BGP (Border Gateway Protocol) border gateway protocol, used to connect the Internet, independent system routing protocol. It is the Internet Engineering Task Force developed an enhanced, comprehensive and scalable protocol. BGP4 supports CIDR addressing scheme that is available on the Internet increase the number of IP addresses. BGP is to replace the original design EGP Exterior Gateway Protocol, is also considered a path vector protocol.
directory
Detailed background operations expanded the definition of message type routing BGP synchronization rules packet format and the basic needs of BGP BGP synchronization solution to synchronize a summary of BGP synchronization rule of BGP protocol BGP protocol level position The message structure properties of BGP path attribute type BGP routing protocol BGP scalability features of application of restricted problem of how to address the constraints of several extended issues summary BGP Route Optimization PathControl Yong pick responsibility ISP network performance testing, see the deployment of commercial value PE, the principle of setting the definition of routing BGP (Border Gateway Protocol) is a dynamic between the autonomous system routing protocol for exchanging routing information. The classic definition of an autonomous system is a regulatory body under the control of a router, which uses the IGP and the general metric system to other autonomous packet forwarding. BGP autonomous system in the use of the term is to emphasize this fact: the management of an autonomous system to other autonomous systems is to provide a unified internal routing program, which for those who can reach through its network to provide a consistent description. BGP, Border Gateway Protocol, is an autonomous network system in the gateway switch routing information between the protocols. BGP is often used between the Internet gateway. Routing table contains a list of known routers, the router can reach the address and the path to reach each router hops. Hosts using Border Gateway Protocol Transmission Control Protocol generally used (TCP). When the network detects changes in a given host, it will send the new routing table. BGP-4, the latest version of border gateway protocol that allows network administrators to configure the policy described in the specifications of hops. Extension is a different autonomous system BGP routers to communicate between the exterior gateway protocol. BGP is used by the old ARPANET EGP's replacement. RFC1267 [LougheedandRekhter1991] on BGP version 3 are described. RFC1268 [RekhterandGross1991] describes how to use the Internet BGP. Most of the following description of the BGP comes from both RFC documents. Meanwhile, development of version 4 in 1993, the BGP (see RFC1467 [Topolcic1993]), to support CIDR. BGP BGP system and other systems of information exchange network can be reached. This information includes data to reach these networks must go through all of the autonomous system AS path. This information is sufficient to construct an autonomous system connection diagram. You can then delete the selection based on Coloane connection diagram to develop routing strategies. First, an autonomous system in IP datagrams into the local traffic and through traffic. In the autonomous system, or termination of local traffic is starting the flow in the autonomous system. In other words, the source IP address or IP address of the destination host specified in the autonomous system. The other is called the flow of through traffic. BGP in the Internet using a purpose is to reduce through traffic. Autonomous system can be divided into the following types: 1) stub autonomous system (stubAS), and other autonomous systems it is only a single connection. stubAS only local traffic. 2) multi-interface autonomous system (multihomedAS), it and other autonomous systems have multiple connections, but refused to send through the traffic. 3) transferred to the autonomous system (transitAS), it and other autonomous systems have multiple connections in a number of policy guidelines under which it can send local traffic and through traffic. This allows the overall topology of the Internet as a stub by a number of autonomous systems, autonomous systems and multi-interface forwarding arbitrary interconnection of autonomous systems. Stub autonomous systems and multi-autonomous system interfaces do not need to use BGP - they are by running the EGP exchange between autonomous systems accessible information. BGP allows policy-based routing. Formulate a strategy by the autonomous system administrator, and through the configuration file specifies the policy to BGP. Development of strategy is not part of the agreement, but the policy allows BGP to achieve the specified number of optional path in the presence of the selection of the path, and control information re-sent. Routing strategy and the political, security or economic factors. BGP and RIP and OSPF BGP is the difference between using TCP as its transport layer protocol. BGP running between the two systems to establish a TCP connection, and then exchange the BGP routing table. From that time, the changes in the routing table, and then send the updated signals. BGP is a distance vector protocol, but with (notice number of hops to the destination address) RIP The difference is, BGP lists the route to each destination address (destination address autonomous system sequence number.) This excludes a number of distance vector protocol issues. Autonomous systems that use 16bit digital identity. BGP by periodically sending keepalive packets to its neighbor stations to detect TCP connection on the side of the link or host failure. Packets between the two proposals is 30 seconds time interval. The application layer and TCP-keepalive keepalive packets option is independent. Detailed background route consists of two basic actions: determine the best path and Information Group (often referred to as packet) transmission through the network. Packet transmitted over the network is relatively simple, and the determination of the path complex. BGP is the routing of today's networks to achieve an agreement. The following summarized the basic operation of BGP, and provide a description of the protocol components. BGP in the TCP / IP network to achieve inter-domain routing. BGP is an Exterior Gateway Protocol (EGP), which is in multiple autonomous systems or routing inter-domain implementation, with other BGP systems to exchange routing and reachability information. BGP design to replace its predecessor (now do not have a) External Gateway Protocol (EGP) as a global Internet standard exterior gateway routing protocols. EGP BGP solves serious problems, to more effectively adapt to the rapid development of the Internet. The following are the core router to use BGP routing data between the AS diagram.
bgp
BGP specified in a number of RFC: RFC1771 - describes the BGP4, BGP is the current version. RFC1654 - describes the first BGP4 specification. RFC1105, RFC1163 and RFC1267 - BGP described earlier version of BGP4. Perform three types of operating BGP routing: AS routing between, AS AS internal routes and throughout the route. Inter-AS routing in two or more different AS BGP routers, these systems use BGP peer router to maintain a consistent view of network topology, AS BGP communication between neighbors must be in the same physical network. The Internet is an instance of using this route, because it is by the number of AS (also known as administrative domains) form, many of the Internet domain for the posed research institutions, companies and entities. BGP is frequently used to provide the best path for the Internet and do routing. AS internal routes in the same AS in two or more BGP routers, the same on the other routers within the AS using BGP to maintain a consistent system topology view. BGP is also used to determine which router as a connection point outside AS. Again, the Internet provides inter-AS routing instance. An organization, such as universities, can take advantage of BGP in its own administrative domain (or AS) to provide the best route. BGP protocol can provide both inter-AS AS can also provide internal routing.
bgp
through (pass-through) AS routing does not run BGP in the AS through the exchange of data between two or more BGP peer routers. In the environment throughout the AS, BGP AS communication is neither from within nor the purpose of the nodes within the AS, BGP must be used within the AS routing protocol interaction in order to successfully transport BGP traffic through the AS, the next figure shows the cross-cutting AS environment: the same routing and other routing protocols, BGP routing table maintenance, message routing updates and routing decisions based on routing metric. BGP main function is to exchange the network up to other BGP systems, including a list of AS path information, this information can be used to establish the AS system connection diagram, in order to eliminate the routing loop, and implementation of strategies to determine AS. Each BGP router maintains to a specific network path consisting of all available routing table, it does not clear the routing table, it maintains the router receives from the peer routing information until you receive the value-added (incremental) update. BGP devices exchange of data and value-added in the initial exchange of routing information update. When a router first connects to the network, BGP routers exchange their entire BGP routing table, Likewise, when the routing table changes, routers send part of the routing table changes. BGP router does not send periodic routing updates, and BGP routing updates contain only the best path to a network. BGP routing with a single metric to determine the best path for a given network. This metric links with the specified priority value of any unit, BGP's metric is usually assigned to each network link. The value assigned to a link can be based on any number of scales, including via the AS number, stability, speed, delay or expense and so on. BGP message type specified in RFC1771 four BGP message types: the initial (open) the message, update message, notification message, and keep-alive message. Other routers in the initial message opens a BGP communications session between, transport protocol is to establish the first message sent after the initial message sent by the peer keep-alive message acknowledgment, and must be confirmed before they can exchange update notice and keep-alive message. Update message to other BGP systems used to provide routing updates, so the router can establish a consistent view of network topology. Updated with the TCP transmit to ensure transmission reliability. Routing table update message can be removed from the one or more invalid routing, also released a number of routes. Notification message to an error in the check is sent. An event notification message used to close the session and notify the other routers why close the session. keep-alive message to inform other BGP routers on the device is still alive. keep-alive message is issued frequently enough to prevent the session expired. Brief initial BGP packet format, update, notification,
air force one high, and keep-alive message type and basic BGP header format. 1, letter head format
bgp
All BGP message types use the basic packet header. Initial,
air force 1 high, update, and notification messages have additional domains, and keep-alive messages using only the basic packet header. The figure below shows the domain used by BGP header: Each BGP packet contains headers, its main purpose is to identify the grouping function. Summarized below for each header field. Mark - with certified values. Length - Indicates the total length of the message, in bytes. Type - identifies the message type is one of the following types: initial update notifications keep-alive data: is an optional field, with the upper information. 2, the initial message format
bgp
BGP initial message from the BGP header and additional tracts, the next picture shows the additional fields BGP initial message: The type field in the header identified as the initial BGP BGP message packet contains the following fields, these fields BGP routers establish two offers exchange programs and other relations: Version - Provides BGP version number, so the receiver can verify that it is running the same version of the sender protocol. Autonomous System - provides the sender's AS number. Hold time (Hold-time) - is considered failure in the sender does not receive the message before the longest number of seconds. BGP ID - provided the identity of the sender (IP address), the decision to start on all local interfaces and all other BGP routers are the same. Optional parameter length - optional parameter field identifies the length (if any). Optional - contains a set of optional parameters. Currently only an optional parameter type is defined: authentication information. Authentication information contains the following two fields: Authentication Code: Identifies the type of authentication used. Authentication data: Contains data used by the authentication mechanism. 3, the update message format
bgp
BGP update messages from the BGP header and additional tracts, the next picture shows the additional fields BGP update message: received update message packet, the router from the routing table can add or delete the specified routing table entry to ensure accuracy. Update message contains the following fields: failure route length - identifies the failure of the total length of the routing domain or the domain does not exist. Failure Routing - IP address contains a set of invalid routing prefixes. Total path attribute length - the path attribute identifies the total length of the domain or the domain does not exist. Path Properties - Description publish path property, the possible values are as follows: Source: Required attributes define the path of information sources. AS path: Required attribute, by a series of AS path segments. Next hop: A required attribute defines the Network Layer Reachability Information field in the application list done Mu Dede's edge router next hop IP address. Multiple export distinction: an optional attribute used to the neighboring AS to distinguish between the various export. Local Priority: optional attribute to specify the release level of priority routes. Atomic Aggregation: optional attributes for the distribution of routing information. Aggregation: an optional attribute that contains the aggregate routing information. Network Layer Reachability Information - IP address contains a set of published routing prefix. 4, the notification message format:
bgp
Below is the additional use of BGP notification message fields: a notification message to the peer group used to notify certain error conditions routers. Error code - identifies the type of error occurred. The following is the definition of types of errors: wrong header: header that is a problem, such as the length of the message is not acceptable, mark value or message type. The initial message was wrong: that the initial message is a problem, such as the version number is not supported and unacceptable AS number or IP address or does not support authentication code. Update error message: update message that is a problem, such as the property list incomplete, incorrect or invalid property list next hop attribute. Hold time expired: that hold time has expired, after which BGP node is considered to have lapsed. Finite state machine error: expected direction outside the event. Termination: When a serious error device according to the request of BGP BGP connection closed. Error sub-code - the error on the report provided more specific information. Erroneous data - contains the error code and error based on sub-code domain data, to detect the reasons for sending notification messages. BGP synchronization rules for the basic needs of 1.BGP Sync synchronization and definition of the rules: in bgp synchronization turned on, a BGP router will not neighbors who learned through ibgp bgp routes to their neighbors ebgp; unless their igb routing table of these routes, we can notice the router to ebgp. 2.BGP synchronization rules aim: to prevent an AS (not all routers are running bgp) route emerged within the black hole, namely an external circular unreachable the AS false route.
BGP synchronization rules topology indicate
3.BGP the basic needs of synchronization rules that exist within the AS if a non-bgp router, then appeared in the border BGP and IGP, you need to BGP border routers Post to igp route in order to ensure that the notices to the external AS BGP routing within the AS is connected. actually requires BGP routing and routing igp sync. 4. to meet the basic needs of BGP synchronization rule if the result of BGP routing Post to igp, due to BGP route is from AS external routes (from internet), then the result is tens of thousands of igp router to maintain external routes, the router's CPU and memeory, and link bandwidth within AS the occupier will have a tremendous overhead. 5. Conclusion usually need to be closed to run BGP protocol synchronization. BGP synchronization solution 1.full mesh iBGP solutions to all routers within the AS are running full mesh iBGP, routers can turn off all synchronization without affecting the circular route and connectivity. The problem:
Full-Mesh IBGP topology hands
When the number of routers within AS often need to create N * (N-1) / 2 个 ibgp conversation with to excessive overhead, scalability is not good. Full-Mesh IBGP topologies indicate: 15 routers of AS, need to build 15 (15-1) / 2 = 105 ibgp session 2. route reflector for all solutions within AS routers are running bgp, in the AS route reflectors within the deployment, building hub and spoke the ibgp (number of sessions for the N-1), then close all the synchronization bgp router. The problem: This program can pass ibgp bgp router route to ebgp, and to ensure the connectivity of bgp routing. but there are substantial limitations physical topology (which requires a star topology) 3.bgp Union Solution:
Bgp route reflector design topology
AS all routers within running bgp, to a network topology based on the original AS is divided into several sub-AS (also known as the Union AS), AS alliance between the bgp neighbor called the Union ebgp, without full mesh bgp session; in the internal operation of each Union AS full mesh ibgp or hub and spoke reflectors, and then you can turn off all the router bgp synchronization. Conclusion: bgp route reflector alliance means a better solution to bring the rules bgp synchronization needs, is the most effective solution program. BGP synchronization rules summary
bgp Union topology diagram
1. In all scenarios, it is necessary to ensure the delivery bgp routing, but also to ensure the connectivity of bgp routing. 2. bgp route close synchronization can be achieved transfer, does not necessarily guarantee as internal connectivity, unless all the routers are running as we can to ensure connectivity bgp; otherwise, still need to route redistribution (bgpàigp) 3. Finally, within the general need to be deployed in as to maintain AS igp connectivity within the network path to ensure that as the notice within the bgp next hop routing reachability. This bgp network to have better flexibility and scalability. Protocol analysis is an autonomous system BGP between Dynamic route discovery protocol, its basic function is to automatically switch between the non-autonomous system routing information loop. RIP, and OSPF and the Ministry of runs in the autonomous region within the corresponding agreement, BGP is a kind of EGP (Edge Gateway Protocol) protocol, and OSPF and agreements. EGP BGP is based on the application to develop. EGP previously in this autonomous region as a route between the discovery protocol, widely used in NFSNET other backbone networks. However, EGP was plagued by routing loops. BGP routing information through the addition of the Autonomous Regions (AS) path attribute to construct the autonomous region of topology, which eliminates routing loops and implementation of user-configured policy. Meanwhile, with the rapid development of INTERNET, the routing table size also increased rapidly, routing information between autonomous regional exchange capacity increases, all affect the network performance. BGP support for the type of inter-regional routes without CIDR (Classless Interdomain Routing), can effectively reduce the growing routing table. BGP autonomous run-time, respectively, with the extra-territorial and regional partners within the BGP connection (using Socket). Connections with partners in the region known as the IBGP (Internal BGP) connections, and the BGP autonomous partners extraterritorial connection is called EBGP (External BGP) connections. Local IBGP and EBGP BGP protocol partners use different mechanisms. BGP BGP protocol level position to establish a connection using the Socket service, port number is 179. BGP BGP message structure has 4 types of messages. Are OPEN, UPDATE, KEEPALIVE and NOTIFY. They have the same header structure header ⑴: Marker: (16 bytes) authentication information Length: (2 bytes) the length of the message Type: (1 byte) message type 0: OPEN 1: UPDATE 2: NOTIFICATION 3: KEEPALIVE ⑵ OPEN message structure: a header plus the following structure: Version: (1 bytes) BGP version number of the originator My Autonomous System: (2-byte unsigned integer) The local AS number Hold Time: (2-byte unsigned integer) originator of the proposed hold time BGP Identifier: (4 bytes) the originator of the router identifier OptParmLen: (1 byte) length of the optional parameters Optional Parameters: (variable length) optional argument ⑶ KEEPALIVE message structure is only one message KEEPALIVE message head. ⑷ NOTIFY message structure, message header plus the following structure: Errsubcode: (1 byte) Auxiliary error code, slightly. Data: (variable length) depends on the different error codes and supporting error code. Used to diagnose the error. ⑸ UPDATE message structure, message header plus the following structure: Unfeasible Routes Len: (2-byte unsigned integer) is not up to the routing length of the Withdrawn Routes: (variable length) exit routes Path Attribute Len: (2-byte unsigned integer) Long Path path attribute Attributes: (longer) path attributes (detailed below) Network Layer Reachability Information: (variable length) network reachability information (the destination) where exit routes and destination address expressed as a binary group. length byte, indicating the address prefix length. prefix for the address prefix, the length of 1 to 4 bytes. BGP path attributes for each path attribute properties from 1 byte flags, 1 byte of attribute type, 1 or 2 bytes and the path length of routing attributes attributes of data. Attribute flags: bits 0:0 indicates that the property must be selected, 1 means the attribute is optional. Bits 1:0 indicates that the property of non-transition property, a property that this property transition. Bits 2:0 are routing all of the properties that the beginning of production, an intermediate AS that added a new property. Bit 3:0 route attribute that indicates the length by 1 byte, and 1 from the 2-byte instructions. Bit 4 to bit 7: not used set to 0, bits 0 and 1, 4 identifies the BGP route attribute class: (01) recognized Required: BGP in the UPDATE message must exist in the property. It must be recognized that all BGP tools. Required attributes recognized UPDATE packet loss means that the error. This is to ensure that all tools are integrated in a standard set of BGP attributes. (01) recognized self-determination: that can be recognized by all BGP attributes, but in the UPDATE packet can be made from time to fat (11) optional transition: If the tool does not recognize the optional BGP attribute it to find the transition attribute bit. If this property is the transition, BGP tools to accept the property and pass it forward to other BGP routers. (10) Optional non-transition: When the optional attribute is not recognized, and the transition properties are not set, this property is ignored and not passed to other BGP routers. Routing attribute type ⑴ ORIGIN (Type Code = 1, recognized Required attribute) indicates the start of this route type: ⑵ AS_PATH (Type Code = 2, recognized Required attributes) AS path attribute by a series of AS path segments (Segment) composition. Each AS path segment for a triple. Path types: Path length AS numbers with a number of bytes that a maximum of 255 AS numbers. AS number is a number of paths, each AS number is 2 bytes. ⑶ NEXT_HOP (Type Code = 3, recognized Required attribute) This attribute in UPDATE messages using the destination address of the next hop. ⑷ MULTI_EXIT_DISC (Type Code = 4, known discretionary attribute) called MED attribute. A 4-byte integer without compliance. It spread among the regional AS, to help one other partner selection BGP AS region into the region of the AS population. ⑸ LOCAL_PREF (Type Code = 5, recognized self-attributes) local preference. A 4-byte integer without compliance. It is spread in the AS region, to help a region of the AS BGP AS partner selection into other areas of the exports. ⑹ ATOMIC_AGGREGATE (Type Code = 6, recognized self-attributes) per polymer properties. Length of zero. It said that in a number of local BGP routing selected a more abstract (less specific) route, but did not choose a more specific (specific) route. ⑺ AGGREGATOR (Type Code = 7, the optional transitional attributes) aggregators property. Length is 6 bytes, respectively, the final routing aggregation router AS numbers (2 bytes) and IP address (4 bytes). Characteristics of BGP protocol is BGP AS (autonomous region) of external routing protocols, mainly responsible for the autonomous region and autonomous areas outside the reach of information between the routing exchange. Therefore, it is concerned with the topology of AS (autonomous region) topology, BGP UPDATE message routing through the AS properties to construct the AS topology map, and further through this structure to select the route. And OSPF, RIP, compared to IGP protocols, BGP topology map to some of the more abstract and rough. IGP protocol is constructed as an internal router AS topology. IGP into a number of the abstract endpoint router,
nike air force 1 high, the router, the link between the abstract into the side of the state and other parameters according to the link and some metrics, each side accompanied by a certain weight to generate topology. According to the topology selection costs (through the edge between two weights and) the smallest routing. There is an assumption that the router (endpoint) to forward the packet is not a cost. In BGP, the topology of the endpoint is an AS region, edge is the link between AS. At this point, the data packets through an endpoint (AS Autonomous Regions), the price can not be assumed to be 0, and the costs to be calculated by the IGP to be responsible. This reflects the EGP and IGP is a hierarchical relationship. IGP within the AS is responsible for the least cost routing options, EGP is responsible for selecting the least cost routing between AS. EGP BGP as a kind of choice is the route to take into account the link between the cost of AS, AS costs within the region (by the BGP router configuration), and other factors. As mentioned above, should be introduced interior gateway protocol IGP within AS Autonomous Region Department of the network topology other points of the route, while sending the endpoint to other endpoints (routers) know the route, such as direct routing, static routing. As an external gateway protocol, BGP routes sent and the introduction of the AS units are autonomous regions, namely the local router to send BGP AS where all the internal routing, the introduction of regional autonomy to all the other AS route (assuming the routing policy control is not used to send and the introduction). The route is clearly much larger than the number sent and the introduction of IGP routing number. Thus, similar to the IGP routing information as regular outside broadcast is not desirable. BGP routing by sending incremental (Incremental) method, all the routing information to complete notice and maintenance: initialization sent to all BGP peer routing (BGP Peer), while kept in the local BGP peer has been sent to the routing information. When the local BGP received a new route (for example, has injected a new route through the IGP or add a new static routes), and saved to compare the information has been sent, if not sent before, then send, and who has sent over the compare the routing has been sent, such as new route takes less, then send the new route, while updating the information has been sent, otherwise not sent. When the local BGP find a route failure (such as the corresponding port failure), so the routing has been sent off, then to BGP peer sends an exit route messages. Application constraints Problems BGP BGP scalability is used in the Internet protocol IP network interconnection for the interconnection between operators to provide a stable and secure routing protocol, has a wealth of routing control mechanism. In order to better control the routing strategy, the current most of the operators are the backbone routers will be deployed to the BGP. As the network continues to expand, increasing the number of routers and routing information entries in the surge, to solve the scalability problem of BGP has become increasingly important. BGP scalability currently facing the following questions. (1) I-BGP in the Full-Mesh routing protocol BGP issues into I-BGP and E-BGP two parts. I-BGP router for autonomous domain, E-BGP is used between the routers between autonomous domains. In order to prevent circular routing, BGP protocol requires a router learned through the I-BGP route, not the other I-BGP neighbor to the radio, so all participants within an autonomous I-BGP protocol router must establish a session with the other routers to ensure the correct routing information can be broadcast to every router. In accordance with this principle, an autonomous region of the I-BGP sessions total number of N × (N-1) / 2 (N to run I-BGP router number), when N is increasing, this figure will be staggering: such as 100 router, the session number is 4950. This is in terms of network equipment is a very big burden, but also to network management and configuration become very complicated. The backbone of the net usually run by a large number of I-BGP router composition, therefore, is able to solve this problem directly affects the size of the network. (2) change the routing strategy BGP route oscillation problem of routing protocol updates are incremental, when a new route to be released, the router will send a Update neighbor information, and if you want to delete a route, it will send a Withdraw information. BGP Route Flap is defined as: when a route is withdrawn at (Withdraw) after he was broadcasting (Update) out, as a Flap. Since any recovery and update a route will lead to a router to recalculate the entire routing table, so when compared to the situation Flap long time, the load on the routers will have a tremendous pressure. According to the author's experience in practical work, under normal circumstances, a high-end router, BGP routing in the calculation of the time, CPU load is basically 80% ~ 90%, sometimes even 100%, taking up almost all CPU resources. Although most of the high-end routers will route calculation module and relay module located in a different hardware, to reduce the busy main CPU performance degradation due to router problems, but frequent changes in the routing table and update the entire device running or has some influence, and such calculations will be back with the routing or broadcast, to continue to expand self-government within the Department, so that the internal routers have the same problem. (3) Other issues to be considered in addition to the above two issues can lead to excessive consumption of router resources, there are other factors, such as the number of routing, BGP routing table size and routing calculation methods, also will affect the performance of the router. In addition, the larger the network, the more routing entries, configure and manage the more complex work, which requires time in the network design configuration as simple as possible, reducing management staff strength, to avoid the human causes failure. How to solve the problem of restricting BGP scalability above problems, introduced a number of related solutions. (1) I-BGP sessions to solve the bottleneck of the method mentioned above I-BGP sessions too many problems, can take two approaches to solve: 1) Union Method (Confederations) Union works is: an autonomous domain of the original network into multiple sub-autonomous domain, through the Confederations id AS number allocated to each of the original router. This has two advantages: First, to retain the original I-BGP attributes, including the Local Preference, MED and NEXT_HOP; the second is the function in Confederations automatically without the need for administrators to configure the filter at the exit of the network within the AS number information operations. 2) route reflector (Route-Reflector) reflector using route is the most widely used method, the method than the previous coalition, has better scalability. Route reflector works: an autonomous region will be divided into several routers Cluster, each Cluster and Client components from the Reflector. Reflector between the Full Mesh, runs the conventional I-BGP; Client Reflector only run I-BGP, the Client is, Reflector is just an ordinary neighbor, Reflector acts as a routing hub role, learn from other Reflector The I-BGP route to forward to the Client, at the same time, learn from the Client's I-BGP route forwarded to other Clients within the Cluster and Cluster other than Reflector, and then broadcast it through the Reflectors in a Cluster. In a real network in order to improve redundancy, usually more than one Client and Reflector establish neighbor relations, but not limited to Client where the Cluster's Reflector. Thus, Client on the I-BGP sessions are generally 1 or 2, compared with the alliance, as long as the performance of Reflector is high enough, Cluster can do much, while the Client will not load with the Cluster's change much change. For the former method, due to a sub-domain of autonomy to do all the router or Full Mesh, so the minimum performance of the router determines the size of a sub-autonomous; and route reflector rule is usually the highest performance of one or more routers Cluster size of the decision, therefore, has better network scalability. In addition, a Reflector Reflector can also become another of the Client, the formation of hierarchical structure, which is particularly suitable for the construction of the network in accordance with the hierarchical structure, the plane can easily be turned into the network management layer management. Of course, when using the route reflector, there are some areas requiring attention, such as: Reflector is not purely forward routing, all received the same route to go through the Reflector on the best terms, then the preferred route out radio Therefore, the choice should be based on Reflector network structure may be, try to make I-BGP neighbor relationship with the actual connection between the corresponding circuit. (2) control the routing oscillations present, the main take Route Flap Damping control method: a BGP router receives the E-BGP routing Penalty values, each route will make the route Flap Penalty value increases, the routing stability Time, Penalty value decreases over time; when the inhibition Penalty exceeds preset limits, the route is no longer being broadcast, and when the Penalty reduced to below the reuse limit, the route will be re-out broadcasting. This approach is for the purposes of E-BGP neighbors. This occurs when a route oscillation within the network when connected to other network defenses by Damping can effectively impact on their respective networks. In the network maintenance, often encounter such a situation: rather a circuit interruption for some time, do not want it to stop shock. After a period of time because of interruption, as long as the circuit recovery, network traffic will follow recovery; In contrast, the oscillation of the circuit is restored, need to wait some time to restore normal flow (all other network routes in the inhibition of oscillations). Damping can effectively ensure the stability of the Internet, but will also lead to some delay fault recovery. The current basic support Damping device functions, and parameters can be configured to provide precise control of Damping. Under normal circumstances, the device provides the default configuration to meet the needs of most networks, if indeed there is special need, be sure to calculate carefully, otherwise, not without effect, is to suppress other people's routes can not be restored after a very long time. (3) Peer Group's application in real network, there is a very common phenomenon, that is, there will be more than a router belongs to a class of BGP Peer. Here, Peer to do a strategic calculations (although the strategies are the same). In addition, the application of Peer Group also has the advantage of the router device is to reduce resource consumption, because the router updates the routing of the same Group, due to use the same strategy, only once a route calculation, thus greatly reducing the CPU-time. In practice, network maintenance, even when the same class Peer little, the generally recommended way of Peer Group because it has good scalability. Peer Group is not only applicable to I-BGP's Peer, also applies to E-BGP's Peer. (4) measures on the route refresh backbone network maintenance, BGP policy often encounter the situation changes, such as updating a Peer of the list as-path constraints, it is required after the configuration changes take effect before the current practice is to interrupt the BGP session, and then re-established, the command is as follows: clear ip bgp xxxx to do so mainly because the other BGP router receives the routing table and made a strategic calculation, then the optimal route is stored in the machine's BGP routing table , not to save all the original route. When the policy changes, must re-establish the connection to get all of the other route, and then re-calculated using the new strategy. Drawbacks of this approach is obvious, such as network blocking, large consumption of the router CPU resources. Now commonly used to solve this problem two ways: One way is to try to save the original BGP routing, so you do not need to re-calculation of the retransmission; Another method is to enable the new strategy, without interrupting the BGP session under the premise of the retransmission request to the other all the BGP routing table. The first method is implemented using software configuration, the configuration of the command is as follows: neighbor 1.1.1.1 soft-reconfiguration inbound at the opening of the new strategy to type the command: clear ip bgp 1.1.1.1 soft [in | out] so that all of the original BGP routes will be stored in a routing table inside another, in the modified strategy, through the operation of this routing table, you can get the best BGP routing table, so that the consumption of resources on the route have smaller; and there is such an original routing table, you can modify the strategy, use the command before the opening to check the effect of the revised strategy. But this method requires additional memory resources to store the routing table. The second method is the ability of the router device BGP (BGP Capabilities) decision, that is, the system software itself is built-in function modules. When the establishment of BGP sessions, this parameter through the BGP OPEN exchanged between the two routers. A device supports BGP Refresh Capabilites, can use the following command to check: show ip bgp n xxxx If you participate in both BGP routers have this feature, you do not need any configuration, when the implementation of clear ip bgp n xxxx in time, the end BGP BGP process will not interrupt the session, but on the client requests to resend the entire BGP routing table. Relative to the first method, this method has the advantage of saving memory resources, the disadvantage is the network administrator can not understand each other sent me the original route, and re-issued all the routing table, the efficiency is not the first method is also high. Currently the equipment is basically the backbone support this feature. In summary BGP scalability of some related issues and related measures were discussed, in practical applications also need to be deployed according to specific circumstances. But the general ideas and principles are the same as that on the one hand to reduce device resource consumption, on the other hand to simplify maintenance and management, from hardware and software aspects to improve the network scalability. Route Optimization
bgp
If you have more than one Internet connection, you will know that only simple to use Border Gateway Protocol (Border Gateway Protocol, BGP) is difficult to get the optimal routing path. At this point, there is no telecommunication networks sophisticated signaling system No. 7 (Signaling System 7, SS7). Control surfaces as a telecommunications network, SS7 before the establishment of a call to determine the best path. There is no such control on the Internet side, although the decision can BGP routing (similar to this point and SS7), but it is not always choose the best path. BGP internal or external, without interrupting the user's situation to the edge router connected via a preferred transmission services, if the preferred connection fails to connect through the second choice to provide services. When a BGP route selection, if the connection can be taking into account each ISP can provide the performance of the routing path and other factors, the situation will certainly be better. RouteScience the PathControl is such a new product, which not only can be measured by each BGP peer's routing performance, you can also use the information to change the priority through the BGP routing. PathControl Yong pick responsibility PathControl 1.1 running on Linux, to provide independent functions,
nike air force one low, one for the report, one for management, and one to the core engine, for each external BGP peer has an interface. PathControl 1.1 provides a robust tool for Java programs. The next version will be allowed from the GUI (graphical user interface) configuration PathControl, but must be in the current version of IOS-based command line interface management device. PathControl can provide not only connectivity and their respective internal information channel, etc., but also allows you to configure a number of variables to control the use of those connections. PathControl can be placed on the edge router to any location. Connection to the router via a dedicated port or a channel connection is established, either way, PathControl need the outside world and the edge router for a visit. See the test performance will PathControl directly connected to a Cisco Catalyst 6500 switch, the switch is configured to receive from two Nortel Web Switch (a 180e and a AD4) external BGP input data. PathControl with the Cisco switch is set on the other, Nortel equipment to monitor statements by each route. PathControl can remain in the passive state, monitor and report on each of the performance of the remote peer; It can also be placed in statement mode, which allows it to determine which of the better performance of a connection after the change in the edge router the preferred route . Users can configure the frequency of PathControl change the routing so that the user's edge router will not be suppressed. Best to allow PathControl at least continue to run in passive mode, a few days, using the report generator to observe PathControl declare a new route frequency, and then determine the best conditions for the specific frequency change. PathControl TCP handshake with a round-trip time as the basis of their performance metrics through a HTTP client to send a 1x1 pixel GIF images to measure the performance of the passive retrieval of data, user settings can also probe the Web site to be active retrieval. For the latter, PathControl open a TCP connection to reach a site and measure the complete three-way TCP handshake time. To achieve a passive measurement device for each measurement interfaces are configured a virtual IP address (Virtual IP Address, VIP). Via HTTP or HTTPS when accessing the address, return a 1x1 pixel GIF image. PathControl direct the GIF, measurement and a client to establish a TCP the call. As long as the VIP address embedded into the web site, a HREF, once a single client issuing the request for the page, start measuring. Herein is the key to the use of policy-based routing, the routing configuration in the edge router, according to the source address of coherent communication data transfer. This PathControl each path will be continuous measurement of performance, without having to consider the other items in the routing table. Default PathControl for the implementation of the calculation of the data measured at least 6 times. Calculation to give a grade for each path, the class will be used to determine which of connection work better. By default, if two or more grades are assigned numerical difference of not more than 25 points, they will be considered If a certain level of connection 25 points higher than the other, it is considered to be the routing. ISP network to deploy
bgp
With the development in China INTERNET, ISP service providers and INTERNET connection is also changing. In the network connection from a single connection to multiple telecommunications companies by a number of lines connected to the INTERNET; in the routing protocol used, from using static routing protocol, developed to use the BGP (Border gateway protocol). So how With the development of ISP, the network within a reasonable and effective deployment of BGP routing agreement? In this article we will briefly explain. In fact, how to deploy the network routing protocol BGP is the dominant thinking of how to make self-government within the Department of the Ministry of the router to obtain routing information extraterritorial autonomy in order to guide the flow out of access to extra-territorial self-government domain. Overall, the domain routing protocol BGP deployment should consider the following points: 1. Actual network topology; 2. Equipment performance meets the requirements of BGP starts, usually very large BGP routing table to be increased from 50,000 to more than 100,000 articles are about. This router's memory and high performance requirements; 3. INTERNET access in a number of lines, the lines back up and flow distribution. To the following specific examples will illustrate these issues. Continue discussion, we first mention a few concepts: Core layer: high-speed switching capacity of backbone networks in the network core. Distribution layer: The outer core layer is usually used to implement various routing strategy or the implementation of access control. Access layer: the outermost layer of the network, in which the router is within the scope of users linked into the ISP's access point. Early stage of development in the network, the network structure is relatively simple (Figure A), ISP only one line connected to the INTERNET. Usually people do not in this simple network using BGP, and will use simple static routing interconnect. This simplifies the router configuration, easy to manage, and reduce the performance requirements of the border router, reducing costs. If you need to enable BGP, operation is simple just enabled on R1 router, BGP, and other routers within the network R1 as the default gateway, so as to achieve injection of external routing purposes within the network. Taking a step forward with the development of the network, ISP need a different network provider, through a number of lines connected with the INTERNET in order to ensure network reliability. BGP then have a real arena. In this case, the network will usually appear the following structure. Figure II, Figure III. Two network structures seem similar to Figure III in the network has only increased by a router to exclude the router R1 in Figure II single point of failure may be produced, but in how to deploy BGP routing Shique completely different. In the second picture just configured on R1 BGP, and other routers within the network R1 as the default gateway, and through the BGP routing protocol provided by the Weight of this route parameters, adjusting the net flow in the distribution of the two lines . This line can achieve mutual backup, but also effectively regulate the flow distribution. Have two routers with external routes, usually within the route to ensure the consistency required in the figure between routers R1 and R2 IBGP connection to establish a consistent BGP routing table. In the process, but also the use of artificially provided by the routing protocol BGP Local-Preference the routing parameters to optimize routing to control the distribution of data traffic on the line. But how will these external routes to other routers within HowNet report, there are two simple ways: 1. The network structure is relatively simple, the network is not the distribution layer, the core layer directly with the access layer. Then the access layer routers often can not transmit large amounts of user data in the same time, meet the performance requirements needed to enable BGP. In this case, only based on the distribution of traffic will be divided into groups of access routers within the network, different groups will be the default gateway point to a different border gateway routers. This design only as a network structure is imperfect, device performance is not high when the temporary program, not as a permanent program. Because it often requires human intervention, based on changes in network traffic division of the router to re-group, otherwise there will be a router overload, while the other is a lighter load, and thereby can not effectively use the equipment resources. And when Internet lines increases, this design will face more problems. 2. The best way is first to establish a complete network structure, network structure should have the core layer, distribution layer and access layer. IBGP routing protocols by injecting the external routes to the distribution layer router, as shown in Figure II routers R3, R4. Then the distribution layer access layer router as a region of the default gateway router, the access to extra-territorial distribution of the load evenly to each distribution layer router from a router to reduce the pressure on the concrete. This design is becoming more perfect network structure, network size widening ISP final choice. CISCO company has recently put forward a different perspective from the BGP routing protocol deployment of new ideas, as an internal routing protocol using IBGP to exchange routing information within the network users. And in the network access layer IBGP routing protocol enabled on the router, and point users to IBGP in the distribution of static routes. This is mainly on account, BGP routing changes in the routing protocol, only a change in the route update would not like OSPF and ISIS, recalculate the SPF database. Thus the use of this feature reduces the BGP routing convergence time and improve network stability. However, using this method, you should pay attention to the performance requirements of BGP on the router should avoid injecting the external routes to the access layer router, cause the router to work performance. In summary, we briefly discuss how to deploy BGP on the ISP's network routing protocols. In fact, BGP application focus and strength is its ability to control the routing of information to achieve data flow control and distribution. This is a very complex task, to be based on the specific circumstances of the case, not much talked about in this article. But one thing to note, just to rely on BGP's own means to meet the actual needs of different is not feasible, the need to work together with the interconnection partners to achieve, because many parameters need to BGP in the Internet the two sides agreed to take effect. Commercial value of China Telecom and China Netcom Internet longstanding North-South This performance of the site, that is, access to the site with Netcom, the use of telecommunications lines slow access speed Internet users, the use of telecommunications line access sites, use the CNC line users to access extremely slow. To solve this problem, currently only two ways: First, similar to large Internet companies like Baidu, the country set the mirror server, a distributed network topology for fast access, but the drawback is that large investment and tens of millions of funding is not an ordinary business can afford; the other is access to the server at the same time Telecom, China Netcom lane, pay double the network costs. Has a customer base of siting the company will be able to afford such costs. However, some web hosting providers because of cost considerations, often quietly determined customers Internet access lines, if a customer online is CNC line, he also access network server lines, but the telecommunications access of Internet users access the customer's site visit speed is very slow; If the client access the telecommunications lines, he can access the server telecommunication lines, but Netcom Internet users access Web site to access the customer's access speed is very slow; and all this, customers themselves are unaware. Want to know: Your site is for the whole of China and the world, not a direct impact on network access speed of your website traffic. Use of BGP can be a good multi-line access technology to solve the above problems, such as BGP-lane, the use of network bandwidth BGP export of technology with the best of China, Beijing Telecom, Beijing Netcom, the country north and south to ensure interoperability, completely solved the Southern Telecom North Netcom interconnection issues. PE set for the technical engineer, BGP is a great difficulty setting, here are BGP multi-line access basic settings: router bgp 100 no synchronization no bgp default ipv4-unicast bgp log-neighbor-changes neighbor 10.1.1.4 remote -as 100 neighbor 10.1.1.4 update-source Loopback0 neighbor 10.1.1.4 activate neighbor 10.3.1.4 remote-as 100 neighbor 10.3.1.4 update-source Loopback0 neighbor 10.3.1.4 activate no auto-summary! address-family ipv4 vrf vrf-1 redistribute connected no auto-summary no synchronization exit-address-family! address-family vpnv4 neighbor 10.1.1.4 activate neighbor 10.1.1.4 send-community both neighbor 10.3.1.4 activate neighbor 10.3.1.4 send-community both exit-address-family! Routing Principles 1. Weight of the route has the largest selection (range 0 to 65,535) A: weight is a CISCO proprietary parameters, the weight after the router is configured in the local force. 2. LOCAL_PREF the maximum value of preferred route (range 0 to 4,294,967,295). 3. Preferred route originating from the router (including the redistribution of the local network configuration, or in the IGP table that need to be configured to have the address of route aggregation in the BGP routing configuration using the command Aggregate polymerization) 4. preferred route has the shortest AS_PATH A. If the configuration of the Bgp bestpath as-path ignore, then this step is ignored B. an AS path set is used as a AS, regardless of the number in the set AS. CA S path length does not include AS_CONFED_SEQUENCE. 5. According to Origin property. Preferred route with the lowest origin type (IGP> EG> Incomplete) 6. Preferred route smallest MED value (range 0 to 4,294,967,295). A. Only path to get through the first two AS (peer) only when the same AS MED comparison; any sub-coalition of autonomous systems will be ignored. That is, only the first sequence number in the AS AS number of a consistent,
air force one low, only the MED comparison; any coalition AS serial number (AS_CONFED_SEQUENCE) will be ignored. B. If the router is configured bgp always-compare-med, all the paths in the MED comparison. But this requires all AS have also enabled this feature, otherwise routing loops may occur. C. If the router is configured bgp bestpath med confed, will all only be MED AS_CONFED_SEQUENCE compare the path (ie path is derived from the local Union). D. If you receive path is not assigned MED value, the distribution of this path is 0, unless the router is configured bestpath missing-is-worst, will be seen as MED value of 4,294,967,295 will be injected into the routing BGP routing table before being replaced by 4,
nike air force one,294, 967,294. E. Clear BGP MED value of 9 (see later in this chapter 7. Precedence over external routes EBGP Union (confederation) is superior to external routes within the routing IBGP (E-BGP route selection) Note that the path of the alliance include AS_CONFEND_SEQUENCE property only in the local force and therefore be seen as the internal path. Can not distinguish external alliances and internal alliances. 8. Preferred to reach through the nearest IGP neighbor path (selection of the BGP next hop with the lowest IGP metric path); 9. If the router is configured maximum-pathsN, but also from the same peer autonomous / child received a number of autonomous external / external alliances path, you can add up to N path section recently received into the IP routing table. This makes the eBGP paths in the multiple load-sharing. N represents the current maximum number is 6; When this feature is not enabled, the default value is 1. After entering the show ip bgp xxxx after the system output can be seen in the earliest received path is marked as the optimal path, the best path forward in this section to an internal peer, you need to perform the same function with next_hop_self role . 10. If the external route, most preferred the old route (first route to be learned). A. This step can be routed to minimize the impact of swing, because the new path is received will not replace the old, even if this received by the new path is the path mentioned below to the additional selection criteria for choice. This makes the iBGP path only apply additional selection step more meaningful. B. This step can be bgp bestpath compare_routerid command statements are closed. C. If the router is the same sign, this step can be shielded, because it shows where the router is receiving from their routes. D. If the current router is not optimal, this step can be masked. 11. 12. 13.
MBT Men's Chapa And when they had come back | Scribble Print Articles
Linear Mode
