Updated: October 30th, 2014
Workers are the processes that allow your application to respond to incoming web requests. Regardless of the application server stack you select, it provides one or more workers per instance to run your application.
For Passenger, Unicorn, Thin or Mongrel, worker processes are allocated using an algorithm that takes into consideration ECUs and memory available to the instance to determine the number of workers to allocate, as shown in the table below.
As of the May 21st, 2013 stack release, we modified the algorithm that generates the worker counts to better use the available resources; the result is increased numbers of worker processes for most instance sizes.
|Instance Type||Memory (MB)||Swap (MB)||ECUs||Pool Size|
|General Purpose (M1) Small||1700||895||1||3||3|
|General Purpose (M1) Medium||3840||1343||2||7||7|
|Compute Optimized (C1) Medium||1740||1343||5||6||6|
|General Purpose (M1) Large||7680||8192||4||6||8|
|General Purpose (M1) Extra Large||15360||8192||8||12||16|
|Compute Optimized (C1) Extra Large||7168||8192||20||24||40|
|Memory Optimized (M2) Extra Large||17500||8192||6.5||8||13|
|Memory Optimized (M2) Double Extra Large||35000||8192||13||8||26|
|Memory Optimized (M2) Quadruple Extra Large||70000||8192||26||24||52|
|Storage Optimized (Hi1) Quadruple Extra Large||62000||8192||35||70||70|
|General Purpose (M3) Medium||3750||8192||3||-||6|
|General Purpose (M3) Large||7444||8192||6.5||-||13|
|General Purpose (M3) Extra Large||14960||8192||13||-||26|
|General Purpose (M3) Double Extra Large||30000||8192||26||-||52|
|Compute Optimized (C3) Large||3750||8192||7||-||14|
|Compute Optimized (C3) Extra Large||7461||8192||14||-||28|
|Compute Optimized (C3) Double Extra Large||14961||8192||28||-||56|
|Compute Optimized (C3) Quadruple Extra Large||30058||8192||55||-||100|
|Compute Optimized (C3) Octuple Extra Large||60066||8192||108||-||100|
|General Purpose (T2) Micro||998||8192||3.25||-||6|
|General Purpose (T2) Small||2004||8192||3.25||-||6|
|General Purpose (T2) Medium||3697||8192||6.5||-||13|
|Memory Optimized (R3) Large||15026||8192||6.5||-||13|
|Memory Optimized (R3) Extra Large||30383||8192||13||-||26|
|Memory Optimized (R3) Double Extra Large||61096||8192||26||-||52|
|Memory Optimized (R3) Quadruple Extra Large||122519||8192||52||-||100|
|Memory Optimized (R3) Octuple Extra Large||245373||8192||104||-||100|
We use "vCPUs" as a generic label for virtual CPU cores (and in some cases hyperthreads) available to your instance. Amazon (IaaS provider for Engine Yard Cloud) uses the term ECU, or Elastic Compute Units, to convey the relative compute power available on each instance type. This value is different than the virtual CPUs reported by the operating system, such as in top or /proc/cpuinfo, but better allows you to determine the amount of effective compute resources you need. According to Amazon, the definition of ECU is:
"One EC2 Compute Unit provides the equivalent CPU capacity of a 1.0-1.2 GHz 2007 Opteron or 2007 Xeon processor."
For the purpose of this feature, you can consider vCPUs to be synonymous with ECUs.
The values in the table (above) are based on generic parameters of an idealized application. However, if your app falls outside these parameters, you should consider tuning your environment, to either:
- Reduce the worker counts to improve responsiveness because your app requires more resources.
- Increase available workers for a lightweight app with high requests per minute.
You can tune your app by setting parameters on your environment. The sections below explain the various parameters that can be tuned. Once you know what values you want tuned, you need to open a support ticket to have them put in effect.
Important: In the case of General Purpose (M1) Medium and Compute Optimized (C1) Medium instances, there is an override in place to ensure that pool size from the values provided in the original algorithm are not reduced. If you tune one or more parameters, these overrides are removed and the generated values (based on your updated parameters) will be used instead.
This is the most direct way of modifying your pool size. By specifying
pool_size, you can set it regardless of resource availability. This is recommended only if you are fully aware of your resource needs and don't plan on changing your instance size in the future.
Slightly less forceful, the
min_pool_size parameter allows you to ensure your application always has the minimum number of workers available. Useful if using the smaller instance types and you don't mind overloading the resources — be careful though, overloaded instances may result in underperforming or even hung instances, potentially triggering a takeover.
If you find yourself using this and having poor responsiveness or takeovers, please consider upgrading to a larger instance size.
max_pool_size parameter allows you to ensure you don't have more workers than you need, regardless of resource availability. You may want to do this if your workers connect to external entities that cannot handle too many simultaneous connections.
This is the best and most accurate way of tuning your app based on memory usage. The standardized values above are based on an app in which each worker uses 250MB of memory, which is close the upper limit of usage we've seen across thousands of apps. If your app is significantly different than this, it is best to tune this value to get accurate resource usage calculations.
Some apps are compute heavy, where others barely use any of the processor capacity. The standard configuration provides 2 workers per ECU; tuning this to your app's needs can improve its responsiveness.
While it is not a good idea to go in to swap in most cases, it is a balancing act to provide the optimal number of workers. We have chosen a modest 25% usage of swap by the workers in order to provide a sufficient worker count without excessive paging. If you find that your app is spending a great deal of time in swap, you may want to change this value.
Note: It is usually due to CPU constraints rather than memory constraints that determines the workers available, but depending on how many workers you allow per CPU, this may not be the case. Under default settings, this value most significantly impacts the Compute Optimized (C1) Small, Medium, and Extra Large instances.
Default: 20 (%)
If you are running background workers or other processes that eat into your memory resources, you may want to specify a chunk of memory to reserve. By default, this is set to a generous value that accommodates the OS and support processes.
Default: 1500 (MB)
reserved_memory but also takes into consideration the database that shared the instance in a single-instance (solo) environment.
Default: 2000 (MB)
These settings are only applicable to single instance environments. By default, no consideration is taken to reserve ECU resources for the database, but if you have a database-heavy application, it would be beneficial to allocate some of the worker resources to the database instead. The number of workers represented by the
db_workers_per_ecu parameter is removed from each ECU, up to the
db_vcpu_max = 0;
db_workers_per_ecu = 0.5
Use the following form to determine how many workers your instances will have, and how much free memory and swap usage will be expected.
|Edit the fields below. Leave blank for default value, which is shown in grey italics when not editing the field.
Why are some of the swap allocations over 100 percent without tuning?
In the default configuration (no tuned fields, so using the values you see in gray), you might have noticed that some of the instances appear allocated above the swap limits of the instance (for example, 131.0%). This is because we wanted to ensure that the minimal worker counts from the previous allocation scheme were not reduced. In practice, this does not cause a problem for the vast majority of our users, because worker_memory_size and reserved_memory do not often reach the upper limit values in the default values above. However, if these values due cause you issues, this tool and the tuned values it produces will allow you to protect your app from being over-allocated.
Concurrent application servers, such as Puma (all Ruby runtimes) use a simple one worker per ECU rule, and rely on threading to scale to the incoming requests. These work best for concurrent Ruby implementations, but threadsafe apps properly configured can benefit from concurrency under MRI by using Puma.
If you have feedback or questions about this page, add a comment below. If you need help, submit a ticket with Engine Yard Support.