Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage...

8
Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System

Transcript of Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage...

Page 1: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

Performers: Yehuda KriefAssaf Cohen

Etay Goldshtein

Academic Advisor:

Dr’ Eitan Bachmat

Storage Management Tools for a Storage System

Page 2: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

• Companies that have a huge amount of data.

• When a user wants to get data from the Storage System, he will wait a long amount of time to get the data that he desires, and it may cause him to lose money, deals, etc…:- For example, in the NASDAQ stock market computers

compete between themselves to buy and sell stocks.

The company which has a faster Storage System than their competitors can get the requested information first, and earn more money.

Problem definition

Page 3: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

Storage System Structure

Host A

Host A

Network

Storage System

Page 4: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

System Storage Structure

Host A

Host A

Network

Storage System

Is it in cache?NO

Page 5: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

Storage Systems have a huge amount of data to store and manage (in many cases, more then 100 TB).

Storage Systems use DRAM memories as cache. DRAM memories are very fast, but they are also very

expensive. Storage Systems have a small capacity of cache

(usually 16 Giga Bytes). Cache capacity is very small compared to the disk drive

capacity that the storage system has. Cache misses lead to poor performance.

Current situation:

Page 6: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

Research on the extension of the Memory-Cache,of the Storage System, with flash disks (size of 150GB each ).

Tracking the trace of the Storage System:- amount of hit ratio before and after adding the Flash memory as extension of the cache.- statistical measures :Kb read/write – in time intervals.

Integrate a clustering algorithm with the trace: - cluster the Storage System : which will map the LUNs of the Storage System.- helps organize the caching algorithm.

Develop a simulator which the IT manager will use to examine and organize his/her Storage System.

Project Requirements

Page 7: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

We will implement the UPGMA clustering algorithm with the use of C++ with threads.

Build a simulator that will predict and examine the benefits of adding Flash into the Memory-Cache.

Investigate the behavior of “local”/”global” caching algorithms for the implementation of the Flash.

Method Of Solution

Page 8: Performers : Yehuda Krief Assaf Cohen Etay Goldshtein Academic Advisor : Dr’ Eitan Bachmat Storage Management Tools for a Storage System.

System Storage Structure

Host A

Host A

Network

Storage System