It has been more than a year since IBM scrapped plans to introduce virtualization for its Shark Enterprise Storage Server and more than three since the announcement of its much-heralded Storage Tank platform, but it appears that IBM is now ready to deliver both.
The first area of IBM's focus is a Linux-based virtualization engine called Compass, which represents Big Blue's move toward virtualizing storage outside of the array. An approach that caused some confusion among end-users in January of 2001, when it was announced that the Shark storage array would not be equipped with virtualization technology within its chassis. IBM said Compass lets administrators view and access a common pool of storage on a network, and to increase storage utilization on existing storage servers.
"What they're really aiming at is taking virtualization that could be done at the array control level and moving that into the network," said John Webster, founder and senior analyst for the Data Mobility Group.
Webster believes IBM, like many other virtualization players, came to the conclusion that it was more advantageous to move virtualization out of the storage array.
The second piece of Big Blue's software puzzle is Storage Tank, a file system optimized for accessing, saving, sharing and managing files on storage networks.
And rounding-out the strategy is a new set of standards-based management tools that are centered on the common software
IBM is offering an olive branch to the rest of the storage industry in the form of free licensing of the Linux-based source code reference implementation of the application server software module that communicates with the Storage Tank metadata controller. The free code will give developers a chance to start prepping applications to run on StorageTank, and, IBM hopes, speed the implementation and adoption of its technology.
IBM's virtualization engine will be implemented on a cluster of IBM eServer xSeries servers running Linux and is designed to serve as a single point of control over disk storage capacity (block level management) within a storage network.
The Storage Tank metadata server will be implemented on an IBM eServer xSeries cluster running Linux. Storage Tank support for AIX, Solaris, HP/UX, Linux, and Windows 2000/XP platforms will also be provided. Since it uses IP networks for communication between application servers and the metadata server at the file level, and it supports SAN transports, Storage Tank is a form of SAN/NAS convergence, according to IBM.
Bruce Hillsberg, director of software strategy and technology for the IBM storage systems group said the ultimate vision of IBM's newly-formed storage software division is to autonomic capabilities in a standard, open way with the SNIA storage model.
Hillsberg said StorageTank is a common file system that is SAN-wide today and will be enterprise wide in the future. He added that StorageTank will allow for heterogeneous file serving and policy-based automation.
"We're working very hard to make sure it's open. The StorageTank clients and source code will be licensed at no cost, Hillsberg said. IBM will also publish its metadata server protocol.
The software push leaves one question: What will become of IBM's Tivoli storage arm?
"[IBM's] role is around the storage infrastructure and Tivoli focuses on enterprise-wide, heterogeneous management," Hillsberg said.
While StorageTank was born of development efforts within Tivoli, it seems Tivoli's future will be in the enterprise backup and recovery and storage resource management arenas.
"We think that by design are totally complimentary," said Hillsberg.
While specific dates and pricing are not yet available, IBM did say the virtualization engine and Storage Tank will be available in 2003.
Let us know what you think about the story, e-mail Kevin Komiega, assistant news editor
FOR MORE INFORMATION: