This year, IP will emerge as an alternative to Fibre Channel for building storage area networks, presenting end users with new options.
By Ron Levine
With the current variety of IP storage protocols (including iSCSI, iFCP, and FCIP) poised to take on Fibre Channel, the level of confusion among end users has reached the point where some have simply abandoned the concept of storage area networks (SANs) until the market can come to a consensus as to how SANs should best be deployed.
Adding to the confusion is the network-attached storage (NAS) option. "It's clear that mixed messages from the NAS and SAN camps are confusing some customers," says Derek Gamradt, chief technology officer at StorNet, a storage solutions and services vendor based in Englewood, CO.
Despite the proliferation of storage networking choices, there are only two options to direct-attached storage: NAS and SAN (see figure).
NAS enables end users to quickly bring storage to the network, but NAS works best in situations where there is a relatively low volume of data traversing the network. SANs may be a superior option for large enterprises that need to store and manage vast amounts of information while maintaining a high-performance network environment.
There are numerous benefits to upgrading from direct-attached storage to networked storage (either NAS or SAN), as shown in the box. "The options for implementing storage networks are simple," says Gamradt. "You can use a file-level, shared storage configuration such as NAS, a high- performance block-level switched fabric approach such as SAN, or a combination of the two approaches."
NAS and SAN architectures can complement each other to provide access to different types of data. Both technologies satisfy the need to remove direct storage-to-server connections to facilitate more-flexible storage access, and both have distinct advantages, as shown in the table.
Fibre Channel vs. IP SANs
Once a decision has been made to implement a SAN, IT managers have to address the issue of whether to build it with Fibre Channel or IP. Until recently, the only viable means of switching data paths to a storage device was through a Fibre Channel switch. However, the emergence of IP storage protocols such as iSCSI, iFCP, FCIP, and others has extended this capability to traditional IP networking switches as well. For local SANs, iSCSI is emerging as the number one challenger to Fibre Channel's dominance.
"A year ago, you could question whether [block-level] IP storage would break through, but it's now obvious that this approach will complement, and ultimately compete with, Fibre Channel," says Farid Neema, president of Peripheral Concepts, a Santa Barbara, CA, storage consulting and research firm.
iSCSI has the lion's share of attention in the IP storage market today and is expected to gain industry-wide acceptance this year when the Internet Engineering Task Force (IETF) ratifies the standard specification (which is expected in the second quarter). When that occurs, the debate surrounding which topology is best (iSCSI or Fibre Channel) for SAN implementations will heat up.
The case for Fibre Channel
Fibre Channel was specifically designed to address server-to-storage interface limitations. Fibre Channel-based SANs deliver operational benefits not previously possible with standard connections such as direct-attached SCSI. For example, by connecting RAID to the back-end of a server over a Fibre Channel bus, higher bandwidth results in quicker I/O transfers over longer distances than is possible with the SCSI interface. The RAID and Fibre Channel combination also improves storage subsystem reliability through fault-tolerant, redundant data paths.
Switch-based Fibre Channel SANs allow virtually unlimited scalability, providing a solid infrastructure for long-term growth. Fibre Channel supports multiple protocols and has a current bandwidth limitation of 200MBps and a connectivity limit of 10km. Each storage unit on a Fibre Channel network is a peer node, allowing for the flexibility of direct device-to-device communications via either arbitrated loop or switched-fabric topologies.
Most Fibre Channel devices are dual-ported. Using both ports in a dual-loop configuration provides a redundant path to/from the device, guaranteeing access should one path fail. Fibre Channel interfaces provide the performance required for a variety of bandwidth- intensive storage applications.
Fibre Channel switches provide simplified storage device scalability, hot plugging of devices, and isolation between functions. This translates into easily scalable bandwidth and improved availability.
The case for iSCSI
What can you gain with iSCSI? One of Fibre Channel's shortcomings is that it requires new skill sets for building and managing the storage network. In addition, the price per port of Fibre Channel is up to five times that of standard Ethernet IP ports. IP storage developers have addressed these issues by making use of the existing TCP/IP network infrastructure, thus eliminating the need for new expertise (and training) while holding down port costs.
The IP-based iSCSI protocol stores/retrieves data to/from any SCSI-compliant storage device over standard Ethernet, allowing for host-to-storage communications over a LAN, WAN, or the Internet. iSCSI is economical because it leverages the existing networking infrastructure and requires no additional storage management training.
How does iSCSI work? Basically, the iSCSI protocol allows the existing SCSI command set to be transported over standard IP networks.
iSCSI encapsulates SCSI commands, which are generated in response to an application's request for data, into an IP frame. The IP frame has a header added, and the data request is sent over a TCP/IP-Ethernet network to a storage device. At the receiving end, the packet is de-encapsulated. The SCSI commands are forwarded to the SCSI controller for transfer to the storage device. When the request for data is filled, the data is returned to the application in the same way (i.e., block-level SCSI data is encapsulated into an IP packet). iSCSI is a bidirectional host-to-storage device protocol that combines standard SCSI commands with TCP/IP frames and routes them, over the network, to their destination.
One of the key benefits of iSCSI is that it can transport standard storage applications and data over the TCP/IP network without modifications. After it is installed on the server, the iSCSI driver operates one layer below the standard SCSI driver, intercepting the SCSI blocks and forming the iSCSI data frame.
This frame is then transmitted via a network interface card (NIC) or host bus adapter (HBA). Applications are unaware of the change from SCSI to iSCSI and therefore continue to operate as they would with standard SCSI.
One of the key enablers of iSCSI will be TCP/IP co-processors that will offload much of the overhead associated with the IP stack (see InfoStor, January 2001, p. 1). Several HBA and NIC vendors are developing these cards, with shipments to end users expected in the first half of this year.
According to StorNet's Gamradt, iSCSI has many benefits, but some of the most prominent are the following:
- The elimination of Fibre Channel's 10km distance limitation, resulting in an inexpensive capability for long-distance remote backup, disaster recovery, and data mirroring;
- The utilization of the standard TCP/IP network reduces the need for specialized hardware, expertise, and training, resulting in significant cost savings;
- The capability to run existing applications and data without modifications;
- The ability to build an inexpensive SAN, which is particularly advantageous for smaller companies.
Another promising iSCSI application is the extension of isolated SAN "islands" across WANs.
"In the short term, there will be a co-existence between Fibre Channel and IP-based storage networks," Gamradt predicts. "The most logical choice for initial IP-based storage networks will be in data-replication applications that require a high-speed connection over long distances. While we are bullish on IP-based SANs in the long term, we are still recommending Fibre Channel for immediate SAN requirements."
"Both Fibre Channel and IP SANs will co-exist for a long time," agrees Peripheral Concepts' Neema. "Fibre Channel has acquired great momentum, and the technology will evolve to a point that will make it difficult for enterprises that have already invested in a Fibre Channel SAN to move away."
Ron Levine is a freelance writer and an Information Systems instructor at Santa Barbara City College. He is a regular contributor to InfoStor and can be contacted at firstname.lastname@example.org.
Benefits of storage networks (NAS and SAN) vs. DAS
- Remote connectivity of storage
- Reduced storage management costs through consolidation
- Ease of scaling storage capacity and processor performance
- Separation of storage and server purchases
- Centralization of storage management and support
- Single-image data availability to users with different operating systems
- Avoidance of single server bottlenecks
- Avoidance of networking bottlenecks
- Removal of backup data path from the LAN
- Server-free backup
- Increased storage resource utilization
- Improved data availability