Streamed and In-Memory Multimedia
Processing
Parallel data processing has been the mainstream
of designing efficient data-processing
platforms so that data could be processed in a
distributed and parallel manner, improving the
throughput of data processing. MapReduce
(http://mapreduce.sandia.gov) is the most representative
paradigm. Today’s multimedia data
is streamed in nature—that is, the data is generated
and updated over time. An effective multimedia
data-processing system must process the
data in a streamed manner. Storm (https://
storm.apache.org) is one such system.
predictions (see Figure 6).