Processes Chapter 3 Processes in Distributed Systems Processes

Processes Chapter 3

Processes in Distributed Systems • Processes and threads – Introduction to threads – Distinction between threads and processes • Threads in distributed systems – Clients/servers • Code migration – Motivation and models – Migration in heterogeneous systems

Processes • Virtual processors – Created by OS to execute a program – Process table to keep track of virtual processors • Process is a program in execution – Executed on one of the virtual processors • Operating systems ensure that processes are independent and transparent – Do not affect each other’s correctness – Resource sharing is transparent • Address space

Costs of Processes • Processes are costly • Creating a process requires creation of an entire new address space – Initializing memory segment, copying instructions to text segment, initializing stack • Switching b/w processes is costly as well – Saving CPU context, modify registers of the memory management unit, TLBs, swapping processes (if number of processors are larger than what can fit in main memory)

Threads • Similar to a process – Perceived as execution of (a part of) program – Information maintained for sharing CPU is minimal • Context of threads is captured by CPU context – May be a little more information is needed for management (like locks) • Very little overheads – Thread switching is easy • Can provide performance gains

Threads in Non-Distributed Systems • With processes a blocking system call blocks the entire process – Interactive programs like spreadsheets – Computations cannot proceed while waiting on user inputs or vice-versa – Having two threads can solve problems • Parallel programs – Multiple threads, single address space • Cooperating programs need to communicate – IPCs are costly

Drawbacks of IPCs • Needs kernel intervention (due to separate address spaces) • Multiple context switches • Context switches are costly (changing MMU, flushing TLB)

Implementing Threads • Implemented as package – Creation, destruction, synchronization • Simpler approach – User level thread library – All thread management is done at user level – Kernel not even that there are multiple threads • Thread creation, destruction and context switching are cheap – Creating stack – Saving registers • Drawback – Blocking system call blocks entire process

Lightweight Process • • A hybrid approach Lightweight process runs in context of single process Reside in kernel System also offers user-level threads – All operations occur in user mode • Thread package can be shared by multiple LWPs • LWP is created by kernel and given its own stack

Light-Weight Process

LWPs Continued • LWP has a scheduling routine and searches for a thread to execute – Thread pool • Context switching to a thread if found • On mutex, context switching happens if a runnable thread is found – LWP need not be informed • LWP context switching happens on a blocking system call • Several advantages – Cheap, will not block entire process, extends to multiprocessing environment