{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}


Add the requester node to the sink table 2 if it is a

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: is Ê༌  Migration to a node for load balancing: transferred. Ê༌  CPU intensive applications. Ê༌  Most system calls are forwarded to the source. Figure 7 from [1] 31 32 LSF (Load Sharing Facility) Ê༌  LSF achieves load balancing through: Ê༌  Initial placements of processes (primary goal). Ê༌  Process migration with checkpointing (using Condor’s checkpointing). Ê༌  Checkpoints, kills, then restart the process on destination node. Ê༌  Centralized load information management: Ê༌  A single master node. Ê༌  If it fails, another node promotes itself (how?). (4) LSF Ê༌  Other nodes send their information to the master periodically. Ê༌  The master is responsible of scheduling. Ê༌  Based on load information and processes requirements. 33 34 Rate of Change Load Balancing (RoC- LB) A (5) Rate of Change Load Balancing (RoC- LB) Campos and Scherson [2] 35 B A A B Equalizing load B A B Utilize idle nodes Ê༌  The idea: the migration decision should depend on how the load changes in time: Ê༌  To predict future overload or starvation. Ê༌  To minimize node idle time. 36 6 10/9/13 Migration Algorithm (When?) Ê༌  Each node periodically computes the change in its load since Migration Algorithm (When?) First Exception Critical t (CT) High t (HT) Low t (LT) the last time interval à༎difference in load (DL). Sink Ê༌  Assuming that the DL will remain the same, each node computes how many time intervals it needs to become idle. Ê༌  If this time less than the network delay (ND) à༎ initiate migration request. Neutral Load Source Ê༌...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online