|
| Copyright © Cloud Software Group, Inc. All Rights Reserved |
Standard multicast and broadcast protocols are not reliable and are unable to detect lost messages. Under normal conditions, Rendezvous reliable multicast protocols ensure that all operational hosts either receive each multicast message or detect the loss of a message. (For details and limitations, see DATALOSS on page 250.)Reliable delivery compensates for brief network failures. The receiving Rendezvous daemon detects missing packets and requests that the sending daemon retransmit them. The sending daemon stores outbound messages for a limited period of time—called the reliability interval—so it can retransmit the information upon request. It discards old messages after the time period elapses, and cannot retransmit after that time.The Rendezvous daemon requires that the physical network and packet recipients are working. The Rendezvous daemon does not guarantee delivery to components that fail and do not recover for periods exceeding the reliability interval. (For stronger assurances of delivery, see Certified Message Delivery.)When a sending daemon receives a retransmission request for data it has already discarded, it notifies the requesting daemon that it cannot retransmit it. One or both daemons present error advisories to indicate that this situation has occurred (see DATALOSS on page 250).For a complete discussion the various ways to control reliability, the interaction among those ways, and reasonable values, see Reliability and Message Retention Time in .The reliable multicast protocol delivers messages once and only once to each subscription, despite multiple transient network failures.
|
| Copyright © Cloud Software Group, Inc. All Rights Reserved |