Leverage existing assets in SANs
The storage industry has gone to great extents to promote the concept and benefits of storage area networks (SANs), but it has done little to show end users, administrators, and systems architects how to deploy the technology without abandoning existing client/server assets and infrastructures.
By Mike Piltoff
This emphasis has left key decision-makers with the impression that the SAN paradigm--and all of its benefits--can only be realized by "ripping and replacing" or "fork-lift upgrading" existing enterprise storage infrastructure.
If SANs are going to go beyond the "pie in the sky" discussion stage and become the de facto standard enterprise storage architectures for the new millenium, this is not a realistic business model.
This article explores issues in designing and deploying SAN architectures with both new and existing storage assets; however, emphasis is placed on leveraging current storage infrastructures. Such SANs feature all the inherent benefits of the "anything to anywhere" storage paradigm, while using installed connectivity and storage peripherals where appropriate. The discussion includes return on investment and total cost of ownership issues.
The basic concepts behind SANs are not new, but many of the smaller players are relative newcomers. IBM has used its ESCON technology in SAN-type architectures since the late 1970s and its SSA technology since the late 1980s to provide many of the promises of SANs (e.g., anything-to-everything connectivity, disk pooling, etc).
The biggest difference between old and new solutions is market acceptance/availability. Early solutions were based on proprietary technology that was not widely available to the vendor community and therefore never reached critical mass. In contrast, today`s SAN paradigm has wide industry support and promises to become the standard for enterprise storage architectures in the new millenium.
For the most part, SANs are now based on "open" technologies--in the sense that they are not proprietary. These technologies are available from a wide array of vendors. SANs promise to allow multiple types of storage devices and networks to interoperate, while providing the highest levels of availability, bandwidth, and flexibility (e.g., connectivity distance, networking topology, and backbone options).
Key technology enablers
Critical to the development of the SAN paradigm has been the broad commercialization of Fibre Channel interconnect technology. This industry effort has produced a wide range of high-performance, relatively low-cost interconnects, switches, hubs, and routers. These components, as well as fiber and copper cabling, are available from a growing number of well-established and start-up storage vendors.
Fibre Channel technology can deliver wide bandwidth over very long distances and it can potentially support millions of devices that are connected to a fabric in a variety of ways (e.g., switches, hubs, and routers). These capabilities are far superior to current interconnect technologies, which are limited in terms of bandwidth, cabling distances, and the number of devices that can be connected.
Vendors first began delivering Fibre Channel devices in production quantities in 1997. Of the initial shipments, most were based on the Fibre Channel-Arbitrated Loop (FC-AL) interconnect scheme, rather than point-to-point or fabric architectures. At this time, the Fibre Channel community at large began promoting the technology, focusing on developing the core concepts and standards necessary for nonproprietary SANs so that open-systems models and standards could be developed for the entire storage industry.
Today, much of the talk in the storage industry centers on the features and benefits of SANs, not on how to get there with existing and new technologies. Without fully understanding how to interface with the existing enterprise infrastructure--much less leveraging existing assets wherever possible and practical--you can not achieve the overall benefits of SANs.
Defining the architecture
All SANs are meant to provide the following attributes:
- Anything-to-anywhere connectivity
- High availability
- Scalability (capacity and bandwidth)
- Centralized enterprise-wide storage resource management
- Comprehensive backup and restore
- Disk and tape pooling
- Third-party copying (or instant copy)
- Heterogeneous data sharing
These attributes and others require an architecture that enables maximum flexibility in connectivity, that has no single points of failure or bottlenecks, and that supports ubiquitous access to any file from any client or server across the enterprise. To achieve this requires a switched fabric based on a high-bandwidth, low-latency interconnect technology. Fibre Channel meets these requirements, but at this stage has little interoperability with existing interconnects and storage assets. To eliminate these shortcomings, some integrators have developed "bridge systems," which allow emerging SANs to integrate with existing interconnects.
For example, storage architects can use existing interconnects (such as SCSI and SSA) and storage assets (JBOD, RAID, tape libraries, optical jukeboxes, etc.) and new architectures (such as Fibre Channel fabrics) to build evolutionary SANs. Bridges also include functionality for data protection, volume management, data backup, and automatic fault recovery (restoration of connections via host and channel failover).
To embrace any new technology paradigm, there must be compelling financial benefits. The two most significant cost considerations in making architectural and purchasing decisions are:
- Total cost of ownership (TCO), a composite metric that includes hardware and software acquisition costs, management and support costs, user-related costs, and loss of business costs.
- Return on investment (ROI ) or more appropriately economic value added (EVA). These calculations are based on formulas that define the ratios between net profits and total assets in an effort to show true returns. EVA calculations correlate near-term results (those that exceed capital costs) with the decisions that actually produced these returns, and are more representative of the true payback.
These calculations relate to the entire costs borne by the organization in choosing a particular architecture and accompanying solution as well as derived financial and competitive advantage paybacks. When making your SAN deployment decision, it`s critical to factor in existing assets and infrastructure. These TCO reductions--along with such indirect ones as reducing losses due to downtime and eliminating a large amount of up-front training for administrators and users--will make your business case much stronger than the "rip-and-replace" approach advocated by many SAN vendors.
In addition, your EVA should also be dramatically higher. EVA is based on the new capabilities that a SAN brings to the organization, which equates to new sources of revenue, and on the timing of the payback, which will occur earlier since existing assets and infrastructure are leveraged.
Bridges can link existing interconnects (e.g., SCSI and SSA) with emerging Fibre Channel SANs.
Mike Piltoff is vice president of advanced products at Champion Computer Corp., in Boca Raton, FL.