The Business Case for Storage Networks [Electronic resources] نسخه متنی

Changing the Support Model Paradigm

As part of the process of moving from early adopter to early majority, storage networking consumers have much to consider during migration and implementation. In addition to varying degrees of standards in hardware Management Information Bases (MIBs) and software drivers, as well as varying degrees of maturity in management software, one of the most crucial issues is supportthe care and feeding of the storage networking solution after product selection and implementation.

Networking and IP Transport Model

Most companies already have some sort of networking or transport support model in place. As more and more storage networks require a long-range networking component (storage over IP or SONET/SDH), storage networking support might fall to the networking infrastructure team. There are some benefits to be reaped from this model. Online support structures and case queues might already be in place. In addition, IP networking knowledge and expertise can be utilized thereby increasing the quality of support. Economies of scale and scope can be achieved with this model, which can boost productivity in the long run.

The issue with this support model, however, is that the end products of networking and of storage are different. The product of the IP network is the IP networkthe delivery of IP bandwidth at a set level of service. The end product of the storage network is, of course, storage. Its delivery to the host and application is highly integrated at both the physical and the application layers.

This is not to say that networking support staff cannot be trained or the support structure modified to support applications above the switching layer, but more than likely, that investment is difficult to justify. The level of changes necessary to make joint storage and networking support teams work would likely diminish the capabilities of the current infrastructure support teams while being cost- and time-prohibitive.

This commingling of resources would be advantageous only in small-to medium-sized businesses where the number of IT support staff is limited. For larger environments, alternative support models should be considered.

Virtual Storage Team

Many organizations find the provisioning and performance tuning needs of storage-intensive applications and environments to be met satisfactorily by the system ownersthe systems administrators themselves. These individuals already have an intimate knowledge of the applications and the hosts, a wide body of corporate knowledge, a sense of ownership over the system, and, often, the ability to tune performance on-the-fly. Generally speaking, the system owners know the application stakeholders and the end users and have a long history providing quality service to all parties involved.

At a certain level, a systems administration team working only part-time on storage, or a virtual storage team, can meet or exceed expectations as far as ongoing storage support is concerned.

There are three caveats with this support model, however. The first of these is the risk of stovepiping. When the storage support model is a virtual team or comprised only of system owners, there are roadblocks to sharing knowledge, best practices, and expertise. Standards become hard to follow and systems tend to be managed differently, preventing any return to economies of scale. Individuals find it hard to manage systems that are not their own, and they therefore tend to get possessive over their own systems.

The second caveat to take into consideration here is the risk of inefficient storage management. In a virtual support model, storage is often not viewed as a capital asset, and as such, it is not uncommon to find a number of gigabytes stashed away to be doled out to applications at the last second. The hoarding of storage prevents any kind of meaningful capacity planning from taking place, and it drives up the TCO for the environment. It can increase the allocation efficiency in the short run, but in the long run it makes the costs of storage and storage support unmanageable.

The third and final caveat to be aware of with a virtual team is that a virtual team by nature is resource-constrained. Because resources are not fully dedicated to storage management, progress toward a storage vision will likely be impaired by the use of part-time resources.

Depending on the size of your storage environment, a virtual team of storage administrators might work well for your organization. As your firm's storage inventory approaches 500 TB to 1 PB, however, the ability to provide adequate storage support and service diminishes. It is at this point that storage support becomes a full-time job. Decision makers managing environments in excess of 500 TB should investigate creating a full-time storage support team.

Dedicated Storage Team

The concept of a storage team is popular in the distributed computing environment, and it was borrowed from the mainframe world in which it was often customary to have a team of storage-focused individuals dedicated to providing data storage expertise, capacity planning, and support to mission-critical applications. Until recently, many client-server environments had not reached a point where data storage needs overwhelmed a dedicated systems administration team. However, as shown in Chapter 1, "Industry Landscape: Storage Costs and Consumption," the rampant growth of data storage across all industries highlights the fact that many environments are now faced with the inability to adequately support and manage the storage filling the environment.

A dedicated storage support team has the benefit of common or shared knowledge and expertise, as well as a clear charter and a mission statement. It is critical for the success of a dedicated storage team to have executive sponsorship as well as a central repository of tools and experiential data designed to preserve and document repeatable tasks and build a basis for sharing best practices.

Ultimately, a dedicated storage team should be able to drive overall productivity in the environment. Projects or programs, such as storage recovery and storage consolidation, are extremely difficult to support without a dedicated storage support team.

Individuals assigned to a full-time storage team should be able to bring increasing returns to scale to the enterprise hosting arena as the amount of terabytes managed by one individual increases. Ultimately, a dedicated storage support team is able to manage far more terabytes than a virtual team or a systems administration team supporting storage on a parttime basis.

From an organizational standpoint, the creation of a storage support team can face some challenges because the systems administrators might feel that his role is diminished and that the environment is becoming a "black box."

This is a real issue. In the client-server environment, systems administrators can resent storage administrators doing their job. They can also present non-trivial hurdles for storage administrators in terms of both non-assistance as well as impedance (physically denying access to the host or politically undermining attempts to do the work).

Nonetheless, as enterprise storage environments increase in size, a dedicated storage team is likely to provide the best support. Outsourced personnel can augment a dedicated team as well, as need dictates.

Outsourced Professional Services

The storage service provider (SSP) market, which originally seemed promising from an investor's standpoint, quickly soured as customers and then venture capital both evacuated the market almost overnight. The entire service provider market seemed to fall victim of the chasm.

Although the SSP business model offered mission-critical storage services to smaller companies lacking capital resources, such as datacenters and large external storage frames, three issues prevented that model from taking flight in any serious capacity, as follows:

The first issue was that many companies felt uncomfortable storing mission-critical data, especially intellectual property, financial data, and other sensitive information outside of their own premises.

The second issue facing SSPs was that they often could not realize the break-even point with a pay-per-use model because firms with significant data storage requirements generally already owned their own storage and datacenters and vice-versa (those who did not have massive amounts of critical data did not need mission-critical datacenter storage).

The final issue with this model (and probably the most telling) was that, generally, the customers who signed up for SSP services were dot-coms with insignificant revenue streams incapable of supporting the SSPs who incurred severe expenses building excess capacity that was never sold.

From a storage support standpoint, customers today can choose to leverage a partially or fully outsourced model on the premises, or utilize the SSP for complete handoff, or engage a combination of the two solutions.

Outsourcing or outtasking (outsourcing only specific tasks) to a vendor can increase returns to scale, especially if the vendor's services are included as part of a larger contract and can be acquired at little or no extra cost. Outtasking specifically can increase returns to scale because the internal staff resources already assigned to projects can be released and reassigned to critical initiatives requiring more internal expertise (which also allows internal staff to grow, broadening their skill sets and increasing job satisfaction).

Migrating to a completely outsourced or SSP model can lead to some of the same support and control issues seen with the hand-off from system administrators to a dedicated storage team, primarily that the storage environment becomes a black box to the internal support staff. Typically, this solution is used only in a scenario where the firm is in a change-intensive transition period with bursts of growth that cannot be accurately modeled for forecasting purposes. Alternately, an outsourced model might succeed during a period of contraction, when declining revenues are likely to hamper the ability to provide best-in-breed support to applications and end users.

Chapter 8, "Cisco Systems, Inc."

None of the decision-makers I have spoken to, however, chose a fully outsourced support model due primarily to the criticality of data and the perceived inability of external staff to perform as well as internal resources.

Organizational changes take time to implement. If project deadlines are tight, it is prudent to avoid initiating large-scale, sweeping organizational changes prior