I have built software that I deploy on Windows 2003 server. The software runs as a service continuously and it's the only application on the Windows box of importance to me. Part of the time, it's retrieving data from the Internet, and part of the time it's doing some computations on that data. It's multi-threaded -- I use thread pools of roughly 4-20 threads.
I won't bore you with all those details, but suffice it to say that as I enable more threads in the pool, more concurrent work occurs, and CPU use rises. (as does demand for other resources, like bandwidth, although that's of no concern to me -- I have plenty)
My question is this: should I simply try to max out the CPU to get the best bang for my buck? Intuitively, I don't think it makes sense to run at 100% CPU; even 95% CPU seems high, almost like I'm not giving the OS much space to do what it needs to do. I don't know the right way to identify best balance. I guessing I could measure and measure and probably find that the best throughput is achived at a CPU avg utilization of 90% or 91%, etc. but...
I'm just wondering if there's a good rule of thumb about this??? I don't want to assume that my testing will take into account all kinds of variations of workloads. I'd rather play it a bit safe, but not too safe (or else I'm underusing my hardware).
What do you recommend? What is a smart, performance minded rule of utilization for a multi-threaded, mixed load (some I/O, some CPU) application on Windows?
Yep, I'd suggest 100% is thrashing so wouldn't want to see processes running like that all the time. I've always aimed for 80% to get a balance between utilization and room for spikes / ad-hoc processes.
An approach i've used in the past is to crank up the pool size slowly and measure the impact (both on CPU and on other constraints such as IO), you never know, you might find that suddenly IO becomes the bottleneck.
CPU utilization shouldn't matter in this i/o intensive workload, you care about throughput, so try using a hill climbing approach and basically try programmatically injecting / removing worker threads and track completion progress...
If you add a thread and it helps, add another one. If you try a thread and it hurts remove it.
Eventually this will stabilize.
If this is a .NET based app, hill climbing was added to the .NET 4 threadpool.
UPDATE:
hill climbing is a control theory based approach to maximizing throughput, you can call it trial and error if you want, but it is a sound approach. In general, there isn't a good 'rule of thumb' to follow here because the overheads and latencies vary so much, it's not really possible to generalize. The focus should be on throughput & task / thread completion, not CPU utilization. For example, it's pretty easy to peg the cores pretty easily with coarse or fine-grained synchronization but not actually make a difference in throughput.
Also regarding .NET 4, if you can reframe your problem as a Parallel.For or Parallel.ForEach then the threadpool will adjust number of threads to maximize throughput so you don't have to worry about this.
-Rick
Assuming nothing else of importance but the OS runs on the machine:
And your load is constant, you should aim at 100% CPU utilization, everything else is a waste of CPU. Remember the OS handles the threads so it is indeed able to run, it's hard to starve the OS with a well behaved program.
But if your load is variable and you expect peaks you should take in consideration, I'd say 80% CPU is a good threshold to use, unless you know exactly how will that load vary and how much CPU it will demand, in which case you can aim for the exact number.
If you simply give your threads a low priority, the OS will do the rest, and take cycles as it needs to do work. Server 2003 (and most Server OSes) are very good at this, no need to try and manage it yourself.
I have also used 80% as a general rule-of-thumb for target CPU utilization. As some others have mentioned, this leaves some headroom for sporadic spikes in activity and will help avoid thrashing on the CPU.
Here is a little (older but still relevant) advice from the Weblogic crew on this issue: http://docs.oracle.com/cd/E13222_01/wls/docs92/perform/basics.html#wp1132942
If you feel your load is very even and predictable you could push that target a little higher, but unless your user base is exceptionally tolerant of periodic slow responses and your project budget is incredibly tight, I'd recommend adding more resources to your system (adding a CPU, using a CPU with more cores, etc.) over making a risky move to try to squeeze out another 10% CPU utilization out of your existing platform.
Related
I have a high-performance software server application that is expected to get increased traffic in the next few months.
I was wondering what approach or methodology is good to use in order to gauge if the server still has the capacity to handle this increased load?
I think you're looking for Stress Testing and the scenario would be something like:
Create a load test simulating current real application usage
Start with current number of users and gradually increase the load until
you reach the "increased traffic" amount
or errors start occurring
or you start observing performance degradation
whatever comes the first
Depending on the outcome you either can state that your server can handle the increased load without any issues or you will come up with the saturation point and the first bottleneck
You might also want to execute a Soak Test - leave the system under high prolonged load for several hours or days, this way you can detect memory leaks or other capacity problems.
More information: Why ‘Normal’ Load Testing Isn’t Enough
Test the product with one-tenth the data and traffic. Be sure the activity is 'realistic'.
Then consider what will happen as traffic grows -- with the RAM, disk, cpu, network, etc, grow linearly or not?
While you are doing that, look for "hot spots". Optimize them.
Will you be using web pages? Databases? Etc. Each of these things scales differently. (In other words, you have not provided enough details in your question.)
Most canned benchmarks focus on one small aspect of computing; applying the results to a specific application is iffy.
I would start by collecting base line data on critical resources - typically, CPU, memory usage, disk usage, network usage - and track them over time. If any of those resources show regular spikes where they remain at 100% capacity for more than a fraction of a second, under current usage, you have a bottleneck somewhere. In this case, you cannot accept additional load without likely outages.
Next, I'd start figuring out what the bottleneck resource for your application is - it varies between applications, but in most cases it's the bottleneck resource that stops you from scaling further. Your CPU might be almost idle, but you're thrashing the disk I/O, for instance. That's a tricky process - load and stress testing are the way to go.
If you can resolve the bottleneck by buying better hardware, do so - it's much cheaper than rewriting the software. If you can't buy better hardware, look at load balancing. If you can't load balance, you've got to look at application architecture and implementation and see if there are ways to move the bottleneck.
It's quite typical for the bottleneck to move from one resource to the next - you've got CPU to behave, but now when you increase traffic, you're spiking disk I/O; once you resolve that, you may get another CPU challenge.
I have 12 tasks to run on an octo-core machine. All tasks are CPU intensive and each will max out a core.
Is there a theoretical reason to avoid stacking tasks on a maxed out core (such as overhead, swapping across tasks) or is it faster to queue everything?
Task switching is a waste of CPU time. Avoid it if you can.
Whatever the scheduler timeslice is set to, the CPU will waste its time every time slice by going into the kernel, saving all the registers, swapping the memory mappings and starting the next task. Then it has to load in all its CPU cache, etc.
Much more efficient to just run one task at a time.
Things are different of course if the tasks use I/O and aren't purely compute bound.
Yes it's called queueing theory https://en.wikipedia.org/wiki/Queueing_theory. There are many different models https://en.wikipedia.org/wiki/Category:Queueing_theory for a range of different problems I'd suggest you scan them and pick the one most applicable to your workload then go and read up on how to avoid the worst outcomes for that model, or pick a different, better, model for dispatching your workload.
Although the graph at this link https://commons.wikimedia.org/wiki/File:StochasticQueueingQueueLength.png applies to Traffic it will give you an idea of what is happening to response times as your CPU utilisation increases. It shows that you'll reach an inflection point after which things get slower and slower.
More work is arriving than can be processed with subsequent work waiting longer and longer until it can be dispatched.
The more cores you have the further to the right you push the inflection point but the faster things go bad after you reach it.
I would also note that unless you've got some really serious cooling in place you are going to cook your CPU. Depending on it's design it will either slow itself down, making your problem worse, or you'll trigger it's thermal overload protection.
So a simplistic design for 8 cores would be, 1 thread to manage things and add tasks to the work queue and 7 threads that are pulling tasks from the work queue. If the tasks need to be performed within a certain time you can add a TimeToLive value so that they can be discarded rather than executed needlessly. As you are almost certainly running your application in an OS that uses a pre-emptive threading model consider things like using processor affinity where possible because as #Zan-Lynx says task/context switching hurts. Be careful not to try to build your OS'es thread management again as you'll probably wind up in conflict with it.
tl;dr: cache thrash is Bad
You have a dozen tasks. Each will have to do a certain amount of work.
At an app level they each processed a thousand customer records or whatever. That is fixed, it is a constant no matter what happens on the hardware.
At the the language level, again it is fixed, C++, java, or python will execute a fixed number of app instructions or bytecodes. We'll gloss over gc overhead here, and page fault and scheduling details.
At the assembly level, again it is fixed, some number of x86 instructions will execute as the app continues to issue new instructions.
But you don't care about how many instructions, you only care about how long it takes to execute those instructions. Many of the instructions are reads which MOV a value from RAM to a register. Think about how long that will take. Your computer has several components to implement the memory hierarchy - which ones will be involved? Will that read hit in L1 cache? In L2? Will it be a miss in last-level cache so you wait (for tens or hundreds of cycles) until RAM delivers that cache line? Did the virtual memory reference miss in RAM, so you wait (for milliseconds) until SSD or Winchester storage can page in the needed frame? You think of your app as issuing N reads, but you might more productively think of it as issuing 0.2 * N cache misses. Running at a different multi-programming level, where you issue 0.3 * N cache misses, could make elapsed time quite noticeably longer.
Every workload is different, and can place larger or smaller demands on memory storage. But every level of the memory hierarchy depends on caching to some extent, and higher multi-programming levels are guaranteed to impact cache hit rates. There are network- and I/O-heavy workloads where very high multi-programming levels absolutely make sense. But for CPU- and memory-intensive workloads, when you benchmark elapsed times you may find that less is more.
We've just bought a 32-core Opteron machine, and the speedups we get are a little disappointing: beyond about 24 threads we see no speedup at all (actually gets slower overall) and after about 6 threads it becomes significantly sub-linear.
Our application is very thread-friendly: our job breaks down into about 170,000 little tasks which can each be executed separately, each taking 5-10 seconds. They all read from the same memory-mapped file of size about 4Gb. They make occasional writes to it, but it might be 10,000 reads to each write - we just write a little bit of data at the end of each of the 170,000 tasks. The writes are lock-protected. Profiling shows that the locks are not a problem. The threads use a lot of JVM memory each in non-shared objects and they make very little access to shared JVM objects and of that, only a small percentage of accesses involve writes.
We're programming in Java, on Linux, with NUMA enabled. We have 128Gb RAM. We have 2 Opteron CPU's (model 6274) of 16 cores each. Each CPU has 2 NUMA nodes. The same job running on an Intel quad-core (i.e. 8 cores) scaled nearly linearly up to 8 threads.
We've tried replicating the read-only data to have one-per-thread, in the hope that most lookups can be local to a NUMA node, but we observed no speedup from this.
With 32 threads, 'top' shows the CPU's 74% "us" (user) and about 23% "id" (idle). But there are no sleeps and almost no disk i/o. With 24 threads we get 83% CPU usage. I'm not sure how to interpret 'idle' state - does this mean 'waiting for memory controller'?
We tried turning NUMA on and off (I'm referring to the Linux-level setting that requires a reboot) and saw no difference. When NUMA was enabled, 'numastat' showed only about 5% of 'allocation and access misses' (95% of cache misses were local to the NUMA node). [Edit:] But adding "-XX:+useNUMA" as a java commandline flag gave us a 10% boost.
One theory we have is that we're maxing out the memory controllers, because our application uses a lot of RAM and we think there are a lot of cache misses.
What can we do to either (a) speed up our program to approach linear scalability, or (b) diagnose what's happening?
Also: (c) how do I interpret the 'top' result - does 'idle' mean 'blocked on memory controllers'? and (d) is there any difference in the characteristics of Opteron vs Xeon's?
I also have a 32 core Opteron machine, with 8 NUMA nodes (4x6128 processors, Mangy Cours, not Bulldozer), and I have faced similar issues.
I think the answer to your problem is hinted at by the 2.3% "sys" time shown in top. In my experience, this sys time is the time the system spends in the kernel waiting for a lock. When a thread can't get a lock it then sits idle until it makes its next attempt. Both the sys and idle time are a direct result of lock contention. You say that your profiler is not showing locks to be the problem. My guess is that for some reason the code causing the lock in question is not included in the profile results.
In my case a significant cause of lock contention was not the processing I was actually doing but the work scheduler that was handing out the individual pieces of work to each thread. This code used locks to keep track of which thread was doing which piece of work. My solution to this problem was to rewrite my work scheduler avoiding mutexes, which I have read do not scale well beyond 8-12 cores, and instead use gcc builtin atomics (I program in C on Linux). Atomic operations are effectively a very fine grained lock that scales much better with high core counts. In your case if your work parcels really do take 5-10s each it seems unlikely this will be significant for you.
I also had problems with malloc, which suffers horrible lock issues in high core count situations, but I can't, off the top of my head, remember whether this also led to sys & idle figures in top, or whether it just showed up using Mike Dunlavey's debugger profiling method (How can I profile C++ code running in Linux?). I suspect it did cause sys & idle problems, but I draw the line at digging through all my old notes to find out :) I do know that I now avoid runtime mallocs as much as possible.
My best guess is that some piece of library code you are using implements locks without your knowledge, is not included in your profiling results, and is not scaling well to high core-count situations. Beware memory allocators!
I'm sure the answer will lie in a consideration of the hardware architecture. You have to think of multi core computers as if they were individual machines connected by a network. In fact that's all that Hypertransport and QPI are.
I find that to solve these scalability problems you have to stop thinking in terms of shared memory and start adopting the philosophy of Communicating Sequential Processes. It means thinking very differently, ie imagine how you would write the software if your hardware was 32 single core machines connected by a network. Modern (and ancient) CPU architectures are not designed to give unfettered scaling of the sort you're after. They are designed to allow many different processes to get on with processing their own data.
Like everything else in computing these things go in fashions. CSP dates back to the 1970s, but the very modern and Java derived Scala is a popular embodiment of the concept. See this section on Scala concurrency on Wikipedia.
What the philosophy of CSP does is force you to design a data distribution scheme that fits your data and the problem you're solving. That's not necessarily easy, but if you manage it then you have a solution that will scale very well indeed. Scala may make it easier to develop.
Personally I do everything in CSP and in C. It's allowed me to develop a signal processing application that scales perfectly linearly from 8 cores to several thousand cores (the limit being how big my room is).
The first thing you're going to have to do is actually use NUMA. It isn't a magic setting that you turn on, you have to exploit it in your software's architecture. I don't know about Java, but in C one would bind a memory allocation to a specific core's memory controller (aka memory affinity), and similarly for threads (core affinity) in cases where the OS doesn't get the hint.
I presume that your data doesn't break down into 32 neat, discrete chunks? It's difficult to give advice without knowing exactly the data flows implicit in your program. But think about it in terms of data flow. Draw it out even; Data Flow Diagrams are useful for this (another ancient graphical formal notation). If your picture shows all your data going through a single object (eg through a single memory buffer) then it's going to be slow...
I assume you have optimized your locks, and synchronization made a minimum. In such a case, it still depends a lot on what libraries you are using to program in parallel.
One issue that can happen even if you have no synchronization issue, is memory bus congestion. This is very nasty and difficult to get rid of.
All I can suggest is somehow make your tasks bigger and create fewer tasks. This depends highly on the nature of your problem. Ideally you want as many tasks as the number of cores/threads, but this is not easy (if possible) to achieve.
Something else that can help is to give more heap to your JVM. This will reduce the need to run Garbage Collector frequently, and speeds up a little.
does 'idle' mean 'blocked on memory controllers'
No. You don't see that in top. I mean if the CPU is waiting for memory access, it will be shown as busy. If you have idle periods, it is either waiting for a lock, or for IO.
I'm the Original Poster. We think we've diagnosed the issue, and it's not locks, not system calls, not memory bus congestion; we think it's level 2/3 CPU cache contention.
To reiterate, our task is embarrassingly parallel so it should scale well. However, one thread has a large amount of CPU cache it can access, but as we add more threads, the amount of CPU cache each process can access gets lower and lower (the same amount of cache divided by more processes). Some levels on some architectures are shared between cores on a die, some are even shared between dies (I think), and it may help to get "down in the weeds" with the specific machine you're using, and optimise your algorithms, but our conclusion is that there's not a lot we can do to achieve the scalability we thought we'd get.
We identified this as the cause by using 2 different algorithms. The one which accesses more level 2/3 cache scales much worse than the one which does more processing with less data. They both make frequent accesses to the main data in main memory.
If you haven't tried that yet: Look at hardware-level profilers like Oracle Studio has (for CentOS, Redhat, and Oracle Linux) or if you are stuck with Windows: Intel VTune. Then start looking at operations with suspiciously high clocks per instruction metrics. Suspiciously high mean a lot higher than the same code on a single-numa, single-L3-cache machine (like current Intel desktop CPUs).
We have a transaction intensive process at one customer site running on a quad core server with four processors. The process is designed to take advantage of every core available. So in this installation, we take an input queue, divide it by 16th's and allocate each fraction of the queue to a core. It works well and keeps up with the transaction volume on the box.
Looking at the CPU utilization on the box, it never seems to go above 33%. Now we have a new customer with at least double the volume of the existing customer. Some of us are arguing that since CPU usage is way below maximum utilization, that we should go with the same configuration.
Others claim that there is no direct correlation between cpu utilization and transaction processing speed and since the logic of the underlying software module is based on the number of available cores, that it makes sense to obtain a box with proportionately more cores available for the new client to accommodate the increased traffic volume.
Does anyone have a sense as to who is right in this instance?
Thank you,
To determine the optimum configuration for your new customer, understanding the reason for low CPU usage is paramount.
Very likely, the reason is one of the following:
Your process is limited by memory bandwidth. In this case, faster RAM will help if supported by the motherboard. If possible, a redesign to limit the amount of data accessed during processing will improve performance. Adding more CPU cores will, on its own, do nothing to improve performance.
Your process is limited by disk I/O. Using faster disk connections (SATA etc.) and/or upgrading to a SSD might help, but more CPU power will not.
Your process is limited by synchronization contention. In this case, adding more threads for more cores might even be counter productive. Redesigning your algorithm might help in this case.
Having said this, I have also seen situations where processes that are definitely CPU bound fail to achieve 100% CPU usage on modern processors (Core i7 etc.) because in certain turbo boost relevant cases, task manager will show less than 100%.
As 9000 said, you need to find out what your bottlenecks are when under load. Perfmon might provide enough data to find out.
Another afterthought: You could limit your process on the existing machine to part of the cores (but still at least 30% so that theoretically, CPU doesn't become a bottleneck due to this limitation) and check if overall throughput degrades. If it does not, adding more cores will not improve performance.
Imagine I have two (three, four, whatever) tasks that have to run in parallel. Now, the easy way to do this would be to create separate threads and forget about it. But on a plain old single-core CPU that would mean a lot of context switching - and we all know that context switching is big, bad, slow, and generally simply Evil. It should be avoided, right?
On that note, if I'm writing the software from ground up anyway, I could go the extra mile and implement my own task-switching. Split each task in parts, save the state inbetween, and then switch among them within a single thread. Or, if I detect that there are multiple CPU cores, I could just give each task to a separate thread and all would be well.
The second solution does have the advantage of adapting to the number of available CPU cores, but will the manual task-switch really be faster than the one in the OS core? Especially if I'm trying to make the whole thing generic with a TaskManager and an ITask, etc?
Clarification: I'm a Windows developer so I'm primarily interested in the answer for this OS, but it would be most interesting to find out about other OSes as well. When you write your answer, please state for which OS it is.
More clarification: OK, so this isn't in the context of a particular application. It's really a general question, the result on my musings about scalability. If I want my application to scale and effectively utilize future CPUs (and even different CPUs of today) I must make it multithreaded. But how many threads? If I make a constant number of threads, then the program will perform suboptimally on all CPUs which do not have the same number of cores.
Ideally the number of threads would be determined at runtime, but few are the tasks that can truly be split into arbitrary number of parts at runtime. Many tasks however can be split in a pretty large constant number of threads at design time. So, for instance, if my program could spawn 32 threads, it would already utilize all cores of up to 32-core CPUs, which is pretty far in the future yet (I think). But on a simple single-core or dual-core CPU it would mean a LOT of context switching, which would slow things down.
Thus my idea about manual task switching. This way one could make 32 "virtual" threads which would be mapped to as many real threads as is optimal, and the "context switching" would be done manually. The question just is - would the overhead of my manual "context switching" be less than that of OS context switching?
Naturally, all this applies to processes which are CPU-bound, like games. For your run-of-the-mill CRUD application this has little value. Such an application is best made with one thread (at most two).
I don't see how a manual task switch could be faster since the OS kernel is still switching other processes, including yours in out of the running state too. Seems like a premature optimization and a potentially huge waste of effort.
If the system isn't doing anything else, chances are you won't have a huge number of context switches anyway. The thread will use its timeslice, the kernel scheduler will see that nothing else needs to run and switch right back to your thread. Also the OS will make a best effort to keep from moving threads between CPUs so you benefit there with caching.
If you are really CPU bound, detect the number of CPUs and start that many threads. You should see nearly 100% CPU utilization. If not, you aren't completely CPU bound and maybe the answer is to start N + X threads. For very IO bound processes, you would be starting a (large) multiple of the CPU count (i.e. high traffic webservers run 1000+ threads).
Finally, for reference, both Windows and Linux schedulers wake up every millisecond to check if another process needs to run. So, even on an idle system you will see 1000+ context switches per second. On heavily loaded systems, I have seen over 10,000 per second per CPU without any significant issues.
The only advantage of manual switch that I can see is that you have better control of where and when the switch happens. The ideal place is of course after a unit of work has been completed so that you can trash it all together. This saves you a cache miss.
I advise not to spend your effort on this.
Single-core Windows machines are going to become extinct in the next few years, so I generally write new code with the assumption that multi-core is the common case. I'd say go with OS thread management, which will automatically take care of whatever concurrency the hardware provides, now and in the future.
I don't know what your application does, but unless you have multiple compute-bound tasks, I doubt that context switches are a significant bottleneck in most applications. If your tasks block on I/O, then you are not going to get much benefit from trying to out-do the OS.