CSC 7103 20090915

CSC 7103 20090915 - FIFO ordering is implemented by...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
CSC 7103 9/15/2009 Page 1 of 1 IPC When using the mailbox source/destination process in the kernel, the kernel must be implemented to choose how many messages are to be kept and for how long. Mailboxes basically provide a global buffer for the process. Once the mailbox is full, we fall back to the idea of discarding the messages and resending them. They also provide an abstraction so that process ids do not need to be known, messages are sent to and received from the mailbox. Ports are specific examples of mailboxes. They are finite and usually implemented with FIFO queues. Mailboxes/ports are the most common implementations of primitive message passing. Multicast or Group Communication Flooding is a trivial method for solving the problem of reliable message delivery.
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: FIFO ordering is implemented by stamping every message. Causal ordering can be implemented by: 1. Each message is time stamped by a vector where each entry in the vector is the number of messages received by the sender from that group member. 2. Accept a message from process i if a) you have received all previous messages from i and b) you have received all messages seen by i. Otherwise, delay accepting the message. 3. Reject any duplicated message. Figure 1 Causal ordering diagram Total ordering means that all messages are received in the order they were sent out in. A multicast is not successful unless all recipients receive the message....
View Full Document

Ask a homework question - tutors are online