The case for application-centric SAM

Storage area management software may be the key to unlocking the true benefits of storage area networks (SANs).


Companies are adopting storage area networks (SANs) to increase efficiency in managing and exploiting data and ensure that applications have continuous access to data. So far, SAN adoption has focused on installing the SAN infrastructure and migrating data from direct-attached storage to SAN-attached storage. These initial steps have yielded tangible benefits but have not allowed businesses to realize the full potential of SANs. A key missing element is comprehensive storage network management.

Storage industry analysts believe management has become a more important part of SANs than the underlying network infrastructure. Gartner Group analysts recently stated that "management software and its related issues should be the focus of SAN projects." Gartner has defined a new concept-storage area management (SAM)-to realign thinking in this direction.

Gartner defines SAM as "the centralized management of resources and data across storage domains, providing shared storage services to groups of servers and their applications." Implementing application-centric SAM may allow companies to fully realize the key benefits of SANs: improved operational efficiency and application service-level management.

First-generation SAN management software today focuses on SAN device discovery, topology rendering, and device-based management. These solutions are an improvement over using a collection of individual vendor-specific tools, but they lack the functionality and architecture to implement the automation and intelligence required for full SAN management. Businesses run on applications, not devices, and first-generation SAN management products do not manage from an application perspective. New approaches that provide intelligence and automation to make scaling SANs easier, more efficient, and foolproof will be required.

The logical path between an application and its data storage consists of the physical connections between the application and data, the devices comprising the connections, and the logical settings of the devices.
Click here to enlarge image

Storage area management provides a framework for evaluating functionality of next- generation SAN management. It dictates that solutions move beyond device-based management and create higher-level abstractions that hide the complexity of the underlying storage network infrastructure. This software should, at a minimum, manage the relationships between applications and their data storage, generate application-oriented views of globally distributed SAN infrastructure, offer policy-based automation of repetitive administrative tasks, and provide application-centric service-level management.

A SAM solution must perform the device discovery and topology rendering provided by first-generation tools and use this to establish a foundation for higher-level abstractions and relationships. One essential relationship is the logical path between an application and its data storage (see diagram). This path should consist of the physical connections between an application and its data, the devices comprising the connections, and the logical settings of these devices. This application-to-data path is, in many ways, analogous to the path established in a Virtual Private Network (VPN). A VPN establishes a secure end-to-end connection over a public network infrastructure, without users needing to know about the intervening devices and their connections. An application-to-data path should establish this same secure end-to-end connection between an application and its data storage over a shared SAN, abstracting the complexity and technologies of the intervening devices and their physical connections. This logical path should be established without placing any hardware and software in the data path, thus avoiding any potential performance or availability penalties.

Creating a logical relationship between applications and their data storage provides the foundation for higher-level abstractions. Next-generation management software will provide an application-centric view of the SAN infrastructure. This view should provide a real-time indication of application status; storage infrastructure problems and configuration changes should instantly and automatically be correlated to their impact on applications. An application view should also display the SAN infrastructure associated with each application-key information for usage chargeback, inventory management, and capacity planning. Ideally, management software should abstract the application view to a line- of-business (LOB) or departmental view. These views should span multiple geographically distributed SANs, allowing administrators to manage a business's global storage infrastructure from a single location.

Establishing logical application-to-data relationships and providing higher-level abstractions will allow businesses deploying SANs to improve operational efficiency and increase application availability. Today, administrators must manage a collection of devices and manually attempt to correlate device settings and events to their impact on applications-a labor-intensive, time-consuming, and error-prone process. With next-generation approaches, administrators will manage and monitor the relationships between applications and their data, and intelligent software will transparently manage and monitor the individual devices.

Automating storage provisioning

Next-generation SAN management software must automate storage provisioning into a single-step operation. Adding storage to new or existing applications, or re-deploying existing storage to other applications, are complex and time-consuming tasks today. Administrators must have a high degree of training and specialized knowledge and perform a large number of difficult and error-prone manual operations. In a next-generation solution, a non-expert should be provided with an intuitive user interface to choose an application and automatically provision storage to it.

The administrator would only have to specify a service-level policy and an amount of storage, and the policy would contain the rules and best practices that the management software follows for provisioning. The software would locate and select the optimal storage paths and storage network devices to meet the policy and automatically configure the devices-switches, host bus adapters, and storage arrays-to provision storage to the application.

As part of the provisioning process, the software could automatically coordinate and configure end-to-end application data security by using all available device-level security features. Config uring security for a storage network is so complex and time-consuming that it is often only partially implemented. In the rare instances where security is fully implemented, it is highly unusual for security settings to be monitored and verified on an ongoing basis. In fact, one erroneously-or maliciously-changed setting may lead to data corruption, denial of access, or unauthorized data access. To ensure ongoing application data security, SAN management software must continuously monitor all devices in the storage network. Any changed setting should be quickly detected and corrective action taken. The type of corrective action should be policy-based and range from notification to automatic device reconfiguration.

Automated storage provisioning and security management will allow businesses to better manage their storage growth, maximize security, and reduce storage network management costs. General operator-class employees will be able to configure and operate storage networks, freeing up storage experts to focus on more strategic tasks. These advanced features can also provide for seamless integration of new devices and technologies into the storage network, and operators will not need to learn new devices or technology specifics. Fewer people, with less training, will be able to operate larger, increasingly complex, and geographically distributed SANs.

Application-based service-level management is a third component of a SAM approach. Not all applications are created equal, and heterogeneous SANs allow applications servers to access multiple types and classes of SAN infrastructures. Mission-critical applications may require a "no-nines" policy of zero downtime, while other applications may safely tolerate 99.999% uptime or less. SAN management must provide service-level management by allowing businesses to set application performance and availability levels by making it simple to measure and track these differentiated service levels and by providing key data required to cost effectively right-size the storage network infrastructure.

SAN management software should automate both the implementation and ongoing monitoring of differentiated service levels.

Today, implementing differentiated service levels to applications and monitoring to ensure service levels are being met are becomingly increasingly difficult. The process relies on the analysis and judgment of an expert who must understand every device in the storage network.

In a next-generation solution, administrators should be able to create service-level policies that define applications' service-level objectives. Policy should at least specify performance and availability requirements. Operators will be able to select service-level policy as part of the storage provisioning process. After the storage is provisioned, the management solution should monitor applications' achieved service levels and alert appropriate personnel if desired service levels are not being achieved.

A SAM solution would also provide key data to allow businesses to make informed and intelligent purchase and deployment decisions. A key component of service-level management is achieving desired application service levels efficiently and cost-effectively.

Since it is so difficult to implement and maintain effective service-level management with current tools, and the consequences can be severe if service levels are not achieved, many organizations overbuy-and overspend-on their storage networks. Next-generation management solutions will track service-level parameters for all devices and connections in a storage network and log this information to a persistent database. Administrators will be able to use this information to determine what service-level devices and configurations actually achieve over time and make intelligent decisions when configuring and growing storage networks.

John Mallory is a senior solutions engineer at InterSAN (www.intersan.net) in Scotts Valley, CA.

Bulk Reprints

Do you need bulk reprints of an InfoStor article?
If so, contact
Dick Arzivian
Tel: (603) 891-9315
Fax: (603) 891-9297

This article was originally published on October 01, 2001