THE DAWN OF CLUSTERED STORAGE

Storage technology has made significant advancements in the recent past, as networked storage, sophisticated storage devices and storage networks become are now integral to IT infrastructure of many businesses. Partly, the advancements have been driven by a sea change in the way business is done in the information era, and partly, by the ever-increasing need to store more and more data by certain 'New Economy', 24x7 businesses like Yahoo! and Google that thrive purely on the Internet. Gone are the days of gigabytes, enterprises now talk, err, store in petabytes and still look for larger capacities and better ways of managing data. It's at this juncture that clustered storage bids for a place in high-end enterprise storage management.

What is Clustered Storage?

Clustered storage originates from the concept of combining an array of storage servers that form a redundant ring of storage devices. Storage solution vendors have created software and hardware devices that integrate disparate file systems into one file system with one name space. These devices and software significantly improve users ability to access data and share it with others regardless of the media or host computer on which it resides. And the technology that seamlessly manages the entire system is known as clustered and Storage Area Network (SAN) file systems.

A SAN is a high-speed, dedicated network infrastructure of shared multi-host storage devices. The network functions in such a way as to make all storage devices available to servers on a LAN or WAN. SANs are being increasingly used for integrated data management and efficient data sharing in heterogeneous server environments. Usually, all storage devices in a SAN are gigabit-level and offer high system availability, extensive fault tolerance, and low cost of ownership. SANs, coupled with clustered storage technology offers the following advantages over legacy distributed storage systems:

• Speedier operations: By clustering systems and sharing applications and data, clustered storage technology performs tasks quicker than individual machines. In clustered storage, data need not to be copied or duplicated from one file system to another.
• Easier management: As one file system needs attention than a file system for each storage device or host computer.
• Uninterrupted data flow and higher uptime. In the event of failure of one of the servers, the other (redundant server) transparently and automatically assumes all server-processing operations.
• More space for files and file systems.
• Concurrent access to all files located on the storage devices on the network.

Further, the modular architecture of clustered storage allows for on-demand scaling of capacity, performance and availability, and dynamic matching of storage capacity to changing business requirements. The resulting savings and efficiencies in operational costs and capital investment significantly improve the bottom line and accelerated ROI.

Who Offers Clustered Storage?

One of the earliest examples of clustered storage application is HP's Tru64 cluster file system used in TruCluster systems. Recent offerings are from Cluster File Systems, Oracle, Red Hat, Panasas and Spinnaker Networks, and others. Red Hat offers the clustered Global File System and Network Appliance's SpinCluster software clusters network-attached storage (NAS) and SAN storage. Oracle uses its Cluster File System on the company's Real Application Clusters (Oracle 9i RAC).

Cluster File Systems' Lustre™ File System is designed to endure demands of world's largest high-performance computer clusters. The object-based storage architecture redefines scalability and provides radical I/O and metadata throughput. Lustre currently scales to thousands of nodes and hundreds of terabytes of storage and supports support tens of thousands of nodes, trillions of files, and petabytes of data.

Panasas is the pioneering leader in object-based storage clustering for scalable Linux clusters. The company's ActiveScale Storage Cluster is a premier storage system for scalable Linux clusters. Built on an object-based storage clustering architecture, the Panasas Storage Cluster offers exceptional scaling in capacity and performance and ease of management to a virtually boundless storage system.

Another SAN vendor, Xiotech Corp. has also released a clustered storage solution called the Magnitude 3D which is designed to automatically respond to outages and ensure continuous data flow by shifting data to another server in the likelihood of a failure.

But Then…

The very sophistication that the clustered storage technology is built on hinders its widespread adoption. Clustered file systems need good, knowledgeable and experienced set of technical people to architect them. And as the technology is still in its nascent stage, it requires exceptional minds with sound theoretical knowledge of file systems to get clustered systems up and running. Otherwise, users (companies) have to go with a vendor-provided solution and depend on their personnel for support. Another stumbling block could be the possibility of choked networks when too many clustered systems exchange data.

Nevertheless, clustered storage has already charted its course in the high-end data storage space, and is set to make it big among the enterprises wanting to manage mega data.

Introduction  |  Contents   |  Top