Calculating SAN return on investment

An analysis of the financial benefits of implementing storage area networks.

By Mark Teter

Storage area networks (SANs) can solve many IT problems. By allowing multiple I/O paths to redundant subsystems, SANs create a virtual storage grid within the IT infrastructure. The adaptive architecture improves both data availability and manageability-and holds great prom-ise for producing a more stress-free enterprise storage model.

SANs permit distributed servers to access large, consolidated storage resources for data-intensive applications. Shared storage pools can be accessed by multiple systems through multiple links, which translates into higher availability. Addi-tional benefits include lower storage management costs and ultimately better asset utilization, better storage investment protection, and the ability to prevent vendor "lock-in."

Though the benefits of a SAN are seemingly obvious, financial justification is of-ten required. What is the return on investment (ROI)? The answer depends on the problems that need to be solved. Are backups becoming unwieldy? Is contingency planning or business continuance a top priority? Is storage growth causing management headaches? A closer look at a SAN's financial impact provides a better view of potential investment returns.

Cost benefit analysis

Many IT areas can benefit from SAN technology. Most often, SANs provide IT departments with an improved structural design for future application platforms. Short-term benefits generally provide immediate cost savings to common IT problems. Whether the problem is a shrinking backup window, supporting greater storage capacity, or building cost-effective disaster recovery solutions, SANs can have a favorable financial impact. See Table 1 for a SAN benefit analysis.

Fig. 1: Host-attached storage involves many management issues.
Click here to enlarge image

Figure 1 illustrates a common IT problem-large amounts of host-attached storage. Typically, this type of storage involves data management, storage allocation, cost, availability, and data recovery issues. SANs can effectively deal with such management concerns.

Fig. 2: A SAN architecture centralizes management and control for better data management.
Click here to enlarge image

Figure 2, on the other hands, shows a SAN archi tecture in which storage re-sources are centrally managed and controlled for improved data man agement. Storage-and servers-can be added without disrupting data access, and storage capacity can be added and made available to all servers, applications, and departments.

With SAN architectures, large libraries can be accessed across the whole IT organization. Investments in existing tape libraries can be re-deployed, or "cascaded," in other areas of the computing environment that aren't yet part of the SAN implementation.

SAN alternatives

Various SAN "solutions" have been available from enterprise system/storage for several years. Known as "SAN in a can," these storage architectures provide multi-hosted data access, creating virtual storage networks within a single storage-array frame.

Click here to enlarge image

Though these approaches enable IT departments to perform sophisticated storage management operations without affecting host CPU cycles, they do have drawbacks. Most notably, they lock IT organizations into proprietary products, preventing administrators to choose "best-of-breed" products from multiple vendors. As a result, IT organizations ultimately pay a premium for these types of architectures.

Also deserving a quick look is network-attached storage (NAS). NAS is often seen as an alternative to SANs. NAS is simply a special-purpose storage server that provides file services (storage is accessed through file-system device drivers) to host computers. In general, NAS and SAN are complementary technologies; however, NAS does not provide all of the benefits of SAN. The major differences are:

  • SANs centralize data management, reducing total cost of ownership; NAS creates distributed management.
  • SANs create private networks for storage; NAS creates shared networks of storage.
  • SANs use storage protocols; NAS uses network protocols.
  • SANs provide remote storage access; NAS provides remote file access.

A look at the numbers

The cost estimates in Table 2 are calculated from the illustrated topology in Figure 2. The calculations provide a basis, or financial reference, for the ROI of a SAN implementation.

In most SAN ROI projections, the cost per MB of data is a significant calculation and should be adjusted to the chosen enterprise storage vendor(s)-the more fault tolerant the storage solution, the higher the cost per MB. Generally, this is money well spent. According to Strategic Research, roughly 30% of application downtime results from data storage failures. From a SAN perspective, cost per MB does not include the cost of the server that hosts the storage or the software, support, and maintenance costs associated with that server.

Another important consideration is management costs (see Table 2). The numbers are based on the fact that central storage pools can be managed more cost-effectively than distributed host-attached environments. Storage is easier to monitor and measure, and service demands are more easily predicted. Additionally, management tools and training can be leveraged across a broader base of systems. By IT Centrix accounts, management costs can be reduced by nearly 57% through the use of SANs. Though the calculation for these management costs varies from company to company and from project to project, the general principle applies.

SANs also significantly improve back-up times since data is moved at Fibre Channel speeds over dedicated storage networks, rather than at Ethernet speeds over shared networks. Most case studies indicate a 5X to 20X reduction in the amount of time required to perform a SAN versus LAN-based backup.

It is important to note that the purpose of a backup is to provide reliable data protection. Under time constraints, IT administrators may perform more incremental (block-level incremental or delta differentials) backups over longer periods of time.

Click here to enlarge image

  1. Centrally managed and controlled storage assures management consistency and reliable data-protection.
  2. Involves the cost of downtime in order to install additional storage resources (servers, along with their applications, and possibly their networks, must be brought down for storage expansion).
  3. Servers must have enough overall disk capacity available, ability to add additional devices (have not exceeded SCSI chain device limitation), and available I/O slots.
  4. Many tape library vendors have or are showing future product roadmaps with native FC interfaces.
  5. Costs are based on MSRP and should only be used for comparison purposes.
  6. Assume cost of downtime is $20,000/hour.
  7. Full-time equivalent.


Current infrastructure
3 enterprise-class UNIX servers ($150,000 ea.)
3 enterprise-class NT servers ($60,000 ea.)
Current IT support
.5 FTE required to support NT systems
.5 FTE required to support UNIX systems
Initial storage requirements
300 GB host-attached to each NT server
900 GB host-attached to each UNIX server
Projected storage growth
10% per month

Often, organizations turn off "write verify," skip or exclude application file systems, or even rely on "RAID protection" to provide data availability. Ultimately, without sufficient data protection, the organization pays a higher price for availability and recoverability in the event of data loss or corruption.

From an ROI perspective, faster backups translate into improved data recovery capabilities. With the backup window a non-issue, backups can be performed more often, more consistently, and more completely-all of which result in faster data recovery. This minimizes downtime associated with application data failures, which greatly contributes to the SAN ROI.

Final analysis roughly indicates a 12-month breakeven point ($60,400 savings). Over a longer period, the ROI looks even better. Obviously, each organization needs to determine the financial impact of specific SAN implementations. Typically, though, they show a 12-to-24-month payback, making the cost benefit analysis a straightforward financial decision. Associated costs and cost-avoidance calculations include:

  1. Initial acquisition costs
  2. Associated IT support costs
  3. Future storage requirement costs
  4. Data availability and recoverability costs.

In general, SANs preserve the storage and server IT investment over a longer period of time than distributed, host-attached storage models. Mileage varies with each organization and with each SAN rollout; however, the cost-saving benefits will be evident.

Mark Teter is director of enterprise solutions at Advanced Systems Group (www.virtual.com), an enterprise computing and storage consulting firm in Denver.



This article was originally published on April 01, 2000