SRM Workflow and Automation - Storage Management

Computer Technology Review, Feb, 2003 by Mike Koclanes

Managing data storage is a continuing challenge for storage administrators. They must steadily improve availability, quality and performance of storage services in order to support their company's business objectives. Now their task is even more challenging as they must manage an ever-increasing amount of physical data in the face of flat headcount and shrinking resources. Consolidating to network storage, SAN, and NAS can improve operational efficiencies can significantly reduce this management burden, especially when as part of an active, integrated storage resource management (SRM) solution.

Effective SRM solutions are based on:

* A holistic data model that incorporates the logical and physical relationship of the storage elements from the application, through file systems, volume managers and databases, to the HBA, through the fabric, to the storage array and down to the spindle.

* Well-defined application centric policy and rules.

* Service-level agreement enablers.

* Service-level objective (SLO) executors.

* Well-defined managed element (devices, file systems, databases, volume managers) characterization, policy, and rules for provisioning and managing these resources.

* Best-practice workflows that enforce the policies, and rules that ensure SLO adherence within the known characteristics based on the holistic data model and the competing application service level objectives.

* Automation of the workflow steps to enable consistent per formance and operational efficiencies in analysis and provisioning of storage.

Best Practice Workflows

Storage management in a shared storage network environment requires storage management processes as well as the SAN infrastructure and device monitoring and provisioning capabilities. An integrated, active SRM solution will incorporate a change management process that utilizes best practice policies and rules through a workflow.

The necessity of the workflow is driven by the fact that there are multiple constituencies involved in the provisioning and management of a shared storage infrastructure:

* Storage consumers, usually the systems administrators/database administrators, of the line of business or application group

* Storage administrators

* Storage architects

* Storage and IT management

No longer are the storage resources under the exclusive control of the local server administrator. Application and line-of-business system administrators and database administrators request storage in this new paradigm based on their application or business needs. Service-level objectives for performance, availability, and recoverability for a given application or application type drive the provisioning process.

Industry best practices show that central storage architects should set up policies on how they want to deliver these classes of service in a standardized fashion through their storage infrastructure. The workflow system must therefore enable the architects to define their policies and rules and should provide a template of best practices out of the box. It must allow for the definition of service level objectives and classes of storage. It must allow for the line of business administrator to choose the class of storage and service level objectives requested and the amount of storage desired.

Furthermore, the workflow should allow the storage group to review the request and even simulate execution to see if the service level objectives can be met. The workflow should enforce best practice policies at each step of the provisioning process. Best practice policies for RAID levels, striping, LUN masking, LUN mapping, zoning, and performance tuning should be able to be enforced by the integrated workflow, policy engine, optimization logic, and the holistic data model. Tasks should be distributed from the workflow to the appropriate automation modules or administrators and complete audit trails must be maintained.

As the complexity, number of ports, arrays, and servers increase in the SAN, it is imperative that automated storage management be an integral part of the system. The workflow system must be able to embrace this complexity and still ensure the implementation of best practices. It should deal with failure in the workflow process due to unplanned obstacles in provisioning, changed states of managed elements failing components, misrouted or disconnected cables, etc. Rollback to a known must be a central part of an automated workflow environment.

In optimum circumstances, these best practices are depicted graphically in a workflow that safely defines individual steps or processes that can be performed either manually or automatically. Executing poorly defined workflows or scenarios will increase the rate at which issues, such as using incorrect worldwide names, or improperly configured ports occur in the organization. The old computer axiom "garbage-in, garbage-out" applies here as well.

Best-practice workflows must be automatically customized to the network storage infrastructure that is already deployed. There are many choices in a typical environment for striping, mirroring, and performance tuning. This varies based on the types of arrays, current RAID pools, fabric layout, port utilization, spindle utilization, volume managers, and replication software available. The workflow suggested should consider these choices in its suggested best practice definition of tasks to fulfill a provisioning request. Finally, best-practice workflows should have a detailed audit and timing capability so that it is clear what steps were taken to achieve a specific objective while assigning accountability to the business units that are being supported by IT.