Method by backboard transmission data
Affiliated technical field:
The invention belongs to the data network communications field, relate in particular to a kind of method by backboard transmission data.
Background technology:
Reliability and bandwidth are most important in the network design, if switching card and interface card can be handled the 10Gbps data flow, backboard also must be accomplished, otherwise it can become bottleneck, along with the increase of microprocessor bus width, the raising of system clock, parallel bus structure begins to influence the performance of system design, wide parallel bus bring such as problems such as deviation of signal, clock synchronization, electromagnetic interference and noise, pcb board route bus difficulty, all make it not too be applicable to network application.
The serial backplane technology can be avoided the parallel transmission problem, but it has utilized the narrow transfer of data of carrying out point-to-point connected in series at a high speed.As " adopting the method for back board high-speed serial line transmitting multiplex circuit data " (CN1384692A) and " a kind of method " based on the backboard transmitting multiplex circuit data (CN1406099A) method of backboard transmission data all disclosed.
As shown in Figure 1, the network equipment by backboard transmission data is made of following several parts usually: there are the lot of data exchange in backboard, power card, main control card, some interface cards and switching card etc. between switching card and the interface card.
Existing backboard transmission technology generally adopts the 8B/10B encoding and decoding technique.The 8B/10B coding is that a kind of 8 bit data bytes are compiled is the method for 10 transmission characters, and this encoding scheme is by IBM Corporation's exploitation and have patent right.Use the serial technology of 8B/10B coding to improve total transmission, walked around the parallel transmission problem, as shown in Figure 2.But it needs extra 25% the expense that increases, and promptly Chuan Shu efficient only is 80%, and when the stream of a 10Gbps bandwidth needed the backboard transmission, backboard must possess the total bandwidth of 12.5Gbps so, and efficient is too low.
GFP is a kind of generic encapsulation technology, and the technology contents of GFP is disclosed in by in the document that T1X1.5 issued " T1X1.5/2000-209GP (GFP) standard " as one of technical committee of the U.S. T1 committee.The SDH technology is also very ripe simultaneously.
Summary of the invention:
The object of the present invention is to provide a kind of GFP of utilization (interchangeable frame encapsulation process, Generic FramingProcedure) and SDH technology to realize the method for backboard transmission data, improved the efficiency of transmission in the transmission of serial backboard.
Method by backboard transmission data of the present invention, its step comprises:
When interface card during to switching card transmission data,
1) at first message is carried out the GFP encapsulation, formed the continuous data flow of a GFP encapsulation format;
2) this data flow is mapped in the virtual container of SDH, forms the SDH frame, enter backboard and transfer to switching card;
3) when switching card receives, at first to the SDH frame (by the CDR of receiver, Clock/Data Recovery, module) carries out bit aligned, from the SDH frame, extract clock and data, and then separate mapping block by SDH and carry out byte-aligned and the alignment of SDH frame, remove the expense of SDH, export the data flow of continuous GFP encapsulation format;
4) data flow of above-mentioned GFP encapsulation format is carried out decapsulation, the reduction message;
When switching card during to interface card transmission data, carry out the GFP encapsulation at switching card in the manner described above, carry out the GFP decapsulation at interface card.
Described GFP is encapsulated as:
Add the gfp frame head, the gfp frame head comprises the PLI territory, the length of expression message, and the CRC check to PLI did is represented in the HEC territory; Respectively frame head and net load are carried out scrambler then;
When channel idle, insert the empty frame of GFP;
Described GRP is descapsulated into:
At first be in search condition, ceaselessly 4 continuous bytes done the HEC verification,, promptly enter presynchronization state in case find a correct HEC;
At presynchronization state, the frame head of each frame is done the HEC verification, utilize the length information of frame head to find postamble, behind the correct frame of the HEC verification that occurs predetermined number continuously, enter synchronous regime; Promptly enter search condition as long as the frame of a HEC check errors occurs;
In synchronous regime, continue each frame is carried out verification, if verification is correct, then remain on synchronous regime, at this moment, remove the gfp frame head, carry out descrambling code again, restore Frame; When the frame that a HEC check errors occurs promptly enters search condition, restart search.
The empty frame of GFP is dropped in the GFP decapsulation process.
Described HEC territory is the CRC-16 verification that PLI is done, and gfp frame head part expense takies 4 bytes.
Abundant, ingenious ripe GFP technology and the SDH technology used of the present invention, under transmission means of the present invention, the expense of GFP encapsulation is fixed as 4 bytes, and the efficiency of transmission of SDH is about 96.3%, and whole efficiency of transmission is
Wherein n is the length of message.When transmission Ethernet message, if the length of message is 64 bytes, then calculating efficiency of transmission according to above formula is 90.6%, if the length of message is 1518 bytes, then efficiency of transmission is 96.0%, and in the practical application, the typical length of message is about 500 bytes, then efficiency of transmission is 95.5%, is higher than the efficiency of transmission of 8B/10B far away.
Description of drawings:
What Fig. 1 represented is the block diagram of a network equipment
What Fig. 2 represented is to adopt the 8B/10B encoding and decoding technique to realize the block diagram of backboard transmission
What Fig. 3 represented is to carry out SDH after the GFP encapsulation to shine upon the block diagram of realizing the backboard transmission
What Fig. 4 represented is the form of the GFP encapsulation of simplification
The content of the empty frame of the GFP that Fig. 5 represents
What Fig. 6 represented is the process of GFP decapsulation
Embodiment:
The invention will be further described below in conjunction with drawings and Examples:
When interface card during to switching card transmission data, at first message is carried out the GFP encapsulation, encapsulation process is as follows: add the gfp frame head, the form of frame head as shown in Figure 4, PLI represents the length of message, and HEC is the CRC-16 verification that PLI is done, and this part expense takies 4 bytes altogether; Respectively frame head and net load are carried out scrambler; When channel idle, insert the empty frame of GFP, the empty frame format of the GFP before the scrambler is as shown in Figure 5.So just formed the continuous data flow of a GFP encapsulation format, this data flow has been mapped in the virtual container of SDH, formed the SDH frame, entered backboard and transfer to switching card.
When switching card receives, at first the SDH frame is carried out bit aligned by CDR (Clock/Data Recovery) module of receiver, from the SDH frame, extract clock and data, and then separate mapping block by SDH and carry out byte-aligned and the alignment of SDH frame, remove the expense of SDH, export the data flow of continuous GFP encapsulation format.
Carry out the GFP decapsulation again, as shown in Figure 6:
At first be in search condition, ceaselessly 4 continuous bytes done the HEC verification,, promptly enter presynchronization state in case find a correct HEC;
At presynchronization state, the frame head of each frame is done the HEC verification, utilize the length information of frame head to find postamble, (the delta parameter can be established, synchronous again robustness when the value of delta has determined decapsulation when occurring delta continuously, when the delta value obtains greatly, so again synchronously need the long time, but it is less pseudo-synchronous probability to occur, when the delta value obtains hour, then very fast synchronously again, increase but pseudo-synchronous probability occurs.The delta value of recommending is 1, sees " T1X1.5/2000-209GP (GFP) standard ".) behind the correct frame of individual HEC verification, enter synchronous regime, promptly enter search condition as long as the frame of a HEC check errors occurs;
In synchronous regime, continue each frame is carried out verification, if verification is correct, then remain on synchronous regime, at this moment, remove the gfp frame head, carry out descrambling code again, restore Frame.When the frame that a HEC check errors occurs promptly enters search condition, restart search.
Empty frame is dropped in decapsulation process.
Like this, just realized the transfer of data of interface card to switching card, same, when switching card during,, on interface card, carry out decapsulation and get final product as long as encapsulate at switching card to interface card transmission data.
Under this transmission means, the expense of GFP encapsulation is fixed as 4 bytes, and the efficiency of transmission of SDH is about 96.3%, and whole efficiency of transmission is
Wherein n is the length of message.When transmission Ethernet message, if the length of message is 64 bytes, then calculating efficiency of transmission according to above formula is 90.6%, if the length of message is 1518 bytes, then efficiency of transmission is 96.0%, and in the practical application, the typical length of message is about 500 bytes, then efficiency of transmission is 95.5%, is higher than the efficiency of transmission of 8B/10B far away.