Conceptual Underpinnings of Storage Management Protocols

To understand management protocols, readers must first understand certain principles of management. Typically, several components work together to compose a management system. At the heart of the management system is a centralized management host called a management station. In some cases, the management station is actually a group of clustered hosts. The management station typically communicates with a software component, called an agent, on each device to be managed. The agent accesses hardware-based instrumentation to make management data (such as events, states, and values) available to the management station. A data model is required to ensure that management data is structured using a well defined format. A well defined protocol facilitates communication between the management station and agents. The types of actions that the management station may perform on the managed devices are determined by the capabilities of the agents, management station, communication protocol, and administrative policies. The administrative policies cover such things as authentication, authorization, data privacy, provisioning, reclamation, alerting, and event response.

Note

Some management stations communicate with agent-less devices. An agent-less device exposes management information via one or more mechanisms embedded in the system BIOS or integrated into the operating system.

Standards play an important role in the cost of management systems. In the absence of standards, each product vendor must develop a proprietary management agent. Each new product results in a new proprietary agent. That increases product development costs. Additionally, management station vendors must adopt multiple proprietary interfaces to manage heterogeneous devices with proprietary agents. That increases the number of lines of software code in the management station product. The increased code size often increases the occurrence of bugs, slows product performance, increases product development costs, slows product adaptation to accommodate new managed devices, and complicates code maintenance when changes are made to one or more existing agents. Standards address these challenges so that product development efforts can be focused on high-level management functionality instead of basic communication challenges. As a result, prices fall and innovation occurs more rapidly. Much of the innovation currently taking place in storage and network management seeks to automate provisioning and reclamation tasks.

Several categories of storage-related management applications exist. One such category is called Storage Resource Management (SRM) that provides the ability to discover, inventory, and monitor disk and tape resources. Some SRM products support visualization of the relationship between each host and its allocated storage resources. Many SRM vendors are actively augmenting their products to include policy-based, automated provisioning and reclamation. SRM applications are sold separately from the storage resources being managed. SAN management is another category. SAN management applications are often called fabric managers. SAN management provides the ability to discover, inventory, monitor, visualize, provision, and reclaim storage network resources. Most FC switch vendors bundle a SAN management application with each FC switch at no additional charge. However, advanced functionality is often licensed or sold as a separate SAN management application. Note that the line between SRM and SAN management is blurring as convergence takes place in the storage management market.

Data management is another category. Data management is sometimes called Hierarchical Storage Management (HSM). HSM should not be confused with Information Lifecycle Management (ILM). HSM provides policy-based, automated migration of data from one type of storage resource to another. HSM policies are usually based on frequency or recency of data access. The goal of HSM is to leverage less-expensive storage technologies to store data that is used infrequently. HSM products have existed for decades. They originated in mainframe environments. The concept of ILM recently evolved from HSM. ILM performs essentially the same function as HSM, but ILM migration policies are based on the business value of data rather than solely on frequency and recency of use. The word information implies knowledge of the business value of the data. In other words, all data must be classified by the ILM application. The phrase tiered storage is often used in conjunction with ILM. ILM applications migrate data between tiers of storage. Each storage tier provides a unique level of performance and reliability. Thus, each storage tier has a unique cost basis. The concept of tiered storage originally derives from the HSM context. However, storage tiers were typically defined as different types of storage media (disk, tape, and optical) in the HSM context. In the ILM context, storage tiers are typically defined as different levels of functionality or business value. For example, high-performance SCSI disk might be used as tier one, while low-performance ATA disk is used as tier two. The underlying premise of all ILM applications is that businesses are willing and able to classify their data accurately so that ILM policies can be executed against the data. In practice, that premise is not entirely sound, and the adoption rate for ILM applications is likely to reflect the challenge of data classification until a practical and efficient data classification methodology is developed.