The distributed file system duplicates three copies of the uploaded
files and stores the copies on different storage servers in order to
achieve fault tolerance. The CloudBook system deploys the vector
graphic converter on all data nodes; therefore, the CloudBook can
dynamically dispatch the file requests from users to the data nodes. If
a data node receives the job request from the job tracker, the data
node loads the required files from its storage and converts the files to
the vector graphic format. Fig. 7 presents the architecture of cloud
platform in the CloudBook system.