Threads and Processes Ariel Caplan 

Concurrency <= Parallelism 

Threads && Processes Both provide concurrent execution of code! Neither guarantee execution order! Both add overhead to your code! 

Threads vs. Processes Process Process Memory Memory Thread Thread Memory 

Threads vs. Processes @var @var @var = 1 @var = 1 @var @var @var = 1 

Threads vs. Processes @var @var @var = 1 @var = 1 @var @var @var = 2 

Threads vs. Processes @var @var @var = 1 @var @var @var = 2 @var = 2 

Threads vs. Processes @var @var @var = 1 @var @var @var = 2 @var = 2 WARNING:
 OVERSIMPLIFICATION 

Threads vs. Processes Threads share memory and state! Processes share nothing 

Threads vs. Processes in MRI Threads share memory and state! Processes share nothing! Threads are concurrent, not parallel, due to the GIL 
 (Global Interpreter Lock)! Processes operate in true parallel 

Parallelism Concurrency Processes Multi-Core System Single-Core System Threads JRuby, Rubinius, Nonblocking IO Standard Calculations in MRI The Bottom Line 

Implication: Sidekiq Concurrency For a server running only Sidekiq on MRI:! 1 Sidekiq process per core! Concurrency (number of threads) = 1 / CPU Time! Example: 8-core machine with 75% IO, 25% CPU:! 8 Sidekiq processes! Sidekiq concurrency of 1/(25%) = 4 

Thank You For Listening! amcaplan.ninja! @amcaplan! ↑! Questions