Fiber Channel over Ethernet
Fiber Channel over Ethernet ( FCoE ) is a protocol for the transmission of Fiber Channel packets in full-duplex Ethernet- based networks. The main aim of the introduction of FCoE is the I / O consolidation on the basis of Ethernet (IEEE 802.3) with a view to reducing the physical complexity of network structures, especially in data centers .
With FCoE it is possible to use a uniform physical infrastructure for both the transmission of Fiber Channel and conventional Ethernet. The scalability and the higher bandwidths of Ethernet-based network structures with the currently common 10 Gbps (soon also 40 or 100 Gbps) represent significant advantages. On the other hand, the use of Ethernet for the transport of Fiber Channel frames also has disadvantages of the classic Ethernet protocol, such as the loss of frames in overload situations, so that some improvements to the Ethernet standard are necessary to ensure reliable transmission based on Ethernet. These extensions are being promoted under the term data center bridging .
The standardization of FCoE began in April 2007 within the FC-BB-5 working group of T11 and was submitted to INCITS for publication as the FC-BB-5 Draft Standard on June 4, 2009 .
Application and motivation
The main application of FCoE can be seen in I / O consolidation (also under the term network convergence ). In this regard, the consolidation of Ethernet-based networks and Storage Area Networks (SAN) on a common, high-performance physical infrastructure should be mentioned. The main advantages here are likely to be the significantly lower cabling effort and the overall lower investment costs for a common infrastructure, as well as the better resource utilization of the physical infrastructure.
Another advantage is certainly to be seen in the currently prevailing virtualization strategy of many data center providers in cooperation with FCoE, since ultimately FCoE in practice also represents a type of virtualization technology based on physical media and as such can partially be transferred to the host systems for virtualized servers. Such consolidation strategies can therefore:
- Reduce effort and costs for a physical infrastructure consisting of network elements and cables
- Number and total costs of the necessary network cards (NICs) in end devices such as B. Reduce servers
- Reduce operating costs (energy supply and heat dissipation).
Demarcation
Like iSCSI, FCoE transmits block data via subordinate protocol layers, but the two protocols are fundamentally different. iSCSI can be classified as an application protocol according to the OSI model , as it is based on TCP as the transport protocol. This makes it a routable protocol and can therefore in principle also be used beyond the boundaries of a data center. With iSCSI, however, due to these properties and the lack of additional requirements for subordinate protocol layers, one accepts a relatively large package overhead , so that the overall transport performance, especially within a data center, is rather low compared to Fiber Channel. FCoE, on the other hand, does not use network and transport layer protocols such as IP and TCP and is based on Ethernet. FCoE benefits from the significantly lower packet overhead and the associated higher transport performance and accepts the local limitation to Layer 2 domains. In addition, there are special challenges for the Ethernet standard used, which data center bridging takes into account. A data center network infrastructure is required to achieve this, which generally makes migration difficult. Therefore, FCoE is particularly important when setting up new network structures.
Functionality
FCoE encapsulates Fiber Channel natively in Ethernet frames, with the FCoE protocol specifications replacing layers FC-0 and FC-1 of the Fiber Channel stack (see frame format). This makes it possible to integrate existing storage area networks into a (new) Ethernet infrastructure without major changes. Since FCoE is based directly on the Ethernet protocol stack, it differs significantly from iSCSI, which also transmits SCSI block data via network protocols, but is based on the TCP stack and is therefore routable in contrast to FCoE.
The standard of the classic Ethernet emerged at a time when only manageable amounts of data were transmitted with the help of networks. During this period (Engl. Thus, there was little need for flow control mechanisms flow control ). Fiber Channel, on the other hand, implements flow control mechanisms, since block data i. A. are very sensitive to transmission disturbances and these cannot be intercepted otherwise. For this reason, FCoE requires extensions to the classic Ethernet standard to include flow control mechanisms to prevent overload situations and the associated frame losses. The IEEE is working on these extensions in the Data Center Bridging Task Group.
The main extensions of the Ethernet standard in data center bridging can be summarized as follows:
- Priority-based Flow Control (PFC) Version 0 Specification (Submitted to IEEE 802.1Qbb working group).
- Enhanced Transmission Selection (ETS) Version 0 Specification (Submitted to IEEE 802.1Qaz Working Group).
- The Data Center Bridging eXchange (DCBX) Version 0 Specification (Submitted to IEEE 802.1Qaz Working Group).
Furthermore, Ethernet must meet the following requirements:
- Encapsulation of native Fiber Channel frames in Ethernet frames.
- A mapping between Fiber Channel N-Port IDs (FCIDs) and Ethernet MAC addresses.
Terminal adapter
End devices are connected directly to the FCoE fabric with the help of so-called Converged Network Adapters (CNAs). Such an adapter provides both fiber channel host bus adapter (HBA) and classic network interface card functions on one piece of hardware. This relieves the CPU of low-level frame processing and the provision of classic SCSI functionality. Alternatively, FCoE encapsulation can also take place in software with sometimes significant CPU utilization.
The first FCoE terminal was implemented by Frederick Knight for NetApp .
Frame format
Both Fiber Channel and classic networks have their own stacks, each of which provides a large number of functionalities. The FC stack consists of five layers (FC-0 to FC-4), while Ethernet is typically mapped in a 7-layer OSI reference model , covering the first two layers (physical layer and data link layer). FCoE is designed to transmit the FC layer FC-2 in Ethernet. This means that the upper Fiber Channel layers FC-3 and FC-4 can also be mapped via IEEE 802.3 Ethernet layers. A typical FC frame has a useful data length of up to 2112 bytes plus header and CRC . The maximum size of an FCoE frame is 2180 bytes. FCoE therefore requires the support of the Ethernet infrastructure for the transmission of so-called baby jumbo frames with a size of up to 2.5 kB so that the FC frames contained are not split up.
The first 48 bits of the frame specify the destination MAC address , while the second 48 bits specify the source MAC address. The 32-bit IEEE 802.1Q marking offers the same functionality as for VLANs and enables multiple virtual local area networks on the same physical infrastructure. FCoE is encapsulated directly in Ethernet frames using a special Ethertype (0x8906). A further 16 bits are provided for this, followed by a 4-bit version field. The next 100 bits are reserved and a further 8 bits indicate the start of the Fiber Channel frame. An 8-bit end-of-frame delimiter (EOF) is followed by a further 24 reserved bits. The frame ends with 32 bits for the FCS function, which provides the error detection for the Ethernet frame.
With encapsulated Fiber Channel, the Ethernet header is followed by a 24-byte long FC header, just like in the actual FC standard, followed by the FC user data, including the Fiber Channel CRC. The usual cyclic redundancy check is (engl. Cyclic redundancy check , therefore usually CRC) used for error correction. Since the classic FC frame is completely retained or used, it is possible to connect an existing Fiber Channel SAN to an FCoE switch, the FCoE switch then acts as a gateway and simply removes the outer Ethernet frame This creates an integration of the FC functions according to Ethernet, without the need for a dedicated gateway.
Another important component of the FCoE standard is the FCoE Initialization Protocol (FIP), which recognizes and initializes the FCoE capabilities of the network components of an Ethernet cloud. FIP again uses a dedicated Ethertype (0x8914).
Similar standards
- ATA over Ethernet : With ATA over Ethernet (ATAoE), ATA / ATAPI packets are encapsulated in Ethernet packets. Similar to FCoE, ATAoE is not encapsulated in TCP / IP, ATAoE is therefore not routable.
- Fiber Channel over IP With Fiber Channel over IP (FCoIP), Fiber Channel packets are also encapsulated in TCP / IP. FCoIP is therefore routable.
- HyperSCSI With HyperSCSI, SCSI packetsare encapsulated in Ethernet packets. Similar to FCoE, HyperSCSI is not encapsulated in TCP / IP, HyperSCSI is therefore not routable.
- iSCSI (SCSI over IP): With iSCSI, SCSI packetsare encapsulated in TCP / IP. This results in a higher overhead, but iSCSI is routing-capable and, like Fiber Channel over IP , can also be used in wide area networks.
See also
- Fiber Channel
- Internet Protocol over Fiber Channel
- Linux iSCSI target with Fiber Channel over Ethernet support
credentials
- ↑ http://www.fcoe.com/
- ↑ The Data Center Bridging (DCB) Task Group (TG)
- ^ Technical Committee T11
- ↑ FC-BB-5 Draft Standard (PDF file; 1.9 MB)
- ^ Data Center Bridging Task Group
- ↑ Priority-based Flow Control (PFC) Version 0 Specification (PDF; 94 kB)
- ^ IEEE 802.1Qbb working group
- ↑ Enhanced Transmission Selection (ETS) Version 0 Specification (PDF; 100 kB)
- ↑ IEEE 802.1Qaz Working Group
- ↑ Data Center Bridging eXchange (DCBX) Version 0 Specification (PDF; 496 kB)
- ↑ 2009 SDC Featured Speakers ( Memento of the original from January 3, 2011 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice.
- ↑ http://www.t11.org/ftp/t11/pub/fc/bb-5/07-479v0.pdf
- ↑ http://www.t10.org/ftp/t11/document.08/08-208v1.pdf ( page no longer available , search in web archives ) Info: The link was automatically marked as defective. Please check the link according to the instructions and then remove this notice.
