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

Table 3-4. Revised Goodrich TCO

Table 3-5. Goodrich Savings by Installing a SAN to Increase Utilization

Item

Cost

Note

Keep in mind that HBAs and training are treated as expenses and are therefore not depreciated.

In terms of ROI, the results from this analysis are encouraging. The sum of all returns is the increased storage available for use, which with the increase in utilization from 50 percent to 75 percent, equals 1.25 terabytes. If you value the reclaimed storage at the original $0.12 per MB acquisition cost, you see a respectable 41.51 percent ROI for the installation of the SAN.

ROI = (Sum of all Returns Sum of all Investments) / (Sum of all Investments)

ROI = ($150,000-$106,000)/$106,000

ROI = 41.51 percent

Increasing allocation efficiency decreases the frequency of storage purchases, increases the storage yield, and decreases the cost of poor quality and the TCO.

Let's take another look at the numbers using NPV. Let's assume that Goodrich still has a required rate of return of 10 percent. The cash flow for Year 0 is $106,000, and the following year, it is $150,000. The NPV for this project over a one-year horizon is calculated as follows:

NPV = $106,000 + ($150,000 / 1.1)

NPV = $30,363.64

The NPV for this project is positive, which is another indication that Goodrich should implement the SAN. Even if Goodrich is a little more conservative and decides to stretch the returns out over two years, knowing that they will not be able to efficiently utilize all of the storage from the increased allocation, the NPV for this project is still positive:

NPV = -$106,000 + ($75,000 / 1.1) + ($75,000 / 1.1²)

NPV = $24,165.29

In fact, even if Goodrich decides to carry the returns out over three years (the average useful life of the products in the environment), the NPV is still positive!

EVA analysis shows us the economic value created by the firm when migrating from DAS to SAN. If you use the same returns for the first year, $150,000, and a cost-of-capital of 10 percent, EVA analysis reveals a similarly positive message:

EVA = $150,000 ($106,000 * 0.10)

EVA = $139,400

The EVA-adjusted ROI shows the true ROI after including the cost of capital:

EVA-Adjusted ROI = ($139,400 $106,000) / $106,000

EVA-Adjusted ROI = 31.51 percent

The DAS-to-SAN migration should create value for the firm, and it should be started as soon as possible.

As the project matures, the TCO for the new SAN environment needs to include an adjustment for any other associated costs, such as software purchased to manage the environment or any additional headcount added to support the new hardware. Although it is often overlooked, post-project analysis needs to be completed to understand the true impact of the changes to the environment, to measure the delivery of the service against its service level agreement (SLA), and to update the TCO for the environment.

Intangible benefits are difficult to quantify, but this does not mean they cannot be counted in the financial analysis. Depending on the firm's strategy, intangible benefits, such as increased business continuance capabilities (through the implementation of FCIP or another type of SAN extension), can be numerically weighted to provide some type of quantitative value and to distinguish them from alternative projects, which might not help execute a firm's strategy.

For example, if one of the firm's top strategic goals is to increase data security and availability, a project that includes long-distance replication over SAN extensions as part of a disaster recovery plan should be weighted heavier than a disk migration strategy that does not offer replication. The business stakeholders need to be able to assign a dollar value to their business continuance requirements (in terms of cost avoidance measured in dollars per hour downtime), and then assign the appropriate weight to the projects in the project pool. The projects with the highest weighting and the projects with positive NPV, ROI, and EVA are then selected for funding and execution.

Storage Consolidation Value Case

Consolidation offers tangible benefits that can be achieved in a relatively short period of time. In particular, disk consolidation projects can:

Decrease annual maintenance bills

Decrease facility expenses

Decrease points of management

Improve operational efficiencies

A few years have passed, and Goodrich now has in production and on its books 80 external storage frames, all of which are configured as DAS and are only 60 percent allocated. These 80 frames (comprised of varying sizes of disk from 9 GB to 36 GB) have been fully depreciated; however, they carry monthly maintenance charges of roughly $6000 per month per frame. Goodrich pays a total of $5,760,000 per year in support for these frames.

In terms of TCO, the annual datacenter utilization cost for 80 storage frames is $1,200,000 (80 * 1.5 tiles * $10,000 per tile). Concurrently, the Goodrich IT team is working with the finance department to determine if they can build another datacenter because they have reached capacity in both datacenters and would like to build a backup datacenter. Original estimates indicate that a new backup datacenter would cost Goodrich roughly $4,000,000.

Goodrich is faced with a prime consolidation opportunity here. With larger and faster disks available on the market and with lower purchase prices for storage, Goodrich, using the highest capacity drives available, has the chance to consolidate from 80 frames down to eight.

The acquisition cost of the eight new frames includes three years of hardware and software maintenance and comes to a total of $3,600,000 million. The sum of all returns is the cost avoidance of the maintenance bill$5,600,000plus the cost reduction in utilities$1,008,000 (72 * 1.5 * $10,000).

The ROI for this project is calculated as follows:

ROI = (Sum of all Returns - Sum of all Investments) / (Sum of all Investments)

ROI = ($6,608,000 - $3,600,000)/$3,600,000.

ROI = 83.56 percent

The NPV for this project is calculated just as the other examples, but takes the returns out over all three years:

NPV = -$3,600,000 + ($6,608,000 / 1.1) + ($6,608,000 / 1.1²) + ($6,608,000 / 1.1³)

NPV = $12,833,117.96

EVA analysis shows the value of the returns for the first year of the project after accounting for the cost of capital:

EVA = $6,608,000 - ($3,600,000 * 0.10)

EVA = $6,248,000

The EVA-adjusted ROI for the first year is calculated as follows:

EVA-Adjusted ROI = ($6,248,000.00 $3,600,000.00) / $3,600,000.00

EVA-Adjusted ROI = 73.56 percent

Note

On the surface, it appears that Goodrich is deferring maintenance costs indefinitely, and if they choose to consolidate every three years, this would in fact be true. However, Goodrich is not taking advantage of the system; they are merely taking advantage of the new technologies available and decreasing the overhead associated with storage assets by migrating to best-in-breed technologies.

With the creation of a storage network in tandem with the consolidation, Goodrich has the opportunity to increase utilization and lower TCO. The additional costs of SAN hardware needs to be analyzed, but those costs should be completely offset by the increase in utilization and the decrease in the TCO per MB.

Consolidation at both the hardware and the application layer provides ample opportunities to reduce points of management and gain economies of scale. Consolidation at the application level trickles down to the physical layer and releases compute cycles, application licenses, and labor resources from the upkeep and maintenance of applications.

From a storage perspective, driving consolidation from the ground up is a great way to raise awareness of the issues previously discussed (cost-of-capital and TCO) and is a quick way to gain momentum for application and server consolidation efforts. When the staff starts to feel relief from managing fewer components (fewer components mean fewer hardware replacements and fewer outages) and experiences the other benefits of consolidation, other consolidation efforts become an easy sell.

Consolidation, after SAN migration, is the most important storage strategy a company can have.

DAS-to-NAS Migration Value Case

The Goodrich external web site and intranet currently reside on older, external RAID arrays that are configured as direct-attached storage with no mirroring or failover capability. The data in these environments is business-critical and read frequently, but written to infrequently, thus performance is of little concern. As part of the TCO study, the Goodrich IT department found that the TCO for the external storage frames in this environment was significantly higher than they expected. The two frames in this environment have been fully depreciated. The teams developing applications for the web site want multiple copies of data to test against. After discussing various vendor proposals with the team members, IT management has determined that this environment is a good candidate for migration to a NAS solution.

The acquisition cost for a NAS solution with failover capability that matches the requirements of this environment is $0.10 per MB, or $250,000.00, which includes hardware and software maintenance costs and redundant hardware for failover capability.

The sum of all returns for this migration (exclusive of a weighted value for the added functionality of failover capability) is the total cost avoidance for the maintenance on the two older frames of $144,000 per year (2 * $6000 per month * 12 months) or $432,000 over three years.

Note

In addition to the decreased acquisitions costs, the benefits from a DAS-to-NAS migration are primarily intangible and difficult to quantify: the flexibility of shared storage, the ease of use and administration, and the economies of scale, which stem from the use of the IP network.

In this case, the maintenance reduction provides the majority of the net benefits of the project.

The scale of this project is such that the ROI becomes positive in the second year. The ROI can be calculated as follows:

ROI = (Sum of all Returns) (Sum of all Investments) / (Sum of all Investments)

ROI = ($288,000 - $250,000) / $250,000

ROI = 15.20 percent

In the third year, the ROI becomes significant.

ROI = (Sum of all Returns) - (Sum of all Investments) / (Sum of all Investments)

ROI = ($432,000 - $250,000) / $250,000

ROI = 72.80 percent

The NPV for the DAS-to-NAS migration would be:

NPV = -$250,000 + ($144,000 / 1.1) + ($144,000 / 1.1²) + ($144,000 / 1.1³)

NPV = $108,106.69

EVA analysis shows us the true economic value of the project:

EVA = Net Benefits (Capital * Cost of Capital)

EVA = $432,000 ($250,000 * .10)

EVA = $407,000

The EVA-adjusted ROI is:

ROI = ($407,000 - $250,000) / $250,000

ROI = 62.80 percent

Network-attached storage is an excellent solution for applications for which availability and cost requirements outweigh the need for high performance. NAS hardware typically has a lower purchase price than higher-tiered solutions, which obviously helps to lower the TCO. Other NAS features also lower the storage TCO. Snapshot technology helps to ease labor costs associated with backups to tape and simplifies the process of providing shared storage to other groups that require access to the same data (as when multiple development environments require separate copies of the same code tree).

With IP networks, as with storage, a fundamental piece of TCO is utilization. Underutilized IP networks have a higher TCO than networks that are optimally utilized. With some planning, excess IP capacity can be successfully allocated for NAS storage requirements. Because NAS solutions do not require a buildup of FC capacity and can utilize resources already in place (network cards on the host; cabling, routers, and switches on the network), they are inherently more cost-efficient.

Migration from DAS to NAS, as application requirements dictate, can increase operational efficiencies and lower the overall storage TCO. Coupled with a consolidation effort, ROI numbers for NAS migrations should approach triple digits.

iSCSI Implementation Value Case

Similar to a NAS storage implementation, an iSCSI solution utilizes preexisting IP infrastructure (hardware, software management tools, and often networking expertise), with the added benefit of requiring no additional FC host-bus adapters for the hosts or additional Fibre Channel switch infrastructure required for a SAN. For applications with less stringent performance requirements, iSCSI can provide a low-cost, easy-to-use storage solution. With the added benefit of being a routable protocol, iSCSI enables the host and the storage to be located great distances apart.

For hosts requiring increased data throughput, TCP offload engines (also known as TCP offload engines [TOE] cards) can be used to relieve some of the processing burden from the host's CPU and from the primary network interface. Even with the added expense of TOE cards, an iSCSI solution can be significantly less expensive than a FC SAN solution. Although prices for FC HBAs are falling, FC HBAs are generally twice the cost of a TOE card and four to five times more costly than a standard Ethernet card.

Note

The performance of iSCSI software drivers has increased over time, which means that TOE cards are often not required for many installations.

Goodrich has an excellent opportunity to utilize iSCSI transport in the following example. Goodrich has a number of Wintel platform hosts dedicated to sharing user home directories in its primary datacenter. This environment uses a terabyte of direct-attached storage, but it has reached its full capacity. In its secondary datacenter, Goodrich has an additional terabyte of storage that is only 25 percent utilized. The acquisition cost for one MB of storage is $0.08.

To provide the additional storage for this environment, Goodrich can do one of the following:

Spend $175,000 for additional direct-attached storage to be hosted in the primary data center

Spend $40,000 for two 16-port multiprotocol switches that utilize iSCSI to attach to the remote storage

The Goodrich team decides to invest in an iSCSI solution because they already have an extensive IP network with extra capacity for growth and a staff that is familiar with IP networking.

Note

The additional port capacity on the IP network required for iSCSI in this example is a minimal cost that does not diminish the financial impact. It has been excluded from the total cost.

The ROI for this investment can be calculated using the cost avoidance of the $175,000, or, more conservatively, using the value of the 750 GB of storage in the secondary datacenter at $0.08 per MB or $60,000.00 (750,000 * 0.08). The total investment is $40,000 for the multiprotocol switches:

ROI = ($60,000 - $40,000) / $40,000

ROI = 50.00 percent

The NPV for the iSCSI migration is calculated as follows:

NPV = -$40,000 + ($60,000 / 1.1)

NPV = $14,545

EVA analysis shows us the true economic value of reclaiming the additional storage:

EVA = Net Benefits - (Capital * Cost of Capital)

EVA = $60,000 - ($40,000 * .10)

EVA = $56,000

The EVA-adjusted ROI is shown as:

ROI = ($56,000 - $40,000) / $40,000

ROI = 40.00 percent