Cloud Storage Technology: Survey on Performance Aspects.

摘要

Today's datasets from every space continue to rise from multi sources and screens: mobile, web enabled devices, surveillance, human generated via social media, machine generated via smart processes, messages, HPC, Financial, Broadcast, Post-Production, etc. These sets are growing exponentially, quickly becoming billions of assets, and consuming petabytes of capacity at a staggering speed - all of which must be quickly and efficiently shared between multiple data centers and accessed by millions of elastic users globally dispersed from anywhere at any time. Compounded by the decrease of disks technology form factor, increased capacity, speed and reliability. As well as advances in: CPU performance, memory speed/size, and faster global network presence; the operating system and access methodology is left with a daunting task of handling multi-threaded billions of calls to the storage sub-system, rendering it therefore inefficient or in need of augmentation. At this capacity and global distribution, traditional storage solutions become complex to deploy, hard to manage, difficult to scale; therefore, a new approach to view the performance aspect is needed.