This book begins with a cloud-based Petabyte-scale RS data
integration and management infrastructure. Based on an OpenStack-enabled
Cloud infrastructure with flexible resources provisioning, a distributed RS
data integration system is built to easily integrate petabyte-scale multi-sensor,
multi-resolution RS data in various formats or metadata standards across
data centers. By virtue of the logical segmentation indexing (LSI) global data
organizing and spatial indexing model as well as the distributed No-SQL
database, an optimized object oriented data technology (OODT) based RS
data management system is put forward to serve RS data on-demand. Then, the
book is followed by high-performance remote sensing clouds for the data center
named pipsCloud. It incorporates the Cloud computing paradigm with clusterbased HPC systems for a quality of service (QoS) optimized Cloud so as to address the challenges from a system architecture point of view. Wherein, baremetal (BM) provisioning of the HPC cluster is proposed for less performance penalty; a data access pattern aware data layout strategy is employed for better data locality and finally optimized parallel data I/O; dynamic directed acyclic graph (DAG) scheduling is used for large-scale workflows. Likewise,
benefitting from an optimal distributed workflow and resource scheduling
strategy on a basis of minimal data transferring, Cloud-enabled RS data
producing infrastructure and services are delivered for planetary-scale RS data
analyzing across data centers.
This book is greatly supported by the National Natural Science Foundation
of China (No. U1711266), and it would appeal to remote sensing researchers
becoming aware of the great challenges lying in most earth sciences as well
as Cloud-empowered infrastructures and techniques available. Meanwhile, the
computer engineers and scientists from other domains with similar issues could
also be benefitted or inspired.