Abstract: Storage Area Network and Cloud Computing are the domains which deal with various storage related computations that lead to better storage management. Storage management techniques for homogeneous and heterogeneous systems aredifferent. Much heterogeneous system uses offloading of various logical parameters and also provides intermediates such as interface engines in case of luster computing to achieve better system performance. while using such system lead to problems such as offloading of resources, large compute time, synchronization required between server and storage devices, large waiting time required for I/O operations, offloading large amount of data and various computational function etc. there are various approaches have been proposed to solve as system base or resource base application. Active storages etc. however, upgrading the system performance often required in existing solutions. The system performance can be improved by increasing speedup, decreasing latency period, etc. To achieve this we have come with a comparative study of tradeoff between.
1) Offloading of data and
2) Offloading of compute function
This will lead in better decision making in order to achieve increased system performance. The tradeoff between these two deals with throughput, resource complexities, I/O time and bandwidth etc. The further advancements can be done in this field by studying more number of parameters and comparing them with respect to offloading of data and offloading of compute functions.
Keywords: Active storages, compute, homogeneous, heterogeneous, interface engine, luster computing, offloading
|
DOI:
10.17148/IJARCCE.2019.81121
