Specific implementation mode
Come that the present invention will be described in detail below with reference to attached drawing and in conjunction with the embodiments.It should be noted that not conflicting
In the case of, the features in the embodiments and the embodiments of the present application can be combined with each other.
An embodiment of the present invention provides a kind of switch type matrix data central site network systems, as shown in Figure 1, including:It exchanges
Machine and multiple main frames server, above-mentioned interchanger include:Row head interchangers, the first interchanger of row and access switch;Wherein,
Row head interchangers are Layer 2 switch with the first interchanger of row, and row head interchangers are same for that will be located at the first interchanger of row
The access switch of a line links together, and arranges first interchanger and is used to that the access switch of same row will to be located at the first interchanger of row
It links together;
Access switch is three-tier switch, for accessing one or more host servers in network.
In above-mentioned preferred embodiment, using Layer 2 switch as the interchanger that row is first and row are first, in logarithm
During being forwarded according to packet, being expert in the first interchanger of first and row to be directly forwarded by link layer without right
Data packet carries out the parsing of network layer, solves and uses three-tier switch as row head and the first access switch of row in the prior art
Caused by data packet forward efficiency reduce, data packet forwarding complexity increase the technical issues of, to reach simplification
The process of Route Selection improves the technique effect of the efficiency of data packet forwarding.
In view of in actual application, certain exchange opportunities break down, in order to reduce due to interchanger breaks down
Caused by data packet the problem of being difficult to normally be forwarded, can be arranged that multiple rows are first or the first interchanger of row.It is preferred real at one
It applies in mode, multiple row head interchangers is set per a line and/or the first interchanger of multiple row is arranged in each row.
For above-mentioned multiple rows, first first interchanger may be used link aggregation technology and be polymerized to a virtual switch
Machine, that is, used corresponding to the first interchanger of the row of every a line as an interchanger, or multiple row head of each row are handed over
It changes planes and is used as an interchanger, so as to avoid the data forwarding caused by an interchanger goes wrong abnormal,
Also improve the ability of data packet forwarding.
Based on above-mentioned switch type matrix data central site network system, the embodiment of the present invention additionally provides a kind of data packet turn
Forwarding method, as shown in Fig. 2, including the following steps:
Step S202:Access switch determine the data packet that the needs that receive forward destination host whether with access
In subnet corresponding to interchanger;
Step S204:If it is not, then access switch with access switch by being located at a line or same row
Access switch is forwarded data packet.
In above-mentioned preferred embodiment, the interactive matrix data central site network system provided through the invention is connecing
It receives when need to be forwarded, is directly forwarded by being used as intermediate equipment positioned at the access switch of same row or column
, data packet is not parsed at first and interchanger that row are first of being expert at, three are used in the prior art to solve
Layer switch is reduced as the forward efficiency of data packet caused by the first first access switch with row of row, the complexity of data packet forwarding
Property the technical issues of increasing, to reach the process of simplified Route Selection, improve the technique effect of the efficiency of data packet forwarding.
Each access switch in network system is likely to occur failure, in order to ensure it is known that in system each
The link state of access switch, can in each access switch in storage system each access switch link shape
State mainly has the link state of setting:Row is unreachable, it is all unreachable to arrange unreachable or ranks, is also exactly state
Normally i.e. ranks are all up.Wherein, unreachable just refer to when the access switch is as next hop address either purpose master is arranged
When access switch corresponding to machine, data packet can not be by being directly forwarded to the access with it in the access switch of same row
Interchanger.Similarly, the unreachable and similar meaning of row, details are not described herein.In a preferred embodiment, access exchanges
Machine is stored with the link state of each access switch in network system, wherein link state is used to indicate to be exchanged with the access
Whether the data packet that machine is located at the access switch with a line or same row can be directly relayed to the access switch.
For the destination host in same subnet, the forwarding of data packet can be directly carried out by Layer Two Forwarding Protocol,
Destination host for being not at same subnet is then needed by the progress of the access switch of same a line or same row
Between forwarding data are forwarded a packet to the purpose of destination host to realize.In this way before forwarding data packet, it is necessary to determine to this
The routing that data packet is forwarded, in a preferred embodiment, access switch are same by being located at access switch
The step of row or the access switch of same row are forwarded data packet is as shown in figure 3, include the following steps:
Step S302:Access switch, which is determined, is turned data packet by the access switch positioned at same a line or same row
It is dealt into the path passed through needed for destination host;
Step S304:Data are forwarded a packet into destination host by determining path.
If having row unreachable in the path of selection or the inaccessible access switch of row, need to exchange these
The forwarding reason of machine is extended, so that can be forwarded to data packet by other paths, a preferred implementation side
In formula, access switch is determined by the way that data are forwarded a packet to destination host positioned at the access switch of same a line or same row
The path of required process includes:If with access switch be located at a line or same row access switch there are rows can not
Reach or arrange inaccessible access switch;Then access switch is redefined forwards a packet to what destination host was passed through by data
Path.
This can be got for each access switch in the network system of the switch type matrix topology structure
The link state of all access switch of other in system can judge whether that corresponding access switch is in time in this way
Normal condition.In a preferred embodiment, as shown in figure 4, the above method further includes:
Step S402:When access switch determines that the link state of itself changes, access switch is by link shape
State is broadcast to the access switch exchanged with the access positioned at a line or same row;
Step S404:It by the link state received and itself is stored positioned at the access switch of same a line or same row
The link state corresponding to the access switch be compared;
Step S406:If it is not the same, then updating the link state corresponding to the access switch being locally stored.
In the above-described embodiment, exactly the case where link state of itself changes is determined using when access switch
Under, just the Link State Advertisement of itself is given and oneself is located at the access switch of a line or same row so that other connect
A situation of change of itself link state can be known by entering interchanger.Preferably, access switch can be by receiving
Protocol Data Unit PDU state messages determine the link state of itself.
In order to enable the link state of the access switch can be known by all access switch of the whole network, connect with this
Enter interchanger be located at same row or column access switch know the link state of the access switch after, can be by the chain
Line state is broadcasted, to realize the self study of whole network.In a preferred embodiment, positioned at same row or together
One row access switch by the link state received and itself storage the link state corresponding to the access switch into
During row compares, the above method further includes:The link that will be received positioned at the access switch of same a line or same row
State is broadcast to and itself is located at the access switch with a line or same row.
The embodiment of the present invention additionally provides specific embodiment and foregoing invention is specifically described.The invention
Concretism it is as follows:
It, can be on the routing road of the access switch influenced because of failure in the scheme that the embodiment of the present invention is provided
Diameter is recalculated, the characteristics of can giving full play to switch type matrixing network topological structure in this way and advantage, with
It solves the problems, such as the communication performance bottleneck in data center network, while routing algorithm is efficiently simple, and has good fault-tolerance.
The switch type matrix data central site network system that the routing algorithm is based on includes mainly:Interchanger and server two
Class equipment, wherein interchanger includes:Row head interchangers, the first interchanger of row and access switch arranged in arrays.
Every every trade head of matrix at least disposes a row head interchanger, and each column row are first at least to dispose the first interchanger of a row,
Server (i.e. host) is connected with access switch, all row head interchangers that each access switch is expert at it and its
All row head interchangers of column are connected, between arbitrary row head interchangers and the first interchanger of row and each access switch
Between be not connected directly.
Meanwhile in the preferred embodiment, the first interchanger of row and the first interchanger of row are Layer 2 switch, it is only necessary to be had
Layer2 switching function does not need three-layer routing function, meanwhile, it is located at the row head or the first interchanger of row with a line or positioned at same row
One virtual switch is formed using equipment and link aggregation technology.
In the network structure, all-network equipment and server can according to its position configuration of IP in a network
Location.The access switch of i-th row jth row can be abbreviated as A (i, j), and IP address is configured to 10.i.j.1.It is connected to A (i, j)
Server be abbreviated as H (i, j, x), IP address is configured to 10.i.j.x, wherein 1 < x < 255.Row is first and is exchanged with row head
Machine need not configure IP address as Layer 2 switch for it.
The routing function of the data centre network system of the present invention is provided by access switch completely, and routing algorithm mainly wraps
Include two parts:One is basic routing algorithm, the other is the routing algorithm of extension.
In basic routing algorithm, the case where not considering network failure, according to basic routing algorithm, access switch according to
If under type forwarding number is according to packet:
1) when destination host is in this subnet, A (i, j) receives the data packet that destination host is H (i, j, k), then directly
It connected Layer2 switching and is transmitted to destination address;
2) when destination host and A (i, j) are in same a line, A (i, j) receives the data that destination host is H (i, j ', k)
The data packet is transmitted to next hop router A (i, j ') by packet, A (i, j);
3) when destination host and A (i, j) are in same row, A (i, j) receives the data that destination host is H (i ', j, k)
The data packet is transmitted to next hop router A (i ', j) by packet, A (i, j);
4) when destination host and A (i, j) belong to and do not go together different lines, it is H (i ', j ', k) that A (i, j), which receives host address,
Data packet, which is transmitted to A (i, j ') or A (i ', j) by A (i, j).
If certain link and nodes break down in network, data packet is forwarded to be possible to encounter according to above-mentioned rule
Faulty link or node fail so as to cause entire communication process.There is event in the link of node A (i, j) or connection A (i, j)
When barrier, it is necessary to recalculate following routing using extension routing algorithm, be mainly the following situation:
1) for the node A of colleague (i, j '):If node A (i, j) rows are unreachable, recalculating destination address is
The route table items of 10.i.j.0/24;If node A (k, j ') it is partly reachable or unreachable, then recalculating destination address is
The route table items of 10.k.j.0/24.
2) for the node A (i ', j) of same column:If node A (i, j) row are unreachable, recalculating destination address is
The route table items of 10.i.j.0/24;For the node A (i ', j) of same column, if node A (i ', k) is partly reachable or unreachable,
Then recalculate the route table items that destination address is 10.i.k.0/24.
3) for the node A of different ranks (i ', j '), if node A (i, j) is partly reachable, destination is recalculated
Location is the route table items of 10.i.j.0/24.
During recalculating route table items, A* algorithms may be used to calculate, the evaluation function f (n) of A* algorithms
=g (n)+h (n), wherein path cost be since current access switch (that is, source node) to destination node need through
The access switch quantity crossed.Wherein, f (n) is the evaluation function via node n to destination node from source node, g (n) be from
For source node to the actual cost of node n, h (n) is the estimate cost from n to destination node optimal path, for specified node,
Can selecting other nodes of colleague and same column, evaluation function h (n) is according to n and purpose section as its adjacent node (child node)
The relative position relation value of point is as follows:
In order to calculate routing, each access switch is required for obtaining the link state (packet of other access switch of the whole network
It includes:Normal condition or malfunction).Therefore, each access switch need the state of oneself be broadcast to one's own profession and Ben Lie its
His access switch, after the state message for receiving other access switch of this row/column, comparison is stored in any access switch
The state of the local node updates state matrix if state changes, while being broadcast to the row or column where its own, excellent
Selection of land, the message received from row can only be broadcast to row, and the message received from row can only be broadcast to row.
The above-mentioned preferred embodiment provided through the invention:Row is first and the first interchanger of row uses Layer 2 switch,
Routing procedure is simplified, basic routing algorithm is simpler;Using interchanger and link aggregation technology, support that link layer is of equal value
Multi-path route (Equal-Cost Multipath Routing, referred to as ECMP), has two layers of balancing link load ability;
It solves routing issue when network failure using the routing algorithm of extension, there is good fault freedom;With general road
It is compared by algorithm (OSPF, SIP), the topological structure of switch type matrix is taken full advantage of by the learning ability of node link state
Feature so that need the message amount exchanged less, convergence rate is faster.
Based on above-mentioned invention thought, the embodiment of the present invention gives concrete implementation mode, is based on above-mentioned switch type
Matrix data central site network system, the topological structure of regularization may be used, and addressed according to certain rule.
Data centre network system (is addressed to by one group of row head interchanger, the first interchanger of row, access switch:
10.i.j.1) (it is addressed to the various servers being connected on access switch:It 10.i.j.X) constitutes, wherein 1 < X <
255, the i line numbers being expert at by access switch or server, j are the row number of access switch or server column.It is all
The subnet mask of equipment is set as 255.255.255.0.
In order to complete route forwarding function, safeguard a routing table on each access switch, packet according to
Routing table is forwarded.Information (including local IP address and MAC are intercoursed between colleague or the access switch of same column
Location), and learn syntople and equipment or the state of link further calculates outlet to obtain the topology information of the whole network
By table.
In the preferred embodiment, data centre network system is made of interchanger and two class equipment of server.It exchanges
Machine equipment provides two layers (link layers) and three layers of (network layer) function of network exchange, and server apparatus provides data operation and storage
Service.Wherein, interchanger is divided into as three types, the referred to as first interchanger of row, the first interchanger of row and access switch.Row is first to be exchanged
Machine and the first interchanger Layer 2 switch of row, are responsible for access switch to link together;Access switch belongs to three-tier switch,
It is responsible for routing and server is linked into network.Row head interchangers, the first interchanger of row and access switch link together structure
At switch type matrix topology structure.
Under normal conditions, in order to avoid communication performance bottleneck and raising network reliability, each row or column can configure multiple
The first head interchanger access switch that either arranges of row is connected on the first interchanger of all rows head or row of one's own profession or this row.Together
Either the row head of same column or the first interchanger of row form a big virtual switch to row using device aggregation technology.
In order to avoid network congestion, the bandwidth of access switch can be divided into 3 parts, a part of bandwidth is used for connecting
Row head interchangers, a part are used for connecting the first interchanger of row, and remaining bandwidth is used for Connection Service device.Such as:Access switch
Using the three-tier switch of 48 ports, 16 first interchangers of port connection row, 16 first interchangers of port connection row can be used to remain
Lower 16 ports are used for Connection Service device.In view of more servers can be accessed simultaneously, 1 can also be used:1:2 ratio
Distribute the port of access switch.
For most of interchanger product currently on the market, the first line of a couplet port that can generally provide more high bandwidth is used for
The bandwidth ratio of switch concatenation, first line of a couplet port and second line of a couplet port is generally 10:1.Such as:The gigabit of one 40 port exchanges
Machine, 4 10,000,000,000 first line of a couplet ports of general offer simultaneously can also use the first line of a couplet in switch matrix data centre network system
Port connection row is first and arranges first interchanger, and second line of a couplet port is specifically used to Connection Service device.
During use, it is addressed as shown in Figure 1, network addressing scheme below may be used:
The IP address of access switch is configured to 10.i.j.1, wherein i is the line number where access switch, and j is access
Row number where interchanger, 0 < i≤total line number, 0 < j≤total columns.
The IP address of server is configured to 10.i.j.X, wherein i is the line number where the server, and j is the server institute
Row number, wherein 0 < i≤total line number, 0 < j≤total columns, 1 < X < 255, defined within the scope of this, server
Address can be with arbitrary disposition (wherein, X can be arbitrarily designated).
In above-mentioned addressing scheme, the position of equipment in a network can be judged according to the IP address of equipment, this has
Help simplify routing plan.According to the connection relation between addressing scheme and equipment, it can be seen that each access switch and
Its whole server connected constitutes a physical subnets, and corresponding subnet mask is 255.255.255.0.
It is as follows in such a way that the above-mentioned network architecture and its corresponding addressing mode are route:
In the switch type matrixing network that the present embodiment proposes, routing function is provided by access switch completely, and routing is calculated
Method includes two parts:Basic routing algorithm and extension routing algorithm.Wherein, basic routing algorithm does not consider the feelings of network failure
Condition need not also obtain full mesh topology, and route table items are calculated using the systematicness of network topology.It is to base to extend routing algorithm
The supplement of this routing algorithm mainly solves routing issue when network failure.
The present invention realizes routing forwarding using the IP route table of standard.Router-table structure is as shown in table 1:
Table 1
Subnet address |
Subnet mask |
Next hop address |
Cost |
Renewal time |
10.1.0.0 |
255.255.0.0 |
10.1.1.1 |
|
|
10.2.0.0 |
255.255.0.0 |
10.1.2.1 |
|
|
For access switch A (i, j), if there are switch A (i, k) in the i-th row, wherein k ≠ i, A (i, j) and A
(i, k) can be by the first interchanger direct communication of row of the i-th row, and it is capable reachable to claim A (i, j).Similarly, it is handed over if there is one
Change planes A (l, j), l ≠ j, and the row head interchanger direct communications that A (i, j) and A (l, j) can be arranged by jth, it is row to claim A (i, j)
Reachable.If A (i, j) is both capable reachable and row are reachable, it is that ranks are reachable to claim A (i, j).Only ranks can
The interchanger reached can just be used as routing intermediate node.The accessibility of A (i, j) is denoted as Rij=rirj, ri and rj and respectively represents
Row accessibility and row accessibility, value 0 or 1, it is preferable that 0 represent it is unreachable, 1 represent it is reachable.Such as:R12=01 then indicates A
(i, j) is capable unreachable, is arranged reachable.
The ranks of itself can be judged according to the case where PDU state messages received for each access switch
Up to property.If A (i, j) does not receive the PDU of any other access switch of one's own profession, it can be determined that it oneself is capable unreachable
, if not receiving this PDU for arranging other access switch, can be determined that it oneself is that row are inaccessible.All for ranks
Inaccessible node, other nodes can not receive the state message of any node, but can be pushed away according to known nodal information
Break and the presence of the node, and assert that its state is that ranks are unreachable.
The state of itself can be broadcast to one's own profession and other access switch of Ben Lie by each access switch.It is any to connect
Enter interchanger to receive one's own profession or after this arranges the state messages of other access switch, compare the shape for being stored in the local node
State information updates state matrix if state changes, while row where being broadcast to or row (preferably, are received from row
The message broadcasting arrived is to row, from the message broadcasting that row receive to row).
By the broadcast and study of above-mentioned link state, each node will be seen that the state of whole network, be stored in
In one two-dimentional reachability matrix M, wherein M (i, j)=Rij.Since network topology structure generally will not frequently change, it connects
Its reachability information need not frequently be sent by entering interchanger.
Above-mentioned basic routing algorithm can be realized as follows:
Basic routing algorithm calculates the basic road for generating access switch according to the data packet forward rule of access switch
By list item.The forward rule of access switch can carry out in the following way:
1) when destination host is in this subnet, A (i, j) receives the data packet that destination host is H (i, j, k), then directly
It is transmitted to destination address by Layer2 switching;
2) when destination host and A (i, j) are in same a line, A (i, j) receives the data that destination host is H (i, j ', k)
The data packet is transmitted to next hop router A (i, j ') by packet;
3) when destination host and A (i, j) are in same row, A (i, j) receives the data that destination host is H (i ', j, k)
Packet, next hop router A (i ', j) is transmitted to by the data packet;
4) when destination host and A (i, j) belong to do not go together also belong to different lines when, it is H that A (i, j), which receives host address,
The data packet is transmitted to A (i, j ') or A (i ', j) by the data packet of (i ', j ', k).
For access switch A (i, j), basic route table items are as follows:
Destination address/the subnet mask and next hop address of communication are in colleague:
10.i.1.0/24 10.i.1.1
10.i.2.0/24 10.i.2.1
……
10.i.N.0/24 10.i.N.1
Destination address/the subnet mask and next hop address of communication are in same column:
10.1.j.0/24 10.1.j.1
10.2.j.0/24 10.2.j.1
……
10.N.j.0/24 10.N.j.1
It does not go together, communication in different lines, the destination address/subnet mask and next hop address of Column Row are:
10.1.0.0/16 10.1.j.1
10.2.0.0/16 10.2.j.1
……
10.N.0.0/16 10.N.j.1
It does not go together, communication in different lines, the destination address/subnet mask and next hop address of Row Column are:
10.0.1.0/255.0.255.0 10.i.1.1
10.0.2.0/255.0.255.0 10.i.2.1
……
10.0.N.0/255.0.255.0 10.i.N.1
Following manner construction may be used for the building method of above-mentioned basic route table items:
Access switch automatically periodically indicates its PDU survived to the broadcast of other access switch of place row and column,
Access switch is to such PDU broadcasting packet received without forwarding.One access switch receives other accesses and exchanges
After the PDU broadcast that machine sends over, which access switch ranks where can not only having known have, but also can be built according to this information
Vertical routing table.Access switch 10.i.j.1 can be according to following rule construct routing table:
1) if the PDU that the A (i, j ') for receiving colleague is sent, then add routing entry " 10.i.j ' .0/
255.255.255.010.i.j ' .1 " arrive routing table.
2) A if (i, j ') ranks are reachable, then routing entry " 10.0.j ' .0/255.0.255.0 10.i.j ' .1 " is added
To routing table.
3) if the PDU that the A (i ', j) for receiving same column is sent, then add routing entry " 10.i ' .j.0/
255.255.255.010.i ' .j.1 " arrive routing table.
If 4) A (i ', j) ranks are reachable, then routing entry " 10.i ' .0.0/255.255.0.0 10.i ' .j.1 " is added
To routing table.
Preferably, if being not received by newer PDU within the predetermined time, corresponding expired routing item is deleted
Mesh.
Above-mentioned basic route table items are described in detail below in conjunction with a specific example.
1) in same subnet equipment data communication, it is assumed that have two-server IP address be respectively 10.1.1.2 (source) and
10.1.1.3 (purpose), the then communication of the two are directly forwarded by access switch 10.1.1.1.
2) in a line equipment data communication, it is assumed that have two-server IP address be respectively 10.1.3.2 and
10.1.5.2, packet needs to be sent to 10.1.5.2 from 10.1.3.2, and routing procedure is specific as follows:
10.1.3.2→10.1.3.1→10.1.5.1→10.1.5.2
3) in same row equipment data communication, it is assumed that have two-server IP address be respectively 10.2.2.2 and
10.4.2.2, packet needs to be sent to 10.4.2.2 from 10.2.2.2, and routing procedure is specific as follows:
10.2.2.2→10.2.2.1→10.4.2.1→10.4.2.2
4) the data communication of different array devices, it is assumed that have two-server IP address be respectively 10.2.2.2 and
10.4.4.2, packet needs to be sent to 10.4.4.2 from 10.2.2.2, and routing procedure is specific as follows:
10.2.2.2→10.2.2.1→10.4.2.1→10.4.4.1→10.4.4.2
Routing algorithm for extension is mainly the supplement to basic routing algorithm, primarily to when solving network failure
Routing issue.The routing algorithm of the extension is learnt by link state, and each access switch can obtain that the whole network is all to be connect
Enter the state of interchanger, and then recalculates the routed path for receiving malfunctioning node influence by A* shortest path firsts.
The building method of extension route table items can be realized in the following way:
Node or link failure can influence the Route Selection of a part of node, when access switch A (i, j) is that part is reachable
Or when unreachable, influence whether to need the communication for A (i, j) by A (i, j) forwardings and destination node, it is therefore desirable to this
A little situations recalculate routed path, and the calculating of above-mentioned routed path can be realized in the following way:
1) for the node A of colleague (i, j '):If node A (i, j) rows are unreachable, recalculating destination address is
The route table items of 10.i.j.0/24;If node A (k, j ') it is partly reachable or unreachable, recalculating destination address is
The route table items of 10.k.j.0/24.
2) for the node A (i ', j) of same column:If node A (i, j) row are unreachable, recalculating destination address is
The route table items of 10.i.j.0/24;For the node A (i ', j) of same column, if node A (i ', k) is partly reachable or unreachable,
Recalculate the route table items that destination address is 10.i.k.0/24.
3) for the node A of different ranks (i ', j '):If node A (i, j) is partly reachable, destination address is recalculated
For the route table items of 10.i.j.0/24.
A specific embodiment is named to illustrate extension routing algorithm:
It is the example of the reachability matrix for the switch type matrixing network that 4 rows × 4 arrange, as shown in above-mentioned matrix, R22=
10, it is reachable to indicate that A (2,2) is arranged, row is unreachable.For access switch A (1,1), extension routing algorithm is according to A* algorithms at it
Increase by 2 extension routing entries in routing table:
Corresponding destination address/subnet mask and next hop address are:
10.2.2.0/24 10.2.1.1
10.3.3.0/24 10.3.1.1
For access switch A (2,3), extension routing algorithm updates destination address 2 according to A* algorithms in its routing table
Item extends routing entry:
10.3.2.0/24 10.1.3.1
10.3.3.0/24 10.2.1.1
It preferably, can be by certain when determination has unreachable row or the inaccessible intermediate forwarding nodes of row
Algorithms selection goes out nearest routing forwarding path to there is the access switch of obstacle around this.For example, in previous example i.e.
Make to select the node positioned at same row nearest with the unreachable node to be accordingly forwarded as intermediate node, in this way in reality
While existing disaster tolerance, the minimal losses in path are realized.
Therefore, following path may be used in H (2,3, x) and A (3,2, y) communication process:
10.2.3.x→10.2.3.1→10.1.3.1→10.1.2.1→10.3.2.1→10.3.2.y
It can be seen from the above description that the present invention realizes following technique effect:
Using Layer 2 switch as the interchanger that row is first and row are first, to during being forwarded to data packet,
Being expert in the first interchanger of first and row to be directly forwarded by link layer without the solution to data packet progress network layer
Analysis solves the forwarding for using three-tier switch as data packet caused by the first first access switch with row of row in the prior art
The technical issues of efficiency reduces, and the complexity of data packet forwarding increases improves number to reach the process of simplified Route Selection
According to the technique effect of the efficiency of packet forwarding.
Obviously, those skilled in the art should be understood that each module of the above invention or each step can be with general
Computing device realize that they can be concentrated on a single computing device, or be distributed in multiple computing devices and formed
Network on, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
It is performed by computing device in the storage device, and in some cases, it can be with different from shown in sequence execution herein
The step of going out or describing, either they are fabricated to each integrated circuit modules or by them multiple modules or
Step is fabricated to single integrated circuit module to realize.In this way, the present invention is not limited to any specific hardware and softwares to combine.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, any made by repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.