GENERIC FRAMING PROTOCOL
The Generic Framing Protocol (GFP), defined in ITU-T G.7041, is a mechanism for mapping constant and variable bit rate data into the synchronous SDH/SONET envelopes. GFP support many types of protocols including those used in local area network (LAN) and storage area network (SAN). In any case GFP adds a very low overhead to increase the efficiency of the optical layer.
Currently, two modes of client signal adaptation are defined for GFP:
- Frame-Mapped GFP (GFP-F), a layer 2 encapsulation PDU-oriented adaptation mode. It is optimized for data packet protocols (e.g. Ethernet, PPP, DVB) that are encapsulated onto variable size frames.
- Transparent GFP (GFP-T), a layer 1 encapsulation or block-code oriented adaptation mode. It is optimized for protocols using 8B/10B physical layer (e.g. Fiber Channel, ESCON, 1000BASE-T) that are encapsulated onto constant size frames.
GFP could be seen as a method to deploy metropolitan networks, and simultaneously to support mainframes and server storage protocols.
Data packet aggregation using GFP. Packets are in queues waiting to be mapped onto a TDM channel. At the far-end packets are drop again to a queue and delivered. GFP frame multiplexing and sub multiplexing. The figure shows the encapsulation mechanism and the transport of the GFP frames into VC containers embedded in the STM frames
Figure GFP frame formats and protocols
Framed-mapped GFP
In Frame-mapped GFP (GFP-F) one complete client packet is entirely mapped into one GFP frame. Idle packets are not transmitted resulting in more efficient transport. However, specific mechanisms are required to transport each type of protocol .
Figure GFP mapping clients format
GFP-F can be used for Ethernet, PPP/IP and HDLC-like protocols where efficiency and flexibility are important. To perform the encapsulation process it is necessary to receive the complete client packet, but this procedure increases the latency, making GFP-F inappropriate for time-sensitive protocols.
Transparent GFP,GFP-T
Transparent GFP (GFP-T) is a protocol-independent encapsulation method in which all client code words are decoded and mapped into fixed-length GFP frames The frames are transmitted immediately without waiting for the entire client data packet to be received. Therefore it is also a Layer 1 transport mechanism because all the client characters are moved to the far-end independently it does not matter if they are information, headers, control or any kind of overhead.
GFP-T can adapt multiple protocols using the same hardware as long as they are based on 8B/10B line coding. This line codes are transcoded to 64B/65B and then encapsulated into fixed size GFP-T frames. Everything is transported, including inter-frame gaps that can have flow control characters or any additional information.
GFP-T is very good for isocronic or delay sensitive-protocols, and also for Storage Area Networks (SAN) such as ECON or FICON. This is because it is not necessary to process client frames or to wait for arrival of the complete frame. This advantage is counteracted by lost efficiency, because the source MSPP node still generates traffic when no data is being received from the client. 
GFP enables MSPP nodes to offer both TDM and packet-oriented services, managing transmission priorities and discard eligibility. GFP replaces legacy mappings, most of them of proprietary nature. In principal GFP is just an encapsulation procedure but robust and standardised for the transport of packetised data on SDH and OTN as well.
GFP uses a HEC-based delineation technique similar to ATM, it therefore does not need bit or byte stuffing. The frame size can be easily set up to a constant length.
When using GFP-F mode there is an optional GPF extension Header (eHEC) to be used by each specific protocol such us source/destination address, port numbers, class of service, etc. Among the EXI types – ‘linear’ supports submultiplexing onto a single path, ‘Channel ID’ (CID) enables sub-multiplexing over one VC channel GFP-F mode.