Storage Virtualization

Digg This!

Page 1 of 2   next page »

The demand for storage will continue to grow. Endless amounts of data are being created driving greater storage capacity requirements and price improvements. That same data must be classified and moved into various tiers of storage to facilitate cost-effective implementations. Information lifecycle management (ILM) offers a set of practices and tools for managing the classification and movement of data in alignment with service-level and cost-of-ownership objectives. For ILM to deliver real end-user value, the storage infrastructure has to provide a foundation that can host the necessary tools and processes. Present day storage infrastructures fall short of providing this foundation because of inherent limitations that include decentralized management, disconnected SAN islands, and inefficient use of storage resources. Intelligent storage infrastructures address these limitations and provide the foundation for a successful deployment for ILM.

Phased Approach to ILM Implementation
Information lifecycle management offers a set of policies and practices that let users apply values and rules to business information. Information is classified at its source based on its business value and stored (or discarded) on a device matching its asset value. For ILM practices to be implemented in a real environment, a number of steps have to occur. The core requirement of the ILM strategy is that an enterprise must understand the relative value of its information and how that value changes over time. This understanding provides the ability to classify and store information and achieve the service-level objectives (SLAs) established.

From the deployment perspective, ILM is all about the classification and movement of data from one storage medium to the other based on its asset value. Initial implementations of ILM are being carried out with tools that exist today but for ILM for deliver its full benefits, the deployment has to be well planned and carried out in multiple phases. To enable a streamlined deployment, Storage Networking Industry Association's (SNIA) Data Management Forum (DMF) is suggesting a multi-phased approach as shown in Figure 1.

The first phase of ILM deployment is to deploy storage on the network and provide a centralized management scheme for the storage services. While deploying networked storage is a common practice in large enterprises, current infrastructures fall short of providing the centralized management of storage resources and services.

Storage Infrastructure Requirements
ILM requires the storage infrastructure to support classification and, more importantly, the movement of data from one storage medium to the other. The infrastructure should also enable storage resources to be allocated on-demand and support non-disruptive data migration to meet service-level objectives. To facilitate these ILM requirements, the storage network should support for the following:

• Multi-tiered storage - The infrastructure must support storage devices ranging from high-end arrays and tape libraries to low-cost storage disks. This enables the movement of data from one tier to another based on the business value of the information.
• Heterogeneous storage - The infrastructure must support storage devices from multiple vendors. Interoperability across vendors enables the migration of data from one vendor's storage device to another.
• Multiple protocols - The infrastructure must support connectivity across storage networks that have implemented protocols such as Fibre Channel and iSCSI.
• Multiple applications - The infrastructure must enable storage and data management applications to be implemented on the network.
• Efficient utilization of resources - The infrastructure must enable efficient use of resources by supporting dynamic resource allocation and reallocation.

Drawbacks of Present Day Infrastructures
Current infrastructures fall short of meeting these requirements because of inherent limitations. Storage arrays from different vendors don't interoperate, locking in customers to a single vendor. Large deployments of SAN infrastructures have led to multiple disconnected SAN islands forcing customers to deploy high-cost resources, tape libraries, for example, in multiple SANs resulting in the underutilization of expensive resources. In cases where SAN islands are connected by a simple switch, the resulting "merged SAN" gives rise to reliability concerns because of changes in network configurations and the limits of the infrastructure's scalability.

Though SMI-S, when implemented, will address the interoperability issues between management application and storage resources, as a management interface it doesn't address the drawbacks related to efficient resource utilization, seamless data migration, or protocol connectivity.

Intelligent Storage Networking
Intelligent storage networking addresses all the current limitations of the infrastructure and lays the foundation for a successful rollout of ILM. Intelligent storage networking involves installing and administering intelligent SAN platforms in the storage network to optimize the transfer of data between servers and storage elements and the transfer of data among storage elements.

Intelligent SAN platforms address the infrastructure requirements of ILM by enabling applications such as network-based virtualization, data movement, and data replication. Network-based virtualization and data management applications enable highly efficient storage resource utilization and enable the movement or replication of data in a manner that's transparent to the applications. Thus virtual volumes (storage) used by the applications can be moved transparently from expensive arrays to inexpensive storage based on their asset value, or replicated to a remote location to meet service-level objectives.

Page 1 of 2   next page »