Abstract: The next generation of storage systems will use new NVMe based devices to store huge amounts of data at a very high price. So, storage users will need to use the new flash media very efficiently to ensure a good return on investment. The efficiency of data reduction is also critical for new generation Cloud Storage as the amount of data explodes impacted by huge data sets used by AI systems. The increase in efficiency can be obtained in different ways but mostly by reducing the size of the data written to the storage arrays using novel deduplication and compression methods. But increased efficiency comes at a price of reduced IO performance of the array due to CPU used for data reduction. So, although the new information theory research defined very efficient data reduction methods, most of these methods are targeting archive and backup storage that are less sensitive to IO performance and IO latency than storage systems that perform data reduction in real time or in-line. But we found out that there are ways to use similar techniques, used for backup storage, but optimize them for IOPS and latency performance and achieve a balance between the data reduction and performance of the arrays.
Keywords: cloud storage, data reduction, latency and performance.
Title: Digital Entropy on Improving Data Reduction for Cloud Storage
Author: Dr. Muchelule Yusuf Wanjala, Jacob Neyole Misiko, Shaaban Bakari
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Research Publish Journals
