SKR 5302: Advanced Distributed Computing

5.7. Asynchronous operation

Asynchronous operation

Making invocation concurrently

• The middleware provides only blocking invocations, but the application spawns multiple threads to perform blocking invocations concurrently. 
• Example is a web browser.
• A web page typically contains several images and may contain many. 
• The browser does not need to obtain the images in a particular sequence, so it makes several concurrent requests at a time. 
• That way, the time taken to complete all the image requests is typically lower than the delay that would result from making the requests serially. 
• Not only is the total communication delay less, in general, but the browser can overlap computation such as image rendering with communication. 
  
  

  Figure : Times for serialized and concurrent invocations
  
  
  
  

Asynchronous invocations

• It is made with a non-blocking call, which returns as soon as the invocation request message has been created and is ready for dispatch. 
• The client uses a separate call to collect the result of the invocation.
• Example: Mercury communication system [Liskov and Shrira 1988]
• An asynchronous operation returns an object called a promise. 
• Eventually, when the invocation succeeds or failed, the Mercury system places the status and any return values in the promise. 
• The caller uses the claim operation to obtain the result from the promise. 
• The claim operation blocks until the promise is ready, whereupon it returns the results or exceptions from the call. 
• The ready operation is available for testing a promise without blocking – it returns true or false according to whether the promise is ready or blocked. 

Operating system architecture

An open distributed system should make it possible to:

• Run only that system software at each computer that is necessary for it to carry out its particular role in the system architecture – 
  • system software requirements can vary between, for example, mobile phones and server computers, and loading redundant modules wastes memory resources; 
• Allow the software (and the computer) implementing any particular service to be changed independently of other facilities;
• Allow for alternatives of the same service to be provided, when this is required to suit different users or applications;
• Introduce new services without harming the integrity of the existing ones. 

Monolithic kernels

• Massive – it performs all basic operating system functions and takes up in the order of megabytes of code and data – and that it is undifferentiated, i.e. it is coded in a non-modular way. 
• To a large extent, it is intractable – altering any individual software component to adapt it to changing requirements is difficult. 
• Example: UNIX operating system kernel, Sprite network operating system [Ousterhout et al. 1988]
• Can contain some server processes that execute within its address space. 

Microkernels

• The microkernel appears as a layer between the hardware layer and a layer consisting of major system components called subsystems. 
• Middleware may use the facilities of the microkernel directly, or uses a language runtime support subsystem, or a higher-level operating system interface provided by an operating system emulation subsystem. 
  
  

  Figure : Monolithic kernel and microkernel
  
  
  

Comparison