Computing grids are collections of computing resources such as servers, networks and storage, that can be managed as single or unified entities. Behind this is the need to scale and support high-capacity and even high-performance environments across an enterprise to meet with and reflect business objectives.
Grid storage is a significant component of this development. As companies focus on their information assets, systems must be able to protect large volumes of information and maintain it in a secure fashion as data volumes explode. This will not only be databases, emails and file systems, and their associated archives, as governance and compliance issues will also need information, in voice and image formats, to be maintained and archived. What grids must be able to do is to dynamically allocate resources in a flexible and transparent fashion across a network, enable cost-effective management of the resources applying policies and procedures to protect and secure the information assets of an organisation.
Who is addressing grid storage?
Network Appliance is building on its success in storage networking to meet, firstly, the high capacity needs of film editors and geophysical organisations. The lessons learnt from building a storage grid with industry standard components that can be addressed with a global name, supporting a unified management environment, will be available to the commercial market.
IBM has been very active in the grid computing area as a whole. Harnessing such technologies as virtualisation, it is developing a distributed Storage Tank offering which will support a distributed file system that can be addressed as a single name space, with the data located across a distributed network. The company also has another initiative developing Collective Intelligent Bricks, storage "bricks" comprising a microprocessor, a small number of disk drives and network communications hardware. This will offer a data storage system that can scale to petabytes of storage, reducing the size, cost and power usage of the vast storage systems needed in the future.
HP announced its Reference Information Storage System (RISS) as its first solution implementing a grid storage architecture. Addressing the need to provide active email archiving, the system is based on a series of storage smart cells that can be added or configured to store, index, search or retrieve information. The system can be scaled using industry standard cell components, with future releases adding additional data management services.
What of the system components?
There are two key issues that are converging to drive the development of grid storage. The first is the utilisation of standard components such as server blades, disk drives and networks. These will harness the Wintel platforms, Linux and other developments that are driving the industry to commoditised components. However, the underlying architecture must be resilient so that any points of failure are automatically and transparently recovered.
The other key driver is managing storage resources across the enterprise, establishing a set of policies that need to be effective across an organisation and building a framework that can scale to support very high capacities of information stored locally or across an organisation. These services will include data protection such as mirroring, snapshots, backup and recovery and RAID. The management framework will add functional cells such as that for indexing, search and retrieval as within the HP RISS system, as well as additional services such as security, provisioning, costing and performance monitoring.
What is different from storage networking?
Consolidating storage resources across a storage network is the first stage for many organisations to improve their data protection practices, to enable a tighter management of all information resources and to reinforce policies and processes that will comply with regulatory or legal guidelines.
The management framework supporting storage networks will be reflected within the grids. While storage networks can grow enabling a flexible management infrastructure, grid storage will offer an integrated management framework which enables scaling to very high volumes with a cost-effective system platform. Grid storage will span locations, bridging system and software architectures, making it possible to simplify access to new or archive information based on service level commitments.