Storage Systems - Projects - Mirage

Mirage

To achieve operational continuity, data center administrators are increasingly turning to automated decision making technology. Mirage provides data center administrators with tools that can automatically analyze their data storage environment and recommend changes they must make to enhance their performance, planning migrations, tiering and consolidating storage in response to storage growth or retirement while maintaining operational continuity.

Mirage solves the critical decision-making problems encountered while managing storage in large firms' data centers. The storage administrators in such data centers must manually evaluate complex storage area networks and determine the best places to layout their firm's data, a task that is solved by the automatic scenario analysis of Mirage. Mirage monitors a data center and recommends the most efficient changes to data placement so that the tight time constraints of demanding applications such as trading, analysis and data mining are satisfied and business continuity achieved. Given that there is no decision-making technology that addresses placement in data centers, Mirage is uniquely positioned to provide the solution.

The key research innovation in Mirage is the storage optimization software designed and implemented at IBM's Almaden Research Center. Mirage is a smart, non-disruptive storage optimization engine that lays out a process for optimal storage allocation and re-allocation using sophisticated statistical methods and detailed storage models. Mirage analyzes the arrays, switches and hosts to identify resource constraints and suggest time-tested solutions. Mirage tells the storage administrator: (a) how to allocate storage for anticipated application requests, and (b) how storage should be configured to optimize existing infrastructure. The inputs to Mirage are application profiles, client optimization policies, configuration and performance information. The output of Mirage is an analysis report with suggested storage allocation and re-organization points.

If you are working on similar areas and interested in collaborating with us, please send an email to Prasenjit Sarkar (psarkar@almaden.ibm.com), Lesley Mbogo (mbogo@us.ibm.com), Karan Gupta (guptaka@us.ibm.com), or Madhukar Korupolu (madhukar@us.ibm.com).

Jianyong Zhang, Prasenjit Sarkar, Anand Sivasubramaniam: Achieving Completion Time Guarantees in an Opportunistic Data Migration Scheme, SIGMETRICS Performance Evaluation Review 33(4), March 2006.

Kostas Magoutis, Prasenjit Sarkar, Gauri Shah: OASIS: Self-tuning storage for applications, 14th IEEE/NASA Conference on Mass Storage Systems and Technologies, May 2006.