Algorithms for Server Rebuild - Scalable Continuous Media Streaming Systems - page 250

14.1.1 Sparing Schemes
Figure 14.1 shows the use of a spare server to store rebuilt data - hot sparing . Note that the
spare server is not used under normal mode operation. When a server fails, the data in the
failed server are then rebuilt and stored in the spare server. When the rebuild process finishes,
the spare server simply replaces the failed server as shown in Figure 14.2.
Figure 14.3 shows another approach of allocating spare units - distributed sparing , in which
the spare stripe units are distributed over all the servers. During normal mode operation, all
servers participate in serving client requests. After a server has failed, data in the failed server
are then rebuilt and stored in the spare stripe units. Unlike hot sparing, the rebuilt system has a
different configuration from the original configuration (Figure 14.4). Therefore, an additional
phase called system restoration is required to copy all rebuilt data onto another spare server.
That spare server can then replace the old server to resume normal system operation.

Spare

Server

S S

S 0 S 1 S 2 S 3 S 4

u 0 u 1 u 2 u 3 p 0 s 0

u 4 u 5 u 6 p 1 u 7 s 1

u 8 u 9 p 2 u 10 u 11 s 2

u 12 p 3 u 13 u 14 u 15

s 3

p 4 u 16 u 17 u 18 u 19

s 4

Figure 14.1 Storage configuration for hot sparing

S 0 S 1 S 2 S 3 S 4 S S

u 2

u 0 u 1 u 2 u 3 p 0

u 4 u 5 u 6 p 1 u 7 u 6

p 2

u 8 u 9 p 2 u 10 u 11

u 12 p 3 u 13 u 14 u 15 u 13

p 4 u 16 u 17 u 18 u 19 u 17

Figure 14.2 Storage configuration after lost-data rebuild in hot sparing