Introducing Storage Area Networks
Here`s a primer to get you started in understanding the benefits of SAN, SAS, and NAS architectures.
By Michael Peterson
As IT organizations re-engineer distributed networks to achieve continuous operations and to host mission-critical applications, they are increasingly considering an architecture that is common in data centers. Mainframe-based data centers use a network storage interface called ESCON to connect mainframes to multiple storage systems and distributed networks, an architecture that is referred to as a storage-area network (SAN).
In a typical data center, SANs account for approximately 25% of all network traffic. What`s new is that SAN architectures are now being adopted in distributed networks using low-cost interconnect technologies such as SCSI, SSA, and Fibre Channel.
A SAN is a high-speed network, similar to a LAN, that establishes a direct connection between storage elements and servers or clients. The SAN is an extended storage bus that can be interconnected using technologies used in LANs and WANs, such as routers, hubs, switches, and gateways. A SAN can be local or remote, shared or dedicated, and it uniquely includes externalized and central storage. SAN interfaces are usually ESCON, SCSI, SSA, Fibre Channel, or HIPPI, rather than Ethernet.
It doesn`t matter whether a SAN is called a storage-area network or system-area network because the architecture is the same. Either way, SANs create a method of attaching storage that is revolutionizing networks, resulting in significant improvements in availability and performance.
SANs are currently used to connect shared-storage arrays, to cluster servers for failover, to interconnect mainframe disk or tape resources to distributed network servers and clients, and to create parallel or alternate data paths for high-performance computing environments.
In essence, a SAN is nothing more than another network, like a subnet, except that it`s implemented with storage interfaces. SANs enable storage to be externalized from the server, allowing storage devices to be shared among multiple host servers without affecting system performance or the primary network.
SANs are not new. The benefits are well proven because the architecture evolved from mainframe DASD. In fact, Digital Equipment`s VAX/VMS network environment is based on a SAN architecture and clustered servers. And vendors such as EMC already have a large installed base of SAN-attached arrays.
So, what`s new? The technology is now moving into mainstream distributed networking and will be the common way of attaching and sharing storage within a few years. Sometimes referred to as "the network behind the server," a SAN represents a new model that has evolved with the advent of shared, multi-host enterprise storage. A SAN bypasses traditional network bottlenecks and supports direct, high-speed data transfer in three different ways: server-to-storage, server-to-server, and storage-to-storage.
However, as various product camps (e.g., networking and storage) praise the merits of their solutions, SAN architectures and terminology often get confused. This article provides a simple set of definitions and terms for potential industry adoption.
NAS and SAS Defined
Storage can be attached to the network in three ways. Virtually all (98%) of today`s server-storage connections are bus-attached, usually via SCSI or IDE. Bus-attached storage operates through the server, so availability and performance are limited by the server`s capabilities. In contrast, storage can be externalized from the server via network-attached storage (NAS) or SAN-attached storage (SAS). From an engineering standpoint, NAS and SAS are very similar, but their implementations differ.
NAS is represented by a storage array that connects directly to the messaging network via a LAN interface such as Ethernet using common communications protocols. A NAS device functions as a server in a client/server relationship; has a processor, operating system, or micro-kernel; and processes file I/O protocols such as SMB and NFS. The NAS architecture was pioneered by vendors such as Auspex and Network Appliance.
SAS, on the other hand, is a shared storage repository that is attached to multiple host servers via storage interfaces such as SCSI, FC-AL, or ESCON. SAS was pioneered by vendors such as EMC.
Another key terminology point deals directly with SAN architectures. The three SAN components are the interface, the interconnect, and the fabric. Though often mixed, these terms are distinct elements of the SAN. Think of the relationship of these three components as a chained sequence: server--interface--interconnect--fabric--interconnect--interface--storage.
Interfaces, Interconnects, and Fabrics
SCSI, FC-AL, SSA, ESCON, and HIPPI are common SAN interfaces. All of them allow storage to be externalized from the server, and each can host shared storage configurations for clustering. Multiple channels or loops can be installed to increase performance and redundancy. (It is incorrect to say that parallel SCSI cannot be extended, multiplexed, switched, and connected via gateways to WANs just like serial interfaces such as Fibre Channel and SSA.)
SAN interconnects include devices such as extenders, multiplexors, hubs, routers, gateways, switches, and directors. SAN interconnects tie storage interfaces together into a variety of network configurations across large distances. Interconnects also link SAN interfaces to SAN fabrics.
Switched SCSI, switched Fibre Channel, and switched SSA are the most common SAN fabrics. With gateways, SANs can also be extended across WANs. The advantages of using switches to build centralized, centrally managed, consolidated storage repositories shared across multiple applications are many.
What Are the Benefits?
SANs improve the performance of almost any application that moves data around the network. Just like conventional "subnets," SANs add bandwidth for specific functions without placing an extra load on the primary network. In this fashion, SANs complement LANs and WANs.
SAN technology enables a network architecture of shared multi-host storage, connecting all storage devices and interconnecting remote sites. Eventually, this configuration will be standard for centralized networks running mission-critical applications. Both disk and tape operations are centralized, attached via the SAN, and are faster and more resilient. As the IT community has learned in the database market, the key to application performance is usually the I/O network, not the storage device.
Key benefits of a SAN architecture include:
- Higher application availability. Storage is externalized, independent of the application, and accessible through alternate data paths, as in clustered systems.
- Better application performance. Because host server and bus overhead degrade performance, independent SAS arrays outperform bus-attached storage configurations, while being compatible with performance clusters.
- Easier, centralized management. SAS configurations encourage centralization, which reduces management time and costs.
- Centralized and consolidated storage. This translates into higher performance, lower cost of management, and greater scalability, flexibility, reliability, availability, and serviceability.
- Practical data transfer, vaulting, and exchange with remote sites. SANs enable cost-effective implementations of high-availability disaster protection configurations, including remote clusters and mirrored arrays.
SANs are being deployed in a variety of applications. At a high level, Strategic Research Corp. has identified six application areas that currently use SAN architectures for data transfer, and there may be more application areas in the future.
Externalized storage, implemented in a network architecture, is a generic application that is applicable to a variety of network applications.
Clustering is usually thought of as a server process that provides failover to a redundant server or scalable processing by running multiple servers in parallel. In a cluster, the SAN provides the data pipe that allows storage to be shared. For example, the Microsoft Cluster Server (MSCS), which is an availability (as opposed to performance) cluster, shares a single array between two servers attached via a SCSI SAN.
Data protection applications require storage redundancy on a dynamic basis. Largely because of their performance and secondary data paths, SANs are ideal interconnects for these applications because they provide storage mirroring, remote clustered storage, and other high availability data protection solutions. SANs do not affect the primary network or the servers, and they provide fault- tolerant redundancy.
Data vaulting is the process of transferring data, usually for the purpose of archiving or logging, to a remote site. For these applications, SANs provide a very efficient transmission medium.
Data interchange and disaster recovery applications are very similar and use SAN architectures the same way (whether local or remote), but for different purposes. SANs provide a very efficient pipe for moving data off-site or between sites. Disaster protection systems can be built on remote vaulting (backup) processes or with high availability remote array mirroring or clustering.
The many facets of implementing SAN technology can be confusing as IS organizations work through the issues of how to best implement a SAN architecture. To solve these problems, an industry organization was recently created--the Storage Networking Industry Association (SNIA).
Membership and participation details are available at www.snia.org.
Bus-attached storage is based on a short-distance connection between the host and the storage device via interfaces such as SCSI or IDE. Network-attached storage and SAN-attached storage configurations externalize storage from host servers.
SAN interfaces include technologies such as Fibre Channel, SCSI, and SSA. SAN interconnects can be switches, gateways, hubs, extenders, multiplexors, routers, or directors. Fabrics are created with switched Fibre Channel, switched SCSI, or WANs.
Strategic Research has identified six application areas that can immediately benefit from SAN implementations.
Michael Peterson is president of Strategic Research Corp. (www.sresearch.com), a market research firm focused on network storage, in Santa Barbara, CA. He is also president of SNIA.