Published: 10 Nov 2006
How to better connect storage to the business
Storage managers can learn how to streamline processes by emulating manufacturing process efficiencies.
Not too long ago, I toured BMW's impressive manufacturing facility in Spartanburg, SC. For each built-to-order car, subassemblies travel separate lines where assorted components are installed. Country-specific features and various customer-requested options intermingle on these lines and, despite all the diversity, the subassemblies create the finished vehicle in a nearly flawless manner.
Manufacturers have made substantial investments to perfect their assembly, quality control and assurance processes. They've also incorporated efficiency methodologies such as just-in-time delivery. The BMW factory stocks remarkably little inventory, and suppliers are required to supply components within days or even hours.
At the core of the modern manufacturing process is the concept of the business supply chain. Each link adds value, with an interdependency among the functions that lets the output of each workflow feed into others, forming larger supply chains.
IT plays a role in virtually every business workflow. The primary point of connection between IT and the various workflow functions is the application. In many cases, applications have become the hub of a given process.
What is storage's relationship to the supply chain? As shown in the diagram "Storage and the business services supply chain" (at right), storage sits deep within the IT infrastructure. Its customers are internal groups and others that are externally focused on customers. While there's talk of aligning storage with business, storage typically doesn't interact directly with the actual business workflow but with the business apps critical to that workflow.
The weakest link?
Being near the bottom of the supply chain doesn't equate to being low in value. On the contrary, almost every function and workflow is dependent on storage services. But there can be a misalignment between storage and its ultimate customer at the top of the chain. At the upper end, the focus is on enabling a business workflow as efficiently as possible. Often, decisions are made that don't adequately consider the impact on data management and the storage infrastructure.
An area of particular weakness is often long-term data retention. We all know that some data needs to be retained for legal, regulatory and business reasons, but most organizations are grappling with how to identify, segregate, store and manage that data. The logical, and often only, place to enable data aging and management policy is within the app; however, few apps provide that mechanism, resulting in the problem of being overrun with inactive data on primary storage. Because the storage group is under pressure to control costs, it has to figure out how to address the problem. This can result in unintended, costly consequences and shortcomings in some supply-chain planning and design.
A supply chain is only as effective as its weakest link. Therefore, service metrics must be clearly defined and understood for each link in the chain. For the business service to be delivered effectively, each component supporting it must perform to required levels. There's little to be gained if one link substantially overdelivers other links.
Each link must understand its interdependencies and perform its responsibilities in proper sequence. Just as in manufacturing, where subassemblies must meet in the right sequences, storage depends on areas like networking and servers; it also needs to provide provisioning and data recovery services to support databases and applications.
Architecting a Service Oriented Infrastructure
As noted earlier, storage impacts many business workflows. To support a wide range of requirements, an effective service architecture must provide the following:
- Responsiveness to business workflow demands
- Tight integration with infrastructure and operations
- Flexibility, scalability and efficiency
- Adherence to standards wherever possible
At the technology infrastructure level, a service provider model lends itself to a similar concept called a Service Oriented Infrastructure (SOI). An SOI is based on the concept of "publishing" a set of services that can be "subscribed" to by those requiring the services. Implicit in the model is the concept of a supply chain. A service is provided to a functional area that can leverage that service's capabilities to provide a function to a higher level within the organization.
Let's look at the attributes of an SOI. The results of the service are visible to the requester, but not the details of how the service is provided. In storage, this means distinguishing between service attributes and the delivery method.
Processes play a major role in an SOI. External processes manage the interaction with customers, focus on customer satisfaction and demonstrate performance; internal processes handle the method of operation, and focus on efficiency and quality of service.
Because of the large number of consumers of SOI-provided services and the desire to leverage the most effective technology options available, adherence to standards is another key attribute. A proprietary solution doesn't fit with the demands for flexibility.
The range of technology components required to fully realize an SOI are evolving at differing rates. Within the IT infrastructure, the three basic components are networks, servers and storage. Networking can be considered the ubiquitous element upon which everything else depends. Bandwidth and reliability are enabling factors and, for the most part, networking technology has evolved to expand predictably on a demand basis. In the server realm, with the adoption of virtualization and tools to better manage virtualized environments, technology is maturing to a level needed to support SOI.
Storage lags servers and networks in its ability to offer an entirely resource-based set of services. One reason is that once a unit of data is moved or processed in a network or server it can be "forgotten"; with storage, data is persistent and must be managed for years. A services-based approach to storage needs to be multidimensional and encompass performance, availability, data protection, data movement and migration, and data retention.
In the manufacturing supply chain, the primary goal is to achieve a quicker time to market by eliminating nonvalue-add functions, cutting lead times and improving communications. The result has been high quality and lower costs. In IT, and particularly in storage, we can learn a lot by emulating these concepts.