Getting From Vendor-Centric To Data-Centric Challenges SANs.

IP storage will play critical role

Are you still confused about where and when to actually deploy SANs (Storage Area Networks) and NAS (Network Attached Storage) technologies? You are not alone and by the time everyone gets on the same page about SAN and NAS, both technologies will be well into their second era of functionality as both are evolving quickly. Each technology connects storage to a network of computing platforms and has its own set of benefits and tradeoffs that not only make the choices more difficult but also raise many questions about where these technologies will be in a few years.

NAS

NAS utilizes IP or network protocols, normally through Ethernet and Gigabit Ethernet that bypass standard host bus adapters. NAS relies on two de-facto standards that are driven by operating system considerations. Interoperability between Windows 2000-based systems and Unix environments is accomplished through the support of the widely used Common Internet File System (CIFS) and Network File System (NFS) and standards respectively. Network Attached Storage becomes easier to manage because of the common file sharing facilities offered by NFS and CIFS. It is well understood that improving storage management ranks as the most pressing issue for the storage industry today. NAS allows users to access data without server intervention reducing overhead and improving performance on the general-purpose application server. Without NAS, general-purpose servers can spend in excess of 40% of the computing cycles on 110 activity creating a very high application "server tax" or I overhead. Because NAS runs on a "thinner ope rating system" than the general-purpose server does, it can deliver similar performance with less compute power.

NAS offers a number of additional benefits as the widely used NFS and CIFS standards provide interoperability and increased manageability. In addition, the ideal state of true data sharing between clients is possible, though difficult, with NAS. True data sharing means sharing heterogeneous access (reads, writes, and updates) to a single copy of data and is one of the final hurdles remaining to be successfully conquered by the storage industry. The primary need for true data sharing is between Unix and NT systems. NAS enables re-centralizing LAN storage and facilitates reducing management costs.

Files And Blocks

The simple plug-and-play install, configuration, and management aspects of NAS are appealing as they reduce overall implementation costs. The use of the common and open NFS and CIFS standards is a plus as it forces vendors to move to non-proprietary storage solutions, often a task met with resistance. Adding storage capacity does not require taking a general-purpose server down and eliminates the need for IT personnel to work at specific or odd hours to add storage capacity. NAS remains a disk storage solution for what is commonly referred to as "File I/O" applications and is best suited for smaller sized applications. File level I/O addresses data stored on disk by its location within a file. This compares to block-level I/O, which addresses data by its physical location on disk. Databases represent an estimated 60-65% of the digital storage on disk today and are not effectively addressed by NAS. Therefore databases are referred to as "Block I/O" applications and are much more effectively addressed by the SA N--today. The NAS market is expected to reach $3 billion in 2001 and exceed $5 billion by 2003 according to most industry sources. While there are over 40 NAS companies today, two companies derive over 50% of the entire worldwide NAS market revenue.

SANs

Typically, a SAN involves the connection of storage devices like RAID (disk) and tape libraries through Fibre Channel switches and channel protocols to heterogeneous application servers. SANs combine the best of LAN networking models with server-storage connectivity and are referred to as a "Block I/O" solution enabling full support for the vast database market. SANs utilize a dedicated high-speed sub-network typically based on Fibre Channel to connect storage peripherals and are best suited for large file traffic, server consolidation, and shared backup resources. Data is directly transferred between storage devices and client computers. Theoretically, this bypasses the server bottlenecks and network control. By separating control from data, increased flexibility and performance should be achieved. File servers still can perform the function of network access and control. SANs shift data transactions and I/O away from application servers. In so doing, application server performance is improved. Storage netwo rks share a common back-end storage network. Data ownership is not assumed, as all authorized servers are provided with equal access to the information.

With a SAN, storage capacity can be added as required, enabling on-demand scalability. SANs do not place the storage subsystems directly on the network. Instead they put an 110 network with switches in between the storage subsystems and the server. This normally adds some degree of throughput delay in the form of network latency. Another potential problem is that SAN standards are still in the formative stages. As a result, we are witnessing continual announcements of proprietary SAN solutions by leading storage vendors as the time to reach agreement on open standards continues to move slowly. Vendor centric solutions will remain the predominant offering for the foreseeable future. SANs are more complex to install than NAS and typically involve solutions from multiple vendors to effectively implement. Customers are also faced with additional training expenses along with integration and testing of multiple vendor components to insure interoperability. With the availability of people resources a major concern, the complexity of a successful installation now looms as SANs biggest challenge. Estimates presently indicate a shortage of well over 600,000 IT workers in the United States alone. As a result of this, a clear 40% of the end users indicate they have no plans to deploy a SAN! Nonetheless, the SAN market is expected to reach $7 billion in 2001 and exceed $14 billion in 2003. Six vendors now command over 85% of the entire combined SAN and NAS market. New entrants in this market need a compelling value proposition or significant customer benefit that distinguishes them from the established businesses.

Next Steps

General understanding of NAS and SANs is improving and the industry is more comfortable with each than at any time previously. Though significant problems remain to be overcome with each implementation, the future for NAS and SAN is evolving quickly and gets even more exciting in terms of the mounting storage related problems that can be solved in light of 70-80% annualized storage growth. A significant new development is upon us referred to with a wide variety of terms. These include IP Storage or Storage over IP (SOIP), iSCSI (Internet SCSI), SAN over IP, Ethernet SAN, or IP-SAN and industry journals are packed each month with endless debates on the pros and cons of these activities. This new area of development refers to the ability to transfer blocks of data over an IP based SAN configuration and poses a longer-term threat to the current Fibre Channel SAN. iSCSI may actually become the early leader in the SOIP bonanza. This is evidenced by the fact that it is presently in the process of being standardized by an IEEE work-group. In the iSCSI environment, SCSI control commands, data, security, and error recovery functions are combined in standardized Ethernet packets that are sent over the Internet. These packets are received, opened, and converted to code that ultimately controls SCSI data storage drives. Similarly, data and responses retrieved from the drives are converted to iSCSI packets.

The evolving IP based storage network will play a critical role in SAN to WAN to SAN connectivity over long distances. The size of this market is unclear and may in reality be overstated. This technology embraces the basic NAS utilization of either native IP or IP tunneling methods over Fibre Channel frames to further expand SANs' reach. We cannot ignore this trend as industry leaders such as IBM, Cisco, Emulex, Nishan, Intel, and others have announced SOIP-based SAN solutions and the trend is gaining significant momentum and visibility. Keep in mind that Ethernet SAN technology is not ready today. As is the case with any major IT development, the long and drawn-out debates over standards can become impediments and must be dealt with. It will take time for a clear leader or de-facto standard to emerge. The IP storage trend further reflects the movement to a data-centric world while suggesting a longer-range convergence between NAS and SAN architectures. The distinction between NAS and SAN is just now beginnin g to blur. To the user, the only thing that really matters is whether or not it works and meets the needs of the business.

Problems always create opportunities. One thing is extremely clear; the significant problems of the past cannot be solved by the same level of thinking that created them.