WO2009103225A1

WO2009103225A1 - A packet forwarding method and equipment

Info

Publication number: WO2009103225A1
Application number: PCT/CN2009/070378
Authority: WO
Inventors: 张�浩; 龙志平; 田向远; 李维; 顾晓浩; 邵建树
Original assignee: 华为技术有限公司
Priority date: 2008-02-18
Filing date: 2009-02-06
Publication date: 2009-08-27
Also published as: CN101247337B; CN101247337A

Abstract

A packet forwarding method and equipment belonging to the communication field is disclosed. The method comprises: receiving a packet and acquiring a key field of the packet; according to the key field, searching whether there is a flow table corresponding to the key field in a storage space of a first type, if yes, forwarding the packet according to the flow table; otherwise, according to the key field, searching whether there is a flow table corresponding to the key field in a storage space of a second type, if yes, forwarding the packet according to the flow table which is found in the storage space of the second type. The equipment comprises a receiving and acquiring module and a searching and forwarding module. By adopting a flow table mechanism that combines the multilevel hash table with the TCAM, the fast-forwarding of packet, the balance between the storage volume of packets and the forwarding speed as well as the fast convergence of flow table are achieved, the packet storage collision is reduced, the request for capability and speed of the network operator is satisfied and the cost is saved.

Description

Method and device for forwarding messages

The present invention relates to the field of communications, and in particular, to a packet forwarding method and device. Background technique

With the continuous development of IP (Internet Protocol) technology, IP networks carry a single Internet data service to carry voice, data, video, large customer lines, 3G (3rd Generation, third generation digital Communication, NGN (Next/New Generation Network), Next/New Generation Network, etc. Because in the past, the IP network mainly carried the data services of the ordinary Internet, such as sending and receiving Emai l, viewing the web page, and so on. Therefore, the bandwidth requirement of the network is not high. Generally, the access speed of 56Kpbs can meet the demand. But with the transformation of the IP network, it needs to serve a variety of new services, such as online games, P2P (Peer-to-Peer, peer-to-peer technology), video, voice, etc. So, in the process of transformation, to IP The network puts forward higher requirements in terms of forwarding performance, capacity, security, and service quality.

At present, the forwarding rate of the backbone network of the IP network has reached 40G or even 100G. The forwarding rate of the aggregation layer and the access layer has generally reached 10G. Usually, the network processor and the multi-core processor are used to implement high-speed forwarding. How to use the network The fast forwarding of large-capacity messages in a processor or a multi-core processor is an urgent problem to be solved.

Currently, the fast forwarding of the packet is performed according to the corresponding key segment of the packet, and the packet is forwarded by the key segment of the packet, such as the MAC (Media Access Control) address of the packet. , the IP address or the quintuple of the packet (for example, the source IP address of the packet, the destination IP address, the protocol number, the source port number, the destination port number, etc.), the seven-tuple (for example, the source of the packet) IP address, destination IP address, protocol number, source port number, destination port number, service type, input interface composition, etc. Correspondingly, in a forwarding device such as a firewall, a deep packet parsing, and a session border controller, a flow table formed by combining message-based key fields (ie, a session table) and an action table composed of corresponding parameters and actions are used. Partial flow table mechanism (in which the flow table and the action table are one-to-one correspondences, according to different implementation manners, the flow table and the action table may be unified together, and the flow table and the action table may be separately stored) Implement fast forwarding of packets. However, the forwarding device needs to be able to provide a large-capacity storage space to store a large number of flow tables consisting of key fields of the message, and at the same time, the forwarding device needs to provide high-speed access and forwarding speed. At present, the storage devices in the forwarding device have different degrees of defects in use: storage devices with large capacity and low price, such as SRAM (Static Random-Access Memory), DRAM (Dynamic Random-Access Memory, dynamic random access memory) ), etc., the access speed is often low; and the access speed of the storage device such as TCAM (Triple Content Addressable Memory) The capacity is often small and the price is high.

At the same time, in the case of message storage, it is also a question that must be considered to store the flow tables composed of the key fields of the message in what kind of organization. If a large number of flow tables are directly stored in the form of a direct table or a linked list, a large amount of storage space is required, and it is difficult to provide high-speed access speed. At present, the main method is to store some key fields in the flow table in a direct table or a linked list, thereby achieving a balance between storage space and access speed, and thus it is possible that some of the key segments of the different messages are the same. This causes conflicts during storage. The use of the tree to store these flow tables in the longest match or exact match can solve the above conflicts, but this method is more complicated and the performance is not high.

In the existing flow table implementation mechanism, the flow table is generally stored in the form of a tree or a linked list. In the implementation manner of the tree, the key segment (such as a quintuple) of the packet is matched by the longest or Complete matching and other methods to establish and find a flow table; In the implementation of the linked list, a multi-level linked list is used, and each level of the lookup table contains multiple auxiliary linked lists to establish and find a flow table. Although these two methods can solve the conflict problem, they are only applicable to the occasions where the access speed and capacity requirements are not high. However, as the requirements for the forwarding speed of packets are increasing, for example, it is generally required to achieve 10G forwarding. In the access layer and the aggregation layer of the speed, a large number of service packets require a large amount of storage space and high-speed forwarding. If the flow table is stored in the form of a tree or a linked list, the depth of the tree becomes deeper and deeper, and the list level is The number is getting larger and larger, which makes the access speed slower and slower, and it is difficult to achieve fast forwarding of packets. Summary of the invention

The present invention provides a method and a device for forwarding a packet, in order to implement the fast forwarding of the packet and the fast convergence of the flow table. The technical solution is as follows:

In one aspect, the embodiment of the present invention provides a method for packet forwarding, where the method includes:

Receiving a message, and obtaining a key field of the message;

Determining, according to the key field, a flow table corresponding to the key field in the first type of storage space, and if yes, forwarding the message according to the flow table; otherwise, according to the key field, Finding whether the second type of storage space has a flow table corresponding to the key field; if yes, forwarding the message according to the flow table found in the second type of storage space.

On the other hand, an embodiment of the present invention provides a packet forwarding device, where the device includes:

a receiving and acquiring module, configured to receive a packet, and obtain a key segment of the packet;

a search and forwarding module, configured to search, according to the key field, a flow table corresponding to the key segment obtained by the receiving and acquiring module in the first type of storage space, and if yes, forwarding according to the flow table The packet; otherwise, according to the keyword segment, searching whether the second type of storage space has a flow table corresponding to the key segment; if yes, the root And forwarding the packet according to the flow table found in the second type of storage space.

The beneficial effects of the technical solutions provided by the embodiments of the present invention are:

By using a multi-level storage space (the first type of storage space and the second type of storage space), the flow table mechanism that combines the multi-level hash table and the TCAM can implement fast forwarding of packets and implement storage of packets in the device. Balance between capacity and forwarding speed, reduce packet storage conflicts, achieve fast convergence of flow tables, and thus meet the network operator's requirements for capacity and speed, save costs, and facilitate more value-added business. DRAWINGS

1 is a schematic flowchart of a method for forwarding a packet according to Embodiment 2 of the present invention;

2 is a schematic flowchart of establishing a flow table according to Embodiment 2 of the present invention;

3 is a schematic flowchart of a method for forwarding a packet according to Embodiment 3 of the present invention;

4 is a schematic diagram of a device for forwarding a packet according to Embodiment 4 of the present invention;

FIG. 5 is a schematic diagram of a device for forwarding a packet according to Embodiment 5 of the present invention; FIG.

6 is a schematic diagram of a device for forwarding a packet according to Embodiment 6 of the present invention;

FIG. 7 is a schematic diagram of a device for forwarding a packet according to Embodiment 7 of the present invention. detailed description

The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Example 1

The solution provided by the embodiment of the present invention uses a forwarding flow table mechanism combining a multi-level hash table and a TCAM, and uses the N+1-level storage space, and the front N-stage uses a low-cost storage space (such as DRAM) to utilize the message. The key segment uses a hash algorithm to store the flow table, and the last level uses the TCAM storage flow table. If the packet is stored in the flow table, if the packet storage conflict exists in the previous N-level, the flow table is created by using the last-level TCAM storage packet to implement fast convergence. Among them, the low-cost storage space used in the previous N-level is called the first-class storage space, and the last-level TCAM is called the second-type storage space, or high-speed storage space. Methods include:

After receiving the packet, the key segment of the packet is obtained. Then, according to the keyword segment, it is found whether there is a flow table corresponding to the key segment in the first type of storage space, and if so, according to the storage space in the first type. The found flow table forwards the packet; otherwise, according to the key field, it finds whether there is a flow table corresponding to the key segment in the second type of storage space; if yes, according to the flow table found in the second type storage space Forward the message. When the first type of storage space is a multi-level storage space, the storage space of each storage space of the first type of storage space needs to be searched in turn, and the packet is forwarded according to the hit flow table. The specific content is as follows:

Obtaining a flow table of the current storage space of the first type of storage space. When the flow table corresponding to the key field is found, the packet is forwarded according to the flow table, otherwise, the next storage space of the current storage space is searched; When no flow table corresponding to the key field is found in the storage space of each type of storage space, the second type storage space is searched for whether there is a flow table corresponding to the key field, and then according to the second type storage space. The hit flow table forwards the message.

Further, when the flow table of the current storage space of the first type of storage space is obtained, the method can be implemented as follows: First, the key field of the packet is hashed, and the key segment of the packet is calculated. The hash value is obtained by the size of the current storage space; the hash value is matched with the mask to obtain an index;

Then, calculating a product value of the index field of the index and the message; obtaining an address of the current storage space according to the sum of the product value and the base address of the current storage space;

Finally, the flow table of the current storage space is obtained according to the address of the current storage space.

Further, when the flow table corresponding to the key field is found, and the flow table is forwarded according to the flow table, it is also required to determine whether the flow table is idle. If the current storage space flow table is not idle, the current storage space needs to be searched. If the flow table corresponding to the key field is found in the flow table of the current storage space, the flow table is forwarded according to the searched flow table; if the current storage space flow table is idle, the next storage space is searched. Storage space.

In summary, the embodiment of the present invention implements a step-by-step search in the first type of storage space according to the keyword segment of the packet, and if the first type of storage space fails to be searched, Quickly find in the second type of storage space. When the flow table is hit, the packet can be quickly forwarded.

According to the technical solution provided by the embodiment of the present invention, when there is no flow table corresponding to the key segment in the storage space of the second type, the packet cannot be hit in the flow table of the first type storage space and the second type storage space, The corresponding flow table of the message has not been established yet, so a flow table needs to be established for the message. The process of establishing a flow table for the packet is as follows:

First, the hash value of the key field of the message is obtained; the size of the current storage space of the first type of storage space is obtained; the hash value is matched with the mask to obtain the index; and the product of the index and the key field is calculated. Value

Then, according to the product value and the base address of the current storage space, obtain an address in the current storage space (the address in the current storage space corresponds to a certain flow table in the storage space), and determine whether the flow table corresponding to the address is idle. If yes, the current storage space address is a flow table for the packet; otherwise, the next storage space of the current storage space is searched; when the flow table corresponding to the address in each storage space of the first type of storage space is non- In the idle state, a flow table is created for the packet in the second type of storage space.

When the packet flow table is forwarded, the idle information of the flow table corresponding to the address in the storage space of the first type of storage space may be recorded; accordingly, the flow table is configured for the packet. The process is as follows: First, obtain a hash value of the key field of the message;

Then, by searching for the idle information of the flow table corresponding to the address in the storage space of the first type of storage space of the record, the address of the idle storage space appears for the first time in the storage space of each level, and the flow table is established for the packet; When the flow table corresponding to the address in the storage space of the first type of storage space is in the non-idle state, the flow table is created in the second type of storage space for the packet.

Through the foregoing method, the embodiment of the present invention implements a flow table for a message. It is important to note that when you create a table for a packet, you need to determine whether the packet with the same hash value is being locked. If yes, wait for the packet with the same hash value to be unlocked. Perform the corresponding action. If no, the message is locked.

Correspondingly, after the flow table is created for the packet, the locked packet is unlocked.

After the packet waiting for the same hash value is unlocked, the corresponding action is performed on the packet, which is specifically:

According to the hash value of the key segment of the obtained packet, it is determined whether the packet and the packet with the same hash value belong to the same type. If yes, the packet waiting for the same hash value is unlocked according to the same hash value. The flow table created by the packet is used to forward the packet. Otherwise, the packet is sent to the packet with the same hash value.

The packet forwarding method provided by the embodiment of the present invention achieves a balance between the capacity of the stored message and the forwarding speed by using a forwarding mechanism combining the storage space of the first type and the storage space of the second type, and reduces the balance between the storage capacity and the forwarding speed of the device. The conflicts in packet storage enable fast convergence of the flow table, which satisfies the requirements of the network operator for capacity and speed, saves costs, and facilitates more value-added services.

For the scheme of packet forwarding provided in this embodiment, a detailed description will be given below for the search and construction of the table in various embodiments. Example 2

Referring to FIG. 1, an embodiment of the present invention provides a method for forwarding a message. This embodiment is described by taking an N+1 level storage space as an example. The first N level is provided by a common storage device such as SRAM\DRAM. The storage space, the N+1 level is a high-speed storage space provided by a storage device such as a TCAM, and the method provided by the embodiment of the present invention uses a hash value obtained by using a hash algorithm according to a key segment of the packet during packet forwarding. The first N-level storage space is searched step by step; if the first N-level storage space fails to find, the fast search is performed directly in the TCAM. The specific contents include the following:

When the device forwards the received packet, it needs to find the flow table corresponding to the packet in the storage space of the device. After the flow table is found, the related action table information is obtained. Forward, where the specific content is as follows:

101: Receive a forwarding packet, obtain a key segment of the packet, and adopt a hash algorithm on the key segment to obtain a hash value. The key field of the packet in this embodiment is described by using a common five-tuple (source IP address, destination IP address, protocol number, source port number, and destination port number) as an example.

102: Obtain a mask according to the size of the storage space of the z'th (l≤z'≤N) level, and obtain the mask and

The hash value obtained in 101 is phase-and-operated to obtain an index; according to the base address (starting address) of the first-level storage space, the quintuple size of the message, and the index, a first level is obtained. The address of the storage space.

When a mask is obtained according to the size of the storage space, the mask is all 1; usually, in order to avoid conflicts, the larger the storage space, the more the number of bits corresponding to the mask. When obtaining the address of the storage space, the calculation method is usually as follows:

Address of storage space = base address of storage space 15-tuple size X index;

103: Determine, according to the address of the obtained storage space, whether the flow table is idle by checking the address of the corresponding flow table, and if yes, execute 105; otherwise, execute 104.

For example, if the flag of the flow table corresponding to the address is “0”, it indicates that the flow table is idle; if the flag of the flow table corresponding to the address is “1”, it indicates that the flow table is not idle.

104: Determine whether the quintuple of the packet matches the quintuple stored in the flow table corresponding to the storage space of the level, and if yes, the packet hits the flow table, and executes 108; otherwise, the packet is When the level storage space performs a flow table lookup, a conflict occurs, and execution 105 is performed.

105: Determine whether the Nth level has been found, if yes, execute 107; otherwise, go to step 106.

106: i = i + l , go to step 102.

107: The quintuple that obtains the message performs a quick lookup in the TCAM to determine whether to hit the corresponding flow table, and if so, executes 108; otherwise, executes 109.

108: Find the related information of the packet, perform fast forwarding of the packet, and end.

In this step, the related information of the packet is searched for, and the corresponding relationship between the flow table corresponding to the packet and the action table is searched for, and the flow table and the action table are combined to complete the fast forwarding of the packet.

109: The packet is the first packet, and a flow table is created for the packet.

In this embodiment, in the N+1 level storage space, the packet that can hit the flow table is called a subsequent packet, and after the subsequent packet obtains the related information, the read action table performs related actions, and the packet can be quickly forwarded. In the N+1 storage space, there is no packet that hits the flow table, which is called the first packet. For the first packet, the flow table needs to be established. At the same time, the first packet needs to be created, and the related actions are stored in the action table to establish and maintain the action. Correspondence between the table and the flow table, so that when the same message is received in the future, the message is processed according to the action in the action table after the hit flow table finds the correspondence between the flow table and the action table.

The following is a detailed description of how to implement the flow table when the message is the first package: The establishment of a flow table means that when a device receives a certain packet for the first time, since the flow table of the packet is not recorded in the storage space of the device, a flow table needs to be established for the packet, so that After receiving the same packet, you can hit the flow table to implement fast forwarding. When the device receives the packet and finds the free storage space, it inserts the corresponding content of the packet into the free storage space. If there is no free storage space in the storage space of the first N levels, the corresponding content of the message is directly stored in the T+1 of the N+1th level, thereby achieving fast convergence.

Generally, since the packet is bidirectional (forward packet and reverse packet), when the flow table is established, correspondingly, a bidirectional flow table, that is, a forward flow table and a reverse flow table, For example, the forward flow table is as follows:

201: Acquire a key field of the message, and adopt a hash algorithm on the key field to obtain a hash value.

The process of the step 201 is similar to the foregoing step 101 and will not be described again.

The key field of the packet in this embodiment is described by using a common five-tuple (source IP address, destination IP address, protocol number, source port number, and destination port number) as an example. Correspondingly, the difference between the forward flow table and the reverse flow table is that the source IP address of the forward packet is the destination IP address of the reverse packet, and the source port number is the destination port number of the reverse packet. Correspondingly, the destination IP address of the forward packet is the source IP address of the reverse packet and the destination port number is the source port number of the reverse packet.

202: To ensure that the flow table is created without errors, perform the lock action.

The purpose of the locking is to ensure that only the atomic operation can be performed when the flow table is created for the first packet (that is, at the time of establishing the flow table, only one message with the same hash value is reported at this time) The file is configured to prevent the hash table having the same hash value as the first packet of the target packet when the flow table is created for the first packet (for example, belonging to the same data packet stream, having the same The second packet of the key segment, or during the time period in which the flow table is created for the first packet, the first packet in the data packet stream from other classes appears, but the first packet is obtained. The hash value is the same as the hash value of the first packet of the target. At this time, the conflict of the packet establishment process occurs. In order to avoid this type of packet conflict, this step needs to be locked. That is, only the atomic operation can be performed when the flow table is established, and the flow table is prevented from being in error. That is, the following operations are performed only for the packet, ensuring that only one packet with the same hash value is constructed at this time. Table processing,

203: Obtain a mask according to the size of the storage space of the zth (l≤z'≤N) level, and perform the AND operation on the obtained hash value with the hash value obtained in 201, thereby obtaining an index; According to the base address of the storage space of the zth stage, the size of the quintuple of the message and the calculated index are calculated to obtain an address of the z-th storage space.

204: Obtain the content of the flow table corresponding to the address according to the obtained address, and determine, according to the flag of the flow table, whether the flow table corresponding to the obtained address is idle, and if yes, execute 205; otherwise, execute 206.

205: Store the corresponding content of the message in the location corresponding to the address to establish a flow table, and set a corresponding flag bit, indicating that the flow table of the location is occupied at this time, and then execute 209. 206: Determine whether z is equal to N at this time, that is, determine whether the Nth level has been found, if yes, execute 207; otherwise, execute 208.

207: It is indicated that in the storage space of the first N levels, there is no free storage space for storing the corresponding content of the packet (that is, a conflict occurs in the storage space of the first N level to establish the packet flow table), The corresponding content of the message is stored in the TCAM of the N+1th level, and the flow table is established in the TCAM, and the corresponding flag bit is set, thereby achieving fast convergence.

208: After z' = z' + l, return to step 203.

209: After the flow table is successfully established, the unlocking is performed, and the forward flow table is successfully established.

The above example takes the establishment of a forward flow table as an example. When a reverse flow table is established, the method is similar and will not be described again.

In summary, the method for searching and establishing a flow table to implement fast forwarding of a packet is provided in the embodiment of the present invention. When searching, a hash algorithm is used according to a key segment of the packet to obtain a key value in the first N level. Find. If the first N-level storage space fails, it is directly searched in the TCAM. When the table is built, the hash value is used according to the key segment of the packet to obtain the key value. The free space is searched in the first N level. If it is idle, it is directly inserted. If there is a conflict, it will be searched step by step. If the previous N-level hash storage spaces conflict, it will be directly stored in the TCAM.

It is important to note that during the flow table creation process, it is necessary to ensure that only one message with the same hash value is processed at a time. As described above, for example: When the flow table of the first packet is created, if a pseudo-first packet belonging to the same type of data packet (that is, a packet having the same key segment) is present, the flow table needs to be established. For example, in the process of establishing the flow table in the first packet A in the data stream, the flow table corresponding to the data flow message has not been established, so that the message in the data flow cannot hit the corresponding flow table, and There is a need to establish a flow table. At this time, the message is said to be a pseudo first packet. For this case, since the first packet is locked, the pseudo first packet B is in a waiting state, waiting for the first packet A to be established. After the flow table is unlocked, the pseudo first packet B can be forwarded according to the flow table established by the first package A.

It should also be noted that when processing the first packet of different types of data packets having the same hash value, if another type of data packet is created when the flow table of the first packet of the first type packet is created, The first packet of the hashed packet needs to be created. The hash value obtained by the first packet of the two types of packets is the same, and the first packet of the two types of packets is processed. A conflict has occurred, causing a mistake in building a table. In this embodiment, in order to avoid conflicts in the flow table, the locking method is adopted, that is, when the flow table is created for the first packet of the first type of packet, the locking is performed. When the first packet is successfully created, after the unlocking, Then, the first packet of the second type of packet is constructed, thereby ensuring that although the packets have the same hash value, the flow tables are established at different flow table locations corresponding to different addresses, thereby ensuring the construction of the table. accuracy.

The embodiment of the invention adopts a combination of multi-level hash table storage and TCAM direct storage in a common storage space, which reduces the collision probability of the storage flow table. In the N+1 storage space mode adopted by the embodiment of the present invention, the first N level storage space can also dynamically configure the number of stages according to the flow table specification requirements and performance requirements, and the last stage adopts TCAM to achieve fast convergence, thereby realizing cost specifications and The balance of performance is conducive to network operators to carry out a variety of rich types of business. Example 3

Referring to FIG. 3, an embodiment of the present invention provides a method for forwarding a packet. This embodiment is described by taking an N+1 storage space as an example. The first N level is provided for a common storage device such as SRAM\DRAM. The storage space, the N+1 level is a high-speed storage space provided by a storage device such as a TCAM, and the method provided by the embodiment of the present invention uses a hash algorithm to obtain a key value according to a key segment of the packet when the packet is forwarded. The N-level storage space is searched step by step. If the first N-level storage space fails to be searched, the fast search is performed directly in the TCAM, and the information about the idle flow table corresponding to the address of the storage space recorded in the process of searching is obtained. When the packet is the first packet, the packet can be directly stored in the idle flow table of the corresponding storage space by using the recorded information. The specific contents include the following:

301: Receive a forwarding packet, obtain a key segment of the packet, and adopt a hash algorithm on the key segment to obtain a hash value.

The key field of the packet in this embodiment is described by using a common five-tuple (source IP address, destination IP address, protocol number, source port number, and destination port number) as an example.

302: Obtain a mask according to the size of the storage space of the z'th (l≤z'≤N) level, and perform the AND operation on the obtained mask and the hash value obtained in 301, thereby obtaining an index. And obtaining an address of the first-level storage space according to the base address (starting address) of the z-th storage space, the quintuple size of the message, and the index.

Address of storage space = base address of storage space 15-tuple size X index;

303: Determine, according to the address of the obtained storage space, whether the flow table is idle according to the address of the obtained storage space, and if yes, execute 305; otherwise, execute 304.

304: Determine whether the quintuple of the packet matches the quintuple stored in the flow table corresponding to the storage space of the level, and if yes, the packet hits the flow table, and executes 309; otherwise, the packet is When the level storage space performs a flow table lookup, a conflict occurs, and execution 306 is performed.

305: After recording the information about the idle flow table, execute 306.

306: Determine whether it is already searching for the Nth level, if yes, execute 307; otherwise, go to step 308.

307: The quintuple that obtains the message performs a fast lookup in the TCAM to determine whether to hit the corresponding flow table, and if so, executes 309; otherwise, executes 310.

308: i = i + l , go to step 302.

309: Find the related information of the packet, perform fast forwarding of the packet, and end.

In this step, the related information of the packet is searched, mainly to find the corresponding flow table of the packet and the corresponding relationship of the action table. Therefore, the flow table and the action table are combined to complete the fast forwarding of the message.

310: The packet is the first packet, and a flow table needs to be established for the packet; the related information of the idle flow table in the TCAM is recorded, and then 311 is performed.

311: Perform the lock action.

312: The message is stored in the corresponding storage space according to the related information of the recorded idle flow table.

For example, in the search process, the information about the idle flow table recorded is specifically the identifier of the storage space, the identifier of the address of the storage space, and the idle state. See Table 1, which provides a schematic table for recording relevant information.

Table 1

As shown in Table 1, the message is stored in the first free storage space (that is, the flow table is established at the address of the second-level storage space B), and the current N-level storage space is In the non-idle state, a flow table is created for the packet in the T+1 storage space of the N+1th level, thereby achieving fast convergence.

313: Perform an unlock action.

The embodiment of the present invention adopts a combination of multi-level hash table storage and TCAM direct storage in a common storage space, which reduces the collision probability of the storage flow table. Meanwhile, after receiving the message, the record is recorded in the process of searching the flow table. In the manner of the information about the idle flow table, when the packet is the first packet, the flow table is created for the packet by using the related information of the recorded idle flow table, thereby saving the time for establishing the flow table. In the N+1 storage space mode adopted by the embodiment of the present invention, the first N level storage space can also dynamically configure the number of stages according to the flow table specification requirements and performance requirements, and the last stage adopts TCAM to achieve fast convergence, thereby realizing cost specifications and The balance of performance is conducive to network operators to carry out a variety of rich types of business. Example 4

Referring to FIG. 4, an embodiment of the present invention provides a device for forwarding a packet, where the device includes:

The receiving and obtaining module 401 is configured to receive a packet, and obtain a key segment of the packet.

The search and forward module 402 is configured to search, according to the keyword segment, whether the flow table corresponding to the key segment obtained by the receiving and obtaining module 401 is in the first type of storage space, and if yes, forward the packet according to the flow table; otherwise, And searching, according to the keyword segment, whether the second type of storage space has a flow table corresponding to the key field; if yes, forwarding the packet according to the flow table found in the second type of storage space.

The device for forwarding a packet according to the embodiment of the present invention uses a multi-level storage space, which reduces the conflict of the storage flow table. Rate. The last level of the multi-level storage space is the second type of storage space (ie, high-speed storage space). The levels can also be dynamically configured according to the flow table specification requirements and performance requirements. The final stage uses high-speed storage space to achieve fast convergence. In order to achieve a balance between cost specifications and performance, it is beneficial for network operators to carry out various types of services. Example 5

Referring to FIG. 5, an embodiment of the present invention provides a device for forwarding a packet, where the device includes:

The receiving and obtaining module 501 is configured to receive a packet, and obtain a key segment of the packet.

The locating and forwarding module 502 is configured to: according to the keyword segment, search for a flow table corresponding to the key segment obtained by the receiving and obtaining module 501 in the first type of storage space, and if yes, forward the packet according to the flow table; otherwise, And searching, according to the keyword segment, whether the second type of storage space has a flow table corresponding to the key field; if yes, forwarding the packet according to the flow table found in the second type of storage space.

The lookup and forwarding module 502 specifically includes:

The obtaining unit 5021 is configured to obtain a flow table of a current storage space of the first type of storage space.

The lookup and forwarding unit 5022 is configured to: according to the flow table of the current storage space of the first type of storage space acquired by the obtaining unit 5021, when the flow table corresponding to the key segment is found, according to the found and the key segment The corresponding flow table forwards the packet. Otherwise, it searches for the next-level storage space of the current storage space. When the flow table corresponding to the key segment is not found in the storage space of the first-class storage space, the search is performed. The second type of storage space.

The obtaining unit 5021 specifically includes:

a calculation subunit, configured to calculate a hash value of the key segment; obtain a mask by the size of the current storage space; and obtain an index by summing the hash value and the mask; calculating a product value of the index and the key segment; according to the product value The sum of the base address of the storage space, and the address of the current storage space;

The obtaining subunit is configured to obtain a flow table for finding a current storage space according to an address of a current storage space obtained by the calculation subunit.

The device for forwarding a packet according to the embodiment of the present invention uses a multi-level storage space, which reduces the probability of collision of the storage flow table. The last level of the multi-level storage space is the second type of storage space (ie, high-speed storage space). The levels can also be dynamically configured according to the flow table specification requirements and performance requirements. The final stage uses high-speed storage space to achieve fast convergence. In order to achieve a balance between cost specifications and performance, it is beneficial for network operators to carry out various types of services. Example 6

Referring to FIG. 6, an embodiment of the present invention provides a device for forwarding a packet, where the device includes:

The receiving and obtaining module 601 is configured to receive a packet, and obtain a key segment of the packet. The locating and forwarding module 602 is configured to: according to the keyword segment, find, in the first type of storage space, a flow table corresponding to the key segment obtained by the receiving and obtaining module 601, and if yes, forward the packet according to the flow table; otherwise, And searching, according to the keyword segment, whether the second type of storage space has a flow table corresponding to the key field; if yes, forwarding the packet according to the flow table found in the second type of storage space.

The table construction module 603 is configured to: when the lookup and forwarding module 602 searches for the flow table that is not corresponding to the key field in the second type of storage space, the flow table is created for the message, where the table construction module 603 specifically includes:

An obtaining unit, configured to obtain a hash value of a key segment of the packet;

a processing unit, configured to obtain a mask according to a size of a current storage space of the first type of storage space according to a hash value obtained by the acquiring unit; and obtain an index by comparing the hash value with the mask; and calculate a product of the index and the key segment The value of the current storage space is obtained according to the product value and the base address of the current storage space, and the flow table corresponding to the address is determined to be idle. If yes, the current storage space address is a flow table for the packet; otherwise, the current search is performed. The next-level storage space of the storage space; when the flow table corresponding to the address in the storage space of the first-class storage space is a non-idle state, the flow table is established for the packet in the storage space of the second type.

Further, the table building module 603 further includes:

a judging unit, configured to determine, after the obtaining unit obtains a hash value of the key segment of the packet, whether the packet with the same hash value is being locked,

a first processing unit, configured to: when the judgment unit determines that the result is yes, the message waits for the message with the same hash value to be unlocked, and performs a corresponding action;

The locking unit is configured to lock the message when the judgment unit determines that the result is no;

Accordingly, the table building module 603 further includes:

The unlocking unit is configured to unlock the locked packet after the flow table is created for the packet.

The first processing unit specifically includes:

a processing subunit, configured to: according to the hash value of the key field of the packet obtained by the obtaining unit; determine whether the packet and the same hash value belong to the same type, and if yes, the message waits for the same hash value After the packet is unlocked, the packet is fast forwarded according to the flow table created by the same hash value. Otherwise, after the packet waits for the same hash value to be unlocked, the flow table is created for the packet.

The device for forwarding a packet according to the embodiment of the present invention adopts a multi-level storage space, which reduces the collision probability of the storage flow table. The last level of the multi-level storage space is the second type of storage space (ie, high-speed storage space). The levels can also be dynamically configured according to the flow table specification requirements and performance requirements. The final stage uses high-speed storage space to achieve fast convergence. In order to achieve a balance between cost specifications and performance, it is beneficial for network operators to carry out various types of services. Example 7

Referring to FIG. 7, an embodiment of the present invention provides a device for forwarding a packet, where the device includes:

The receiving and obtaining module 701 is configured to receive a packet, and obtain a key segment of the packet.

The locating and forwarding module 702 is configured to search, according to the keyword segment, whether the flow table corresponding to the key segment obtained by the receiving and obtaining module 701 is in the first type of storage space, and if yes, forward the packet according to the flow table; otherwise, And searching, according to the keyword segment, whether the second type of storage space has a flow table corresponding to the key field; if yes, forwarding the packet according to the flow table found in the second type of storage space.

The recording module 703 is configured to: when the lookup and forwarding module 702 searches and forwards, record idle information of the flow table corresponding to the address in each storage space of the first type of storage space;

The table construction module 704 is configured to: when the lookup and forwarding module 702 searches for the flow table that is not corresponding to the key field in the second type of storage space, the flow table is created for the message, where the table construction module 704 specifically includes:

a processing unit, configured to: according to the hash value obtained by the obtaining unit, according to the idle information recorded by the recording module 703, the address of the idle storage space appears for the first time in each storage space, and the flow table is established for the packet; The flow table corresponding to the address in the storage space of the space is a non-idle state, and the flow table is established for the packet in the second type of storage space.

Further, the table building module 704 further includes:

Accordingly, the build module 704 also includes:

The first processing unit specifically includes:

The device for forwarding a packet according to the embodiment of the present invention adopts a multi-level storage space, which reduces the collision probability of the storage flow table. The last level of the multi-level storage space is the second type of storage space (ie, high-speed storage space). The levels can also be dynamically configured according to the flow table specification requirements and performance requirements. The final stage uses high-speed storage space to achieve fast convergence. , From The balance between cost specifications and performance is realized, which is beneficial to network operators to carry out various types of services.

Some of the steps in the embodiment of the present invention may be implemented by software, and the corresponding software program may be stored in a readable storage medium such as an optical disk or a hard disk.

The above is only a specific embodiment of the present invention, and is not intended to limit the present invention, and any modifications, equivalents, and improvements made by those skilled in the art without departing from the principles of the present invention. And so on, should be included in the scope of protection of the present invention.

Claims

Claim

A method for forwarding a packet, the method comprising:

Receiving a message, and obtaining a key field of the message;

The packet forwarding method according to claim 1, wherein the first type of storage space is a multi-level storage space, and correspondingly, the first type of storage space is searched according to the key segment. Whether there is a flow table corresponding to the key field, and if yes, forwarding the message according to the flow table; otherwise, searching for the second type of storage space and the keyword according to the key segment The flow table corresponding to the segment, including:

Obtaining a flow table of the current storage space of the first type of storage space, and when the flow table corresponding to the key field is found, forwarding the packet according to the flow table, otherwise, searching for the current storage The storage space of the next level of space; when no flow table corresponding to the key field is found in each storage space of the first type of storage space, whether the storage space of the second type is found The flow table corresponding to the key field.

The method for forwarding a packet according to claim 2, wherein the obtaining a flow table of the current storage space of the first type of storage space comprises:

Calculating a hash value of the key field;

Obtain a mask by the size of the current storage space;

And obtaining an index by comparing the hash value with the mask;

Calculating a product value of the index and the key segment;

And obtaining an address of the current storage space according to the sum of the product value and the base address of the storage space; and acquiring a flow table of the current storage space according to the address of the current storage space.

The packet forwarding method according to claim 2, wherein, when the flow table corresponding to the key field is found, the message is forwarded according to the flow table, otherwise, Finding the next-level storage space of the current storage space includes:

Determining whether the flow table is idle, if the flow table is not idle, searching for the flow table, and when the flow table corresponding to the key field is found, forwarding the message according to the flow table; The flow table is idle, and the next-level storage space of the current storage space is searched.

5. The method of packet forwarding according to claim 2, wherein the method further comprises:

When the flow table corresponding to the key segment is not found in the second type of storage space, a flow table is created for the packet.

The method for forwarding a packet according to claim 5, wherein the establishing a flow table for the packet includes:

Obtaining a hash value of the key field of the packet; obtaining a mask by using a size of a current storage space of the first type of storage space; acquiring an index with the hash value and the mask; and calculating the index a product value of the key field; obtaining an address in the current storage space according to the product value and a base address of the current storage space, and determining whether the flow table corresponding to the address is idle, and if so, The address of the current storage space is a flow table for the packet; the second U is used to find the next-level storage space of the current storage space; and the addresses in the storage spaces of the first-class storage spaces are corresponding to each other. The flow table is in a non-idle state, and a flow table is created for the message in the second type of storage space.

The method for forwarding a packet according to claim 5, wherein the method further comprises: recording idle information of a flow table corresponding to an address in each storage space of the first type of storage space;

Correspondingly, the establishing a flow table for the packet includes:

Obtaining a hash value of the key field of the packet, and finding idle information in the storage space of each level by searching for the idle information of the flow table corresponding to the address in each storage space of the first type storage space of the record The address of the storage space is used to establish a flow table for the packet; when the flow table corresponding to the address in each storage space of the first type of storage space is a non-idle state, the second type of storage space is The message is created in a flow table.

The method for forwarding a packet according to claim 6 or 7, wherein after the obtaining the hash value of the key field of the packet, the method further includes:

Determining whether a packet having the same hash value is being locked. If yes, after waiting for the packet with the same hash value to be unlocked, performing a corresponding action on the packet, and if not, the packet is Locked.

The method of packet forwarding according to claim 8, wherein the method further comprises:

After the flow table is created for the packet, the packet is unlocked.

The method for forwarding a packet according to claim 8, wherein after the message waiting for the same hash value is unlocked, performing a corresponding action on the packet, specifically:

Determining, according to the hash value of the key segment of the packet, whether the packet and the packet with the same hash value belong to the same type, and if yes, the packet waits for the same hash value. After the packet is unlocked, the packet is quickly forwarded according to the flow table established by the packet with the same hash value; otherwise, after the packet waits for the packet with the same hash value to be unlocked, The packet is created with a flow table.

The packet forwarding method according to claim 4, wherein the determining whether the flow table is idle is specifically:

Whether the flow table is idle is determined by checking the identifier bit corresponding to the address of the current storage space.

The packet forwarding method according to claim 1 or 2, wherein the second type of storage space package Includes tri-state content addressable memory.

A packet forwarding device, the device comprising:

a search and forwarding module, configured to search, according to the key field, a flow table corresponding to the key segment obtained by the receiving and acquiring module in the first type of storage space, and if yes, forwarding according to the flow table If the message is found, according to the keyword segment, whether the second type of storage space has a flow table corresponding to the key field; and if so, according to the flow found in the second type of storage space. The table forwards the message.

The packet forwarding device according to claim 13, wherein the searching and forwarding module specifically includes: an obtaining unit, configured to acquire a flow table of a current storage space of the first type of storage space;

a lookup and forwarding unit, configured to: according to the flow table obtained by the obtaining unit, when the flow table corresponding to the key field is found, forwarding the message according to the flow table, otherwise, searching for the message The next-level storage space of the current storage space; when the flow table corresponding to the key segment is not found in the storage spaces of the first-class storage space, the second-type storage space is searched.

The packet forwarding device according to claim 14, wherein the obtaining unit specifically includes: a calculating subunit, configured to calculate a hash value of the key segment; and acquiring a mask by using a size of a current storage space a code; obtaining an index on the hash value and the mask; calculating a product value of the index and the key segment; obtaining a location according to a sum of the product value and a base address of the storage space The address of the current storage space;

And an obtaining sub-unit, configured to obtain, according to the address of the current storage space acquired by the computing sub-unit, a flow table that searches for the current storage space.

The packet forwarding device according to claim 14, wherein when the searching and forwarding module searches for the flow table corresponding to the key segment in the second type of storage space, the device further Including the construction of the module;

The building module specifically includes:

a processing unit, configured to acquire, according to a hash value obtained by the acquiring unit, a size of a current storage space of the first type of storage space; and obtain an index by using the hash value and the mask; Determining, by the product value, an address in the current storage space, determining whether the flow table corresponding to the address is idle, if And establishing, in the current storage space, a flow table for the packet; otherwise, searching for a next-level storage space of the current storage space; and corresponding to an address in each storage space of the first-type storage space The flow table is in a non-idle state, and a flow table is created for the message in the second type of storage space.

The packet forwarding device according to claim 14, wherein the device further comprises: a recording module, configured to record idle information of a flow table corresponding to an address in each storage space of the first type of storage space when the searching and forwarding module performs the searching and forwarding;

Correspondingly, the building module includes:

a processing unit, configured to: according to the hash value acquired by the acquiring unit, the address of the idle storage space first appears in the storage space of each level according to the idle information recorded by the recording module, and establish a flow table for the packet; When the flow table corresponding to the address in the storage space of the first type of storage space is in a non-idle state, the flow table is established in the second type of storage space for the packet.

The packet forwarding device according to claim 16 or 17, wherein the table construction module further comprises: a determining unit, configured to: when the obtaining unit acquires a key segment of the packet After the Greek value, it is judged whether the message with the same hash value is being locked;

a first processing unit, configured to: when the result of the determining by the determining unit is yes, after the message waits for the message of the same hash value to be unlocked, perform a corresponding action;

a locking unit, configured to lock the message when the result of the determining by the determining unit is negative;

The unlocking unit is configured to unlock the locked message after the flow table is created for the packet.

The packet forwarding device according to claim 18, wherein the first processing unit specifically includes: a processing subunit, configured to: according to the key segment of the packet acquired by the acquiring unit a value of the same hash value after the message of the same hash value is unlocked, if yes, if the message of the same hash value is the same type. The flow table is configured to perform fast forwarding on the packet; otherwise, after the packet is forged to be unlocked, the flow table is created for the packet.