{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}


2013 2 101513 information for migration decision

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

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Ê༌  Sprite, LSF à༎ Centralized. Ê༌  MOSIX estimates load by the average length of the ready queue over a fixed time period. Ê༌  Distributed algorithms: Ê༌  RoC- LB uses load change over time. Ê༌  More scalable. Ê༌  Load information storage issue. (Inefficiency) Ê༌  Problem: Ê༌  Load units (processes) are not equal: Ê༌  Centralized algorithms: Ê༌  Bottleneck. (Inefficiency) Ê༌  Different sizes (running times) à༎Many short tasks can be equivalent to a long one. Ê༌  Single point of failure. (Inefficiency) Ê༌  CPU- bound vs. I/O- bound processesà༎ must have a mixture to Ê༌  Fast, up- to- date information about all nodes. effectively utilize the resources. 49 50 If I would build a system, how would I do it? If I would build a system, how would I do it? Ê༌  (1) Load balancing through full process migration (no home- node Ê༌  (2) Migration decision is distributed as in MOSIX à༎ more dependency) à༎ more failure tolerant. scalable. Ê༌  Checkpoints at application level à༎ not transparent but produces Ê༌  Ê༌  Ê༌  Ê༌  Ê༌  But change the load representation to instead of basing on significantly smaller checkpoint files that are also machine- independent. Replicate these checkpoints at a configurable level on different nodes. Periodically save checkpoints and update replicas. If a node holding a replica does not respond, choose a replacement. On the other hand, if a timeout occurred at these nodes, one will take over the process. When a process has terminate...
View Full Document

{[ snackBarMessage ]}