As articulated several years ago, the mission of the Storage Networking Industry Association (SNIA) is to ensure that storage networks become efficient, complete, and trusted solutions across the IT community. While this is a noble goal for a vendor association, it implies that SANs have not been efficient, have been incomplete, and have not established widespread trust within the broader end user community. While all of these assumptions are still valid, significant progress has been made over the five years of the SNIA’s existence.
Efficiency of SANs
Efficiency is a broad category that spans everything from performance to ease of implementation and management. SAN technology has demonstrable efficiency in terms of performance. Host bus adapters (HBAs) operate at wire speed and with minimal CPU utilization. SAN fabrics offer high performance 1 and 2 gigabit per second (Gbps) switching, as well as the ability to aggregate interswitch links to optimize throughput in a multi-switch fabric. SAN-attached storage devices offer multiple 1 and 2 gigabit ports and high performance RAID controllers and caches that minimize the latency of read and write operations.
New IP storage technologies have also demonstrated efficiency in terms of wire-speed throughput and optimum utilization of local and wide area IP networks. With 4 gigabit and 10 gigabit interfaces on the horizon, even greater performance is expected, making SAN technology the obvious choice for efficiently moving block data from servers to storage assets.
The same level of efficiency has not been reached, however, in terms of ease of implementation and management. Beyond the underlying transport, storage is a complex discipline that has required manual administration to create, assign, and monitor storage resources. This inherent complexity is exacerbated by a storage network, since the ability to share storage assets between multiple servers also requires close monitoring of asset assignment.
Installing and optimizing a SAN configuration is far more complex than a comparable direct-attached solution, but once established, SANs offer significantly more value in terms of consolidated storage, backup, and other applications. The result is that while customers may feel some new pains in implementing SANs, many old and familiar pains thankfully disappear.
Making SAN solutions efficient requires addressing all the remaining inhibitors to configuration and management. The SNIA Storage Management Initiative (SMI) is attempting to resolve inefficiencies in storage networking by offering a uniform management interface that spans all vendors and all aspects of storage transport and placement. This goal has been central to the recent activity of the SNIA and a major beneficiary of SNIA funding and volunteer effort. The SMI effort has been elevated as an official SNIA Architecture, with shepherding by SNIA through ANSI and the appropriate standards bodies.
In the long term, a viable SAN will offer the dual benefits of efficient performance and efficient management. In the near term, customers frustrated with the current inefficiencies of SANs can expect steady progress towards resolution, but no immediate remedies. SMI and virtualization initiatives are attempting to mask the complexity of SANs and simplify management, but these are intricate problems that cannot be fixed overnight. Multi-vendor cooperation between storage, switch, and software suppliers is at least accelerating the development of new management solutions, particularly compared to the sluggish performance of vendors on this front a few years ago.
Complete SAN Solutions
Providing a complete, end-to-end SAN solution is predicated on many components, including efficient management, standardization of protocols, and the interoperability of products. To date, customers have had to rely on VARs or integrators that can source and pre-stage the requisite boxes and software to build a SAN, or simply buy a preconfigured SAN solution from a major storage provider. Few customers have had the internal resources to select the best of breed products from the marketplace, tailor a SAN to their own specific application requirements, and perform their own interoperability testing and debugging to get the solution to actually work.
Progress on standardization has focused primarily on the SAN transport. The SNIA, FCIA, ANSI, and IETF have been successful in codifying standards and standards-compliance testing for Fibre Channel and IP transport protocols. This is the fruit of many years of work by multiple vendors, spanning the Fibre Channel SANmark certifications of the FCIA, standards compliance programs of the University of New Hampshire (UNH), and standards test suites being developed by the SNIA Interoperability Committee.
Standards development and standards compliance increase the likelihood of interoperability, and therefore the ability to more easily build complete end-to-end storage networks. But increasing the odds of interoperability alone is not sufficient. No customer wants their data center to become a lab for debugging vendor interoperability issues.
The SNIA has made steady if slow progress in advancing interoperability through showcase demonstrations at Storage Networking World and other venues. What the customer does not see at these events, though, is the months-long effort and commitment of resources by vendors behind the scenes to configure, debug, and stabilize the demo solutions. Although the pre-staged configurations may not map directly to a customer’s specific requirements, the interoperability initiatives are teaching vendors themselves what is required to deploy and support real-world, complex multi-vendor installations.
Vendors that are committed to a solutions approach to customers assume the responsibility for validating all the components required to build a complete and operational storage network. This forces cooperation and information sharing between the various hardware and software vendors, proactive testing and verification of products, and most importantly, a concrete understanding of the customer’s actual needs.
Storage networking is inherently complex, and although canned storage consolidation or tape backup packages may be useful as a starting point, customers will always require additional components, features, or software to satisfy their unique business requirements. In an open systems world, providing complete, end-to-end solutions is always a work in progress.
Of the SNIA’s mission goals, trust is perhaps the most challenging. Trust for customers equates to reliable and predicable performance, vendor responsiveness to issues or new requirements, consistency in product development to avoid forklift upgrades or obsolescence, and the ability to scale working solutions throughout the broader enterprise. In a society nurtured on planned obsolescence, exploitation of the market, and profit above all else, trust is already a rare commodity.
Establishing trust in the more conservative and demanding storage market may often require vendors to sacrifice their immediate self-interest in order to ensure the success of the industry as a whole. Switch interoperability, for example, was retarded by several years due to the stubborn effort of one vendor to dominate the market by subverting open systems initiatives.
As an association of overtly for-profit vendors, the SNIA itself has had to deal with the trust issue. Some industry pundits view the SNIA as a conspiratorial society whose mission is to dupe poor customers into buying an immature and unproven technology. SNIA vendor members, however, are not Goa’uld System Lords intent on ruling the universe.
Like other industry associations, the SNIA brings together both complementary and competing vendors to promote a technology on the understanding that the success of individual members is tied directly to the success of the industry as a whole. Progress towards the goal of providing efficient, complete, and trusted shared storage solutions is thus accelerated or constrained by the degree to which participating members cooperate or compete under the SNIA umbrella.
Bringing trusted solutions to a broader customer base is also facilitated by the involvement of customers themselves. The SNIA’s end user initiatives and new customer channels such as www.storagenetworking.org are intended to more directly engage the consumers of storage networking technologies into the cycle of solutions discovery and features prioritization. This will make the goal of providing trusted shared storage solutions more attainable and more closely aligned with real customer needs.