Basics of the Fibre Channel Standard
The standard of choice for many of the high-performance video media storage platforms is the Fibre Channel (FC), which is more than just switching or a type disk drive interface. FC is a vast set of applications with a depth not unlike those of SCSI, SATA (serial ATA), and other storage device and network technologies.
For defining conventional 10/100/1,000 Mbps Ethernet networks, today’s end user is offered an abundance of choices in products and methodologies in which to assemble their network architecture. By contrast, FC, when used in high-performance video storage systems, will require an application-specific, sophisticated configuration and selection process.
You’ve probably wondered why video server manufacturers utilizing FC switches have such a strong emphasis on employing their own thoroughly qualified and configured devices. This is primarily because the performance of their storage systems depends upon the precise applications of the FC fabric (switches) in order to meet and maintain a set of mission critical parameters that broadcasters rely on for their principle assets.
In getting reacquainted with FC, we’ll explore the technologies and terminologies which have evolved over the previous decade and a half. FC has a language unto itself; not unlike that found in Internet protocol, Ethernet, SSA (serial storage architectures), ATM and other networking and storage technologies. Even though the mainstream broadcast technical staff may seldom need to utilize the details of these technologies, it is useful to understand why FC is configured and for what applications it supports.BREAKING IT DOWN
At its most general entry level, FC initially was intended as a practical and expendable means for the rapid transfer of data between computing systems—from workstations through supercomputers and from desktop computers to sophisticated storage devices and other peripherals.
FC is a general terminology for an integrated set of standards that evolved through the T10 Technical Committee of the InterNational Committee on Information Technology Standards (INCITS, pronounced “insights”), which is responsible for SCSI storage interfaces; including all SCSI standards and the mappings of SCSI to FC. INCITS is accredited by and operates under rules as approved by ANSI.
| Table 1: Fibre Channel protocols for fabric (switches) typically found in video server applications that employ external SAN-like storage architectures.|
An additional T11 INCITS committee is responsible for FC interfaces and for storage (network) management. Each undertaking by these committees is referred to as a project. T11 FC interface projects are varied and cover such topics as a 10 Gb interface (FC-10GFC); FC for the avionics environment (FC-AE); Fibre Channel-Arbitrated Loop (FC-AL) and FC 2nd Generation Arbitrated Loop (FC-AL-2).
As an example, in the 1990s, a project titled “Fibre Channel Audio-Visual (FC-AV)” was established with a goal to develop definitions for the transport of ITU-R BT-601 and ISO/IEC 13818 (MPEG) protocols over FC. FC-AV defines an FC mapping layer (i.e., FC-4, which maps upper level protocols to FC-2) that uses the services as defined by INCITS project 1331-D, Fibre Channel Framing and Signaling Interface (FC-FS), to transmit audio and video information in such environments as broadcast and avionics systems.
The FC-AV project looked to support definitions and applications for film formats and compressions schemes, YUV and RGB formats and more, and for the synchronization of the FC segment with existing analog and digital segments. The latest draft (Revision 1.71, T11/01-344v2) was dated Dec. 3, 2001, and has, as of yet, not become a standard and at this point looks to be stalled on any further development.
In similar fashion to the beloved serial digital video (SDI) transport, FC must also address issues that deal with the complications of jitter, which is covered under the Methodology of Jitter Specification (FC-MJS). FC project working groups also cover topics related to FC tape, FC virtual interfaces, and FC fabrics. While the full technical details of FC can be found in the T11 committee documents, more fundamental information may be found at www.fibrechannel.org.BACK TO THE BASICS
For the basics, we’ll look at the two basic types of data communication between processors and peripherals: channels and networks.
A channel provides either a direct or a switched point-to-point connection between the devices. A channel is typically hardware intensive, thus permits the transport of data at high speed but with low overhead.
| Table 2: Classes of Service for Fibre Channel&mdashClass 2 is usually employed in video server applications with larger SAN systems.|
By contrast, a network is an aggregation of distributed nodes with their own protocols that support interaction among these nodes. Nodes connect devices such as workstations, file servers and other peripherals. Being more software-oriented, a network has relatively high overhead and is consequently slower than a channel, but more diverse. The benefit to a network is it’s capable of a more extensive range of tasks. Networks operate in an environment of unanticipated connections and channels operate with a finite, predefined set of addresses and generally among far fewer devices.
FC sought to combine the best-of-breed properties for these two communications methods into an interface that meets the needs of both channel users and network users.
Like other networking systems, FC has elements under which it is structured. These include specifications of FC protocols, classes of services, and modes of operation. You’ll find which of these FC specifications are supported in the manufacturers’ data sheets.
FC protocols, defined in T11 as projects, cover a wide range of topics. Some of those applicable to video servers can be found in Table 1. FC services classes—defined as Class-1 through Class-3—are used to ensure the efficiency of transmission for different types of traffic (see Table 2). Users select appropriate service classes for their applications, for example packet length and transmission duration, and then allocate those services according to the fabric login protocol.
In FC terms the switch connecting the devices is called the fabric. The physical link is the two unidirectional signal carriers (either optical fiber or copper) that transmit in opposite directions with a pair of associated transmitters and receivers. Each link is attached to port, with a transmitter at one end and a receiver at the other. When the configuration utilizes a fabric, the fiber or copper connection attaches to a node, called the N_Port; and to the fabric at an F_Port.
| Fig. 1: Three Fibre Channel topologies|
FC systems require that ports log-in with each other, with the fabric using a defined protocol for the application. It doesn’t matter if the fabric is a circuit switch, an active hub or a loop. The topology is selected depending on the system performance requirements or physical packaging. Recent storage systems offered by video and nonlinear editing systems have the fabric built into the actual drive chassis, assuring solid reliable connectivity between the hard drives and the external peripherals or hosts attached to the storage array.
FC topologies include point-to-point, crosspoint switched or arbitrated loop (see Fig. 1). For ensured reliability, video server manufacturers will provide dual sets of connections between the play-out and record (I/O) chassis and other elements including storage, fabric or other I/O-chassis.
The performance of FC allows data transfer and system bandwidths grow to the levels necessary for multiple instances of HD and SD content to move between storage and I/O-device. The capabilities of FC are enormous and the performance becomes highly extensible because it combines the best of both the channels and networking data communications systems.
Touching only on a basic introduction to FC—mainly the physical and signaling interfaces (FC-PH) portion—the depth of FC networking and storage is vast. Looking forward is gigabit Ethernet, now totally intermingled with high-performance video server architectures. FC and GigE interfaces are necessary to extend the reach and capabilities of both. Gateways, sometimes called bridges, are employed as interfaces for transferring data between platforms such as catch servers or archives, or over great distances over WANs.
In part, the development of both the hardware and software systems that match the protocols and physical layers between these data communications networks is driving force for putting the video server in the forefront of media services storage and delivery.