Java Servlet Specification which deals specifically with 'asynchronous' responses.
Wherefore, ASP.NET?
So, does ASP.NET have a similar API for asynchronous responses (aka. continuations, aka. (a bit loosely) 'reactor pattern')? Is there authoritative documentation which directly addresses this matter and its interaction with IIS? What nomenclature does ASP.NET / IIS use to refer to these properties?
I am coming to IIS/.NET shortly, and don't want to write any code that's unnecessarily 'thread-per-request' if I can help it. I'll be using C# by preference.
This question is about limits imposed to me by ASP.NET (like script timeout etc').
I have a service running under ASP.NET and I want to create a counterpart service for monitoring.
The main service's data is located at a database.
I was thinking about having the monitor service query the database in intervals of 1 second, within a loop, issued by an http request done by the remote client.
Now the actual serving of this monitoring will be done by a client http request, which will make the script loop (written in C#) and when new data is detected it'll aggregate that data into that one looping request output buffer, send it, and exit the loop, thus finishing the request.
The client will have to issue a new request in order to keep getting updates.
This is actually exactly like TCP (precisely like Windows IOCP); You request the service for data and wait for it. When it arrives you fire another request.
My actual question is: Have you done it before? How did it go? Am I limited by some (configurable) limits imposed by the IIS/ASP.NET framework? What are my limits in such situation, or, what are better options without complicating things too much?
Note that I do not expect many such monitoring requests at a time, maybe a few dozens.
This means however that 10 such concurrent monitoring requests will keep 10 threads busy, and the question is; Can it hurt IIS/performance? How will IIS handle 10 busy threads? Will it issue more? What are the limits? This is just one example of a limit I can think of.
I think you main concern in this situation would be timeouts, which are pretty much configurable. But I think that it is a wrong solution - you'd be better of with some background service, running constantly/periodically, and writing the monitoring data to some data store and then your monitoring page would just return it upon request.
if you want your page to display something only if the monitorign data is available- implement it with ajax - on page load query monitoring service, then if some monitoring events are available- render them, if not- sleep and query again.
IMO this would be a much better solution than a reallu long running requests.
I think it won't be a very good idea to monitor a service using ASP.NET due to the following reasons...
What happens when your application pool crashes?
What if you decide to do IISReset? Which application will come up first... the main app, or the monitoring app?
What if the monitoring application hangs due to load?
What if the load is already high on the Main Service. Wouldn't monitoring it every 1 sec, increase the load on the Primary Service, as well as IIS?
You get the idea...
Background:
Enterprise application - very will written for its time in 2004.
Stack:
.NET, Heavy use of Remoting, ASMX style web services, SQL Server
Problem:
The application allows user to go through various wizards for lack of a better term, all of their actions are stored in what we call "wiz state", which is essentially XML that is persisted to a SQL server database very frequently because we allow users to pause/resume their application. Often in these wizards, the XML that comprises the wizard state grows very large, I'm talking 5-8 MB of data, and we noticed that when we had a sudden influx of simultaneous users, we started receiving occasional timeouts against the database, because a lot of what the wizard state is comprised of, is keeping track of collections of "things". Sometimes these custom collections grow very large.
Question:
We were in a meeting today and we're expecting a flurry of activity in October that will test the system like never before, and possibly result in huge wizard states that go back and forth from the web server to the database. The crux of the situation is that there is only one database and one web server.
For arguments sake, because of the complexity of the application, lets say adding any kind of clustering/mirroring to increase database throughput is out of the question. I spoke up in the meeting and said the quickest way to address this in the shortest time period would be to add more servers to the front end web application so the load could be distributed amongst web servers. The development lead said I was completely wrong and it would have no effect because we only have one database, so adding more web power would do nothing. He is having one of the other developers reduce the xml bloat that we persist frequently to the database. Probably in the long run, reducing the size of the xml that we pass back and forth is the right idea, but will adding additional web servers truly have no effect, I just think in terms of simultaneous users, it should help.
Any responses thoughts are appreciated, proof that more web servers would help would be pure win.
Thanks.
EDIT: We use binary serialization to store the XML in the database in an image field.
I haven't heard anything about locating the "bottlenecks". Isn't that the first thing to do? Here's the method I use.
Otherwise you're just investing in guesses. That won't work.
I've been in meetings like that, where everybody gets excited throwing ideas around, and "management" wants to make "decisions", but it's the blind leading the blind. Knuckle down and find out what's going on. You can't do that in meetings.
Some time ago I looked at a performance problem with some similarity to yours. The biggest "bottleneck" was in writing and parsing XML, with attendant memory allocation, setup, and destruction. Then there were others as well. You might find the same thing, or something different.
P.S. I keep quoting "bottleneck" because all the performance problems I've found have been nothing at all like the necks of bottles. Rather they are like way over-bushy call trees that need radical pruning, such as making and reading mountains of XML for no good reason.
If the rate at which the data is written by SQL is the bottleneck, feeding data to SQL more quickly should have no effect.
I am not sure exactly what the data structure is, but perhaps compressing the XML data on the web server(s) before writing may have a positive effect.
If the bottleneck is the database, then more web services will not help you a lot.
The problem may be that the problem is not only the size of the data, but the number of concurrent request to the same table. The number of writes will be the big problem. If your XML write is in a transaction with other queries you may try to break out the XML write from that transaction to reduce locking time of the XML table.
As stated by vdeych you may try compression to reduce the data size. (That would increase the load on the web servers.)
You may also try caching the data. Only read from the SQL server if the data is not already in the cache. Make sure you don't update the SQL server if your data has not changed.
No one seems to have suggest this, what about replacing your XML serialization of your wizard with JsonSerialization.
Not only should this give you a minor boost in performance in the serialization itself since both the DataContractSerializer (faster) and Newtonsoft Json.NET (fastest) out perform the XML serializers in .NET. This should easily reduce the size of your object graph by upwards of 50% or more (depending on number of properties vs large strings in the XML).
This should dramatically lower the IO that is inflicted upon Sql server. This should also limit the amount of scope required to alter your application significantly (assuming it's well designed and works through common calls for serialization/deserialization).
If you choose to go this route also invest time comparing BSON vs JSON as I think it would be likely that the binary encoded one will offer even more space savings (and further IO reduction) due to the size of your object graphs.
I'm not a .NET expert but maybe using a binary serialization would increase throughput. Making sure that the XML isn't stored as text (fairly obvious but thought I'd mention it). Also relational databases are best for storing relational data, so perhaps substituting an ORM layer in place of the serialization (sounds feasible) could speed things up.
Mike is spot on, without understanding the resource constaint leading to the performance issues, no amount of discussion will resolve the problem. I'll add that socket timeouts that affect running statements are a symptom, and are never imposed by SQL Server, they're an artifact of your driver configuration or a firewall or similar device between app and db imposing them (unless you're talking about timeouts for new connections, then you have a host in serious distress under load).
Given your symptom is database timeouts, you need to start there. If they're indicative of long running statements that result in a socket timeout, use SQL Server profiler to capture the workload while simultaneously monitoring system resources. Given it's a mature application and the type of workload you mention, it's unlikely to be statement tuning related, it probably boils down to resource limitations CPU, memory or disk IO capacity
This Technet guide is a very good place to start:
http://technet.microsoft.com/en-us/library/cc966540.aspx
If it's resource contention, then it's a simple discussion about how the resource contention can be tuned, configured for or addressed by adding more of whatever is needed.
Edit: I should add that given a database performance issue, more applications servers is likely to worsen the problem as you increase the amount of concurrency, that might otherwise be kept in check by connection pool, request processing or other limits.