The 802.1x protocol (802.1x for short) was developed by IEEE802 LAN/WAN committee to address security issues of wireless LANs. It was then used in Ethernet as a common access control mechanism for LAN ports to address mainly authentication and security problems.

802.1x is a port-based network access control protocol. It authenticates and controls devices requesting for access in terms of the ports of LAN access devices. With the 802.1x protocol employed, a user-side device can access the LAN only when it passes the authentication. Those fail to pass the authentication are denied when accessing the LAN.

This section covers these topics:

l Architecture of 802.1x Authentication

l The Mechanism of an 802.1x Authentication System

l Encapsulation of EAPoL Messages

l 802.1x Authentication Procedure

l Timers Used in 802.1x

l 802.1x Implementation on an S3100-52P Switch

1.1.1 Architecture of 802.1x Authentication

As shown in Figure 1-1, 802.1x adopts a client/server architecture with three entities: a supplicant system, an authenticator system, and an authentication server system.

Figure 1-1 Architecture of 802.1x authentication

l The supplicant system is an entity residing at one end of a LAN segment and is authenticated by the authenticator system at the other end of the LAN segment. The supplicant system is usually a user terminal device. An 802.1x authentication is triggered when a user launches client program on the supplicant system. Note that the client program must support the extensible authentication protocol over LAN (EAPoL).

l The authenticator system is another entity residing at one end of a LAN segment. It authenticates the connected supplicant systems. The authenticator system is usually an 802.1x-supported network device (such as a H3C series switch). It provides the port (physical or logical) for the supplicant system to access the LAN.

l The authentication server system is an entity that provides authentication service to the authenticator system. Normally in the form of a RADIUS server, the authentication server system serves to perform Authentication, Authorization, and Accounting (AAA) services to users. It also stores user information, such as user name, password, the VLAN a user belongs to, priority, and the Access Control Lists (ACLs) applied.

The four basic concepts related to the above three entities are PAE, controlled port and uncontrolled port, the valid direction of a controlled port and the way a port is controlled.

I. PAE

A port access entity (PAE) is responsible for implementing algorithms and performing protocol-related operations in the authentication mechanism.

l The authenticator system PAE authenticates the supplicant systems when they log into the LAN and controls the status (authorized/unauthorized) of the controlled ports according to the authentication result.

l The supplicant system PAE responds to the authentication requests received from the authenticator system and submits user authentication information to the authenticator system. It also sends authentication requests and disconnection requests to the authenticator system PAE.

II. Controlled port and uncontrolled port

The authenticator system provides ports for supplicant systems to access a LAN. Logically, a port of this kind is divided into a controlled port and an uncontrolled port.

l The uncontrolled port can always send and receive packets. It mainly serves to forward EAPoL packets to ensure that a supplicant system can send and receive authentication requests.

l The controlled port can be used to pass service packets when it is in authorized state. It is blocked when not in authorized state. In this case, no packets can pass through it.

l Controlled port and uncontrolled port are two properties of a port. Packets reaching a port are visible to both the controlled port and uncontrolled port of the port.

III. The valid direction of a controlled port

When a controlled port is in unauthorized state, you can configure it to be a unidirectional port, which sends packets to supplicant systems only.

By default, a controlled port is a unidirectional port.

IV. The way a port is controlled

A port of a H3C series switch can be controlled in the following two ways.

l Port-based authentication. When a port is controlled in this way, all the supplicant systems connected to the port can access the network without being authenticated after one supplicant system among them passes the authentication. And when the authenticated supplicant system goes offline, the others are denied as well.

l MAC-based authentication. All supplicant systems connected to a port have to be authenticated individually in order to access the network. And when a supplicant system goes offline, the others are not affected.

1.1.2 The Mechanism of an 802.1x Authentication System

IEEE 802.1x authentication system uses the Extensible Authentication Protocol (EAP) to exchange information between the supplicant system and the authentication server.

Figure 1-2 The mechanism of an 802.1x authentication system

l EAP protocol packets transmitted between the supplicant system PAE and the authenticator system PAE are encapsulated as EAPoL packets.

l EAP protocol packets transmitted between the authenticator system PAE and the RADIUS server can either be encapsulated as EAP over RADIUS (EAPoR) packets or be terminated at system PAEs. The system PAEs then communicate with RADIUS servers through Password Authentication Protocol (PAP) or Challenge-Handshake Authentication Protocol (CHAP) packets.

l When a supplicant system passes the authentication, the authentication server passes the information about the supplicant system to the authenticator system. The authenticator system in turn determines the state (authorized or unauthorized) of the controlled port according to the instructions (accept or reject) received from the RADIUS server.

1.1.3 Encapsulation of EAPoL Messages

I. The format of an EAPoL packet

EAPoL is a packet encapsulation format defined in 802.1x. To enable EAP protocol packets to be transmitted between supplicant systems and authenticator systems through LANs, EAP protocol packets are encapsulated in EAPoL format. The following figure illustrates the structure of an EAPoL packet.

Figure 1-3 The format of an EAPoL packet

In an EAPoL packet:

l The PAE Ethernet type field holds the protocol identifier. The identifier for 802.1x is 0x888E.

l The Protocol version field holds the version of the protocol supported by the sender of the EAPoL packet.

l The Type field can be one of the following:

00: Indicates that the packet is an EAP-packet, which carries authentication information.

01: Indicates that the packet is an EAPoL-start packet, which initiates the authentication.

02: Indicates that the packet is an EAPoL-logoff packet, which sends logging off requests.

03: Indicates that the packet is an EAPoL-key packet, which carries key information.

04: Indicates that the packet is an EAPoL-encapsulated-ASF-Alert packet, which is used to support the alerting messages of Alerting Standards Forum (ASF).

l The Length field indicates the size of the Packet body field. A value of 0 indicates that the Packet Body field does not exist.

l The Packet body field differs with the Type field.

Note that EAPoL-Start, EAPoL-Logoff, and EAPoL-Key packets are only transmitted between the supplicant system and the authenticator system. EAP packets are encapsulated by RADIUS protocol to allow them successfully reach the authentication servers. Network management-related information (such as alarming information) is encapsulated in EAPoL-Encapsulated-ASF-Alert packets, which are terminated by authenticator systems.

II. The format of an EAP packet

For an EAPoL packet with the value of the Type field being EAP-packet, its Packet body field is an EAP packet, whose format is illustrated in Figure 1-4.

Figure 1-4 The format of an EAP packet

In an EAP packet:

l The Code field indicates the EAP packet type, which can be Request, Response, Success, or Failure.

l The Identifier field is used to match a Response packet with the corresponding Request packet.

l The Length field indicates the size of an EAP packet, which includes the Code, Identifier, Length, and Data fields.

l The Data field carries the EAP packet, whose format differs with the Code field.

A Success or Failure packet does not contain the Data field, so the Length field of it is 4.

Figure 1-5 shows the format of the Data field of a Request packet or a Response packet.

Figure 1-5 The format of the Data field of a Request packet or a Response packet

l The Type field indicates the EAP authentication type. A value of 1 indicates Identity and that the packet is used to query the identity of the peer. A value of 4 represents MD5-Challenge (similar to PPP CHAP) and indicates that the packet includes query information.

l The Type Date field differs with types of Request and Response packets.

III. Newly added fields for EAP authentication

Two fields, EAP-message and Message-authenticator, are added to a RADIUS protocol packet for EAP authentication. (Refer to the Introduction to RADIUS protocol section in the AAA Operation for information about the format of a RADIUS protocol packet.)

The EAP-message field, whose format is shown in Figure 1-6, is used to encapsulate EAP packets. The maximum size of the string field is 253 bytes. EAP packets with their size larger than 253 bytes are fragmented and are encapsulated in multiple EAP-message fields. The type code of the EAP-message field is 79.

Figure 1-6 The format of an EAP-message field

The Message-authenticator field, whose format is shown in Figure 1-7, is used to prevent unauthorized interception to access requesting packets during authentications using CHAP, EAP, and so on. A packet with the EAP-message field must also have the Message-authenticator field. Otherwise, the packet is regarded as invalid and is discarded.

Figure 1-7 The format of an Message-authenticator field

1.1.4 802.1x Authentication Procedure

A H3C S3100-52P Ethernet switch can authenticate supplicant systems in EAP terminating mode or EAP relay mode.

I. EAP relay mode

This mode is defined in 802.1x. In this mode, EAP packets are encapsulated in higher level protocol (such as EAPoR) packets to enable them to successfully reach the authentication server. Normally, this mode requires that the RADIUS server support the two newly-added fields: the EAP-message field (with a value of 79) and the Message-authenticator field (with a value of 80).

Four authentication ways, namely EAP-MD5, EAP-TLS (transport layer security), EAP-TTLS (tunneled transport layer security), and Protected Extensible Authentication Protocol (PEAP), are available in the EAP relay mode.

l EAP-MD5 authenticates the supplicant system. The RADIUS server sends MD5 keys (contained in EAP-request/MD5 challenge packets) to the supplicant system, which in turn encrypts the passwords using the MD5 keys.

l EAP-TLS allows the supplicant system and the RADIUS server to check each other’s security certificate and authenticate each other’s identity, guaranteeing that data is transferred to the right destination and preventing data from being intercepted.

l EAP-TTLS is a kind of extended EAP-TLS. EAP-TLS implements bidirectional authentication between the client and authentication server. EAP-TTLS transmit message using a tunnel established using TLS.

l PEAP creates and uses TLS security channels to ensure data integrity and then performs new EAP negotiations to verify supplicant systems.

Figure 1-8 describes the basic EAP-MD5 authentication procedure.

Figure 1-8 802.1x authentication procedure (in EAP relay mode)

The detailed procedure is as follows:

l A supplicant system launches an 802.1x client to initiate an access request by sending an EAPoL-start packet to the switch, with its user name and password provided. The 802.1x client program then forwards the packet to the switch to start the authentication process.

l Upon receiving the authentication request packet, the switch sends an EAP-request/identity packet to ask the 802.1x client for the user name.

l The 802.1x client responds by sending an EAP-response/identity packet to the switch with the user name contained in it. The switch then encapsulates the packet in a RADIUS Access-Request packet and forwards it to the RADIUS server.

l Upon receiving the packet from the switch, the RADIUS server retrieves the user name from the packet, finds the corresponding password by matching the user name in its database, encrypts the password using a randomly-generated key, and sends the key to the switch through an RADIUS access-challenge packet. The switch then sends the key to the 802.1x client.

l Upon receiving the key (encapsulated in an EAP-request/MD5 challenge packet) from the switch, the client program encrypts the password of the supplicant system with the key and sends the encrypted password (contained in an EAP-response/MD5 challenge packet) to the RADIUS server through the switch. (Normally, the encryption is irreversible.)

l The RADIUS server compares the received encrypted password (contained in a RADIUS access-request packet) with the locally-encrypted password. If the two match, it will then send feedbacks (through a RADIUS access-accept packet and an EAP-success packet) to the switch to indicate that the supplicant system is authenticated.

l The switch changes the state of the corresponding port to accepted state to allow the supplicant system to access the network.

l The supplicant system can also terminate the authenticated state by sending EAPoL-Logoff packets to the switch. The switch then changes the port state from accepted to rejected.

& Note:

In EAP relay mode, packets are not modified during transmission. Therefore if one of the four ways are used (that is, PEAP, EAP-TLS, EAP-TTLS or EAP-MD5) to authenticate, ensure that the authenticating ways used on the supplicant system and the RADIUS server are the same. However for the switch, you can simply enable the EAP relay mode by using the dot1x authentication-method eap command.

II. EAP terminating mode

In this mode, EAP packet transmission is terminated at authenticator systems and the EAP packets are converted to RADIUS packets. Authentication and accounting are carried out through RADIUS protocol.

In this mode, PAP or CHAP is employed between the switch and the RADIUS server. Figure 1-9 illustrates the authentication procedure (assuming that CHAP is employed between the switch and the RADIUS server).

Figure 1-9 802.1x authentication procedure (in EAP terminating mode)

The authentication procedure in EAP terminating mode is the same as that in the EAP relay mode except that the randomly-generated key in the EAP terminating mode is generated by the switch, and that it is the switch that sends the user name, the randomly-generated key, and the supplicant system-encrypted password to the RADIUS server for further authentication.

1.1.5 Timers Used in 802.1x

In 802.1 x authentication, the following timers are used to ensure that the supplicant system, the switch, and the RADIUS server interact in an orderly way.

l Handshake timer (handshake-period). This timer sets the handshake period and is triggered after a supplicant system passes the authentication. It sets the interval for a switch to send handshake request packets to online users. You can set the maximum number of transmission attempts by using the dot1x retry command. An online user will be considered offline when the switch has not received any response packets after the maximum number of handshake request transmission attempts is reached.

l Quiet-period timer (quiet-period). This timer sets the quiet-period. When a supplicant system fails to pass the authentication, the switch quiets for the set period (set by the quiet-period timer) before it processes another authentication request re-initiated by the supplicant system. During this quiet period, the switch does not perform any 802.1x authentication-related actions for the supplicant system.

l Re-authentication timer (reauth-period). The switch initiates 802.1x re-authentication at the interval set by the re-authentication timer.

l RADIUS server timer (server-timeout). This timer sets the server-timeout period. After sending an authentication request packet to the RADIUS server, the switch sends another authentication request packet if it does not receive the response from the RADIUS server when this timer times out.

l Supplicant system timer (supp-timeout). This timer sets the supp-timeout period and is triggered by the switch after the switch sends a request/challenge packet to a supplicant system. The switch sends another request/challenge packet to the supplicant system if the switch does not receive the response from the supplicant system when this timer times out.

l Transmission timer (tx-period). This timer sets the tx-period and is triggered by the switch in two cases. The first case is when the client requests for authentication. The switch sends a unicast request/identity packet to a supplicant system and then triggers the transmission timer. The switch sends another request/identity packet to the supplicant system if it does not receive the reply packet from the supplicant system when this timer times out. The second case is when the switch authenticates the 802.1x client who cannot request for authentication actively. The switch sends multicast request/identity packets periodically through the port enabled with 802.1x function. In this case, this timer sets the interval to send the multicast request/identity packets.

l Client version request timer (ver-period). This timer sets the version period and is triggered after a switch sends a version request packet. The switch sends another version request packet if it does receive version response packets from the supplicant system when the timer expires.

1.1.6 802.1x Implementation on an S3100-52P Switch

In addition to the earlier mentioned 802.1x features, an S3100-52P switch is also capable of the following:

l Checking supplicant systems for proxies, multiple network adapters, and so on (This function needs the cooperation of a CAMS server.)

l Checking client version

l The guest VLAN function

& Note:

H3C's CAMS Server is a service management system used to manage networks and to secure networks and user information. With the cooperation of other networking devices (such as switches) in the network, a CAMS server can implement the AAA functions and rights management.

I. Checking the supplicant system

An S3100-52P switch checks:

l Supplicant systems logging on through proxies

l Supplicant systems logging on through IE proxies

l Whether or not a supplicant system logs in through more than one network adapters (that is, whether or not more than one network adapters are active in a supplicant system when the supplicant system logs in).

In response to any of the three cases, a switch can optionally take the following measures:

l Only disconnects the supplicant system but sends no Trap packets.

l Sends Trap packets without disconnecting the supplicant system.

This function needs the cooperation of 802.1x client and a CAMS server.

l The 802.1x client needs to be capable of detecting multiple network adapters, proxies, and IE proxies.

l The CAMS server is configured to disable the use of multiple network adapters, proxies, or IE proxies.

By default, an 802.1x client program allows use of multiple network adapters, proxies, and IE proxies. In this case, if the CAMS server is configured to disable use of multiple network adapters, proxies, or IE proxies, it prompts the 802.1x client to disable use of multiple network adapters, proxies, or IE proxies through messages after the supplicant system passes the authentication.

& Note:

l The client-checking function needs the support of H3C’s 802.1x client program.

l To implement the proxy detecting function, you need to enable the function on both the 802.1x client program and the CAMS server in addition to enabling the client version detecting function on the switch by using the dot1x version-check command.

II. Checking the client version

With the 802.1x client version-checking function enabled, a switch checks the version and validity of an 802.1x client to prevent unauthorized users or users with earlier versions of 802.1x client from logging in.

This function makes the switch to send version-requesting packets again if the 802.1x client fails to send version-reply packet to the switch when the version-checking timer times out.

& Note:

The 802.1x client version-checking function needs the support of H3C’s 802.1x client program.

III. The guest VLAN function

The guest VLAN function enables supplicant systems that are not authenticated to access network resources in a restrained way.

The guest VLAN function enables supplicant systems that do not have 802.1x client installed to access specific network resources. It also enables supplicant systems that are not authenticated to upgrade their 802.1x client programs.

With this function enabled:

l The switch sends authentication triggering request (EAP-Request/Identity) packets to all the 802.1x-enabled ports.

l After the maximum number retries have been made and there are still ports that have not sent any response back, the switch will then add these ports to the guest VLAN.

l Users belonging to the guest VLAN can access the resources of the guest VLAN without being authenticated. But they need to be authenticated when accessing external resources.

Normally, the guest VLAN function is coupled with the dynamic VLAN delivery function.

Refer to AAA Operation for detailed information about the dynamic VLAN delivery function.

IV. Enabling 802.1x re-authentication

802.1x re-authentication is timer-triggered or packet-triggered. It re-authenticates users who have passed authentication. With 802.1x re-authentication enabled, the switch can monitor the connection status of users periodically. If the switch receives no re-authentication response from a user in a period of time, it tears down the connection to the user. To connect to the switch again, the user needs to initiate 802.1x authentication with the client software again.

& Note:

l When re-authenticating a user, a switch goes through the complete authentication process. It transmits the username and password of the user to the server. The server may authenticate the username and password, or, however, use re-authentication for only accounting and user connection status checking and therefore does not authenticate the username and password any more.

l An authentication server running CAMS authenticates the username and password during re-authentication of a user in the EAP authentication mode but does not in PAP or CHAP authentication mode.

Figure 1-10 802.1x re-authentication

802.1x re-authentication can be enabled in one of the following two ways:

l The RADIUS server has the switch perform 802.1x re-authentication of users. The RADIUS server sends the switch an Access-Accept packet with the Termination-Action attribute field of 1. Upon receiving the packet, the switch re-authenticates the user periodically.

l You enable 802.1x re-authentication on the switch. With 802.1x re-authentication enabled, the switch re-authenticates users periodically.

& Note:

802.1x re-authentication will fail if a CAMS server is used and configured to perform authentication but not accounting. This is because a CAMS server establishes a user session after it begins to perform accounting. Therefore, to enable 802.1x re-authentication, do not configure the accounting none command in the domain. This restriction does not apply to other types of servers.

1.2 Introduction to 802.1x Configuration

802.1x provides a solution for authenticating users. To implement this solution, you need to execute 802.1x-related commands. You also need to configure AAA schemes on switches and specify the authentication scheme (RADIUS or local authentication scheme).

Figure 1-11 802.1x configuration

l 802.1x users use domain names to associate with the ISP domains configured on switches

l Configure the AAA scheme (a local authentication scheme or a RADIUS scheme) to be adopted in the ISP domain.

l If you specify to use a local authentication scheme, you need to configure the user names and passwords manually on the switch. Users can pass the authentication through 802.1x client if they provide user names and passwords that match those configured on the switch.

l If you specify to adopt the RADIUS scheme, the supplicant systems are authenticated by a remote RADIUS server. In this case, you need to configure user names and passwords on the RADIUS server and perform RADIUS client-related configuration on the switches.

l You can also specify to adopt the RADIUS authentication scheme, with a local authentication scheme as a backup. In this case, the local authentication scheme is adopted when the RADIUS server fails.

Refer to the AAA Operation for detailed information about AAA scheme configuration.

1.3 Basic 802.1x Configuration

1.3.1 Configuration Prerequisites

l Configure ISP domain and the AAA scheme to be adopted. You can specify a RADIUS scheme or a local scheme.

l Ensure that the service type is configured as lan-access (by using the service-type command) if local authentication scheme is adopted.

1.3.2 Configuring Basic 802.1x Functions

Follow these steps to configure basic 802.1x functions:

To do…		Use the command…	Remarks
Enter system view		system-view	—
Enable 802.1x globally		dot1x	Required By default, 802.1x is disabled globally.
Enable 802.1x for specified ports	In system view	dot1x interface interface-list	Required By default, 802.1x is disabled on all ports.
	In port view	interface interface-type interface-number
		dot1x
		quit
Set port access control mode for specified ports	In system view	dot1x port-control { authorized-force \| unauthorized-force \| auto } [ interface interface-list ]	Optional By default, an 802.1x-enabled port operates in the auto mode.
	In port view	interface interface-type interface-number
		dot1x port-control { authorized-force \| unauthorized-force \| auto }
		quit
Set port access method for specified ports		dot1x port-method { macbased \| portbased } [ interface interface-list ]	Optional The default port access method is MAC-address-based (that is, the macbased keyword is used by default).
		interface interface-type interface-number
		dot1x port-method { macbased \| portbased }
		quit
Set authentication method for 802.1x users		dot1x authentication-method { chap \| pap \| eap }	Optional By default, a switch performs CHAP authentication in EAP terminating mode.
Enable online user handshaking		dot1x handshake enable	Optional By default, online user handshaking is enabled.
Enter Ethernet port view		interface interface-type interface-number	—
Enable the handshake packet protection function		dot1x handshake secure	Optional By default, the handshake packet protection function is disabled.

Caution:

l 802.1x configurations take effect only after you enable 802.1x both globally and for specified ports.

l The settings of 802.1x and MAC address learning limit are mutually exclusive. Enabling 802.1x on a port will prevent you from setting the limit on MAC address learning on the port and vice versa.

l The settings of 802.1x and aggregation group member are mutually exclusive. Enabling 802.1x on a port will prevent you from adding the port to an aggregation group and vice versa.

l When a device operates as an authentication server, its authentication method for 802.1x users cannot be configured as EAP.

l With the support of the H3C proprietary client, handshake packets are used to test whether or not a user is online.

l As clients that are not of H3C do not support the online user handshaking function, switches cannot receive handshake acknowledgement packets from them in handshaking periods. To prevent users being falsely considered offline, you need to disable the online user handshaking function in this case.

l The handshake packet protection function requires the cooperation of the client and the authentication server. If either of the two ends does not support the function, you need to disable it on the other one.

1.3.3 Timer and Maximum User Number Configuration

Follow these steps to configure 802.1x timers and the maximum number of users:

To do…		Use the command...	Remarks
Enter system view		system-view	—
Set the maximum number of concurrent on-line users for specified ports	In system view	dot1x max-user user-number [ interface interface-list ]	Optional By default, a port can accommodate up to 256 users at a time.
	In port view	interface interface-type interface-number
		dot1x max-user user-number
		quit
Set the maximum retry times to send request packets		dot1x retry max-retry-value	Optional By default, the maximum retry times to send a request packet is 2. That is, the authenticator system sends a request packet to a supplicant system for up to two times by default.
Set 802.1x timers		dot1x timer { handshake-period handshake-period-value \| quiet-period quiet-period-value \| server-timeout server-timeout-value \| supp-timeout supp-timeout-value \| tx-period tx-period-value \| ver-period ver-period-value }	Optional The settings of 802.1x timers are as follows. 1) handshake-period-value: 15 seconds 2) quiet-period-value: 60 seconds 3) server-timeout-value: 100 seconds 4) supp-timeout-value: 30 seconds 5) tx-period-value: 30 seconds 6) ver-period-value: 30 seconds
Enable the quiet-period timer		dot1x quiet-period	Optional By default, the quiet-period timer is disabled.

& Note:

l As for the dot1x max-user command, if you execute it in system view without specifying the interface-list argument, the command applies to all ports. You can also use this command in port view. In this case, this command applies to the current port only and the interface-list argument is not needed.

l As for the configuration of 802.1x timers, the default values are recommended.

1.4 Advanced 802.1x Configuration

Advanced 802.1x configurations, as listed below, are all optional.

l Configuration concerning CAMS, including multiple network adapters detecting, proxy detecting, and so on.

l Client version checking configuration

l DHCP–triggered authentication

l Guest VLAN configuration

l 802.1x re-authentication configuration

l Configuration of the 802.1x re-authentication timer

You need to configure basic 802.1x functions before configuring the above 802.1x features.

1.4.1 Configuring Proxy Checking

Follow these steps to configure proxy checking:

To do...		Use the command...	Remarks
Enter system view		system-view	—
Enable proxy checking function globally		dot1x supp-proxy-check { logoff \| trap }	Required By default, the 802.1x proxy checking function is globally disabled.
Enable proxy checking for a port/specified ports	In system view	dot1x supp-proxy-check { logoff \| trap } [ interface interface-list ]	Required By default, the 802.1x proxy checking is disabled on a port.
	In port view	interface interface-type interface-number
		dot1x supp-proxy-check { logoff \| trap }
		quit

& Note:

l The proxy checking function needs the cooperation of H3C's 802.1x client (iNode) program.

l The proxy checking function depends on the online user handshaking function. To enable the proxy detecting function, you need to enable the online user handshaking function first.

l The configuration listed in the above table takes effect only when it is performed on CAMS as well as on the switch. In addition, the client version checking function needs to be enabled on the switch too (by using the dot1x version-check command).

1.4.2 Configuring Client Version Checking

Follow these steps to configure client version checking:

To do...		Use the command...	Remarks
Enter system view		system-view	—
Enable 802.1x client version checking	In system view	dot1x version-check [ interface interface-list ]	Required By default, 802.1x client version checking is disabled on a port.
	In port view	interface interface-type interface-number
		dot1x version-check
		quit
Set the maximum number of retires to send version checking request packets		dot1x retry-version-max max-retry-version-value	Optional By default, the maximum number of retires to send version checking request packets is 3.
Set the client version checking period timer		dot1x timer ver-period ver-period-value	Optional By default, the timer is set to 30 seconds.

& Note:

As for the dot1x version-user command, if you execute it in system view without specifying the interface-list argument, the command applies to all ports. You can also execute this command in port view. In this case, this command applies to the current port only and the interface-list argument is not needed.

1.4.3 Enabling DHCP-triggered Authentication

After performing the following configuration, 802.1x allows running DHCP on access users, and users are authenticated when they apply for dynamic IP addresses through DHCP.

Follow these steps to enable DHCP-triggered authentication:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable DHCP-triggered authentication	dot1x dhcp-launch	Required By default, DHCP-triggered authentication is disabled.

To do...

Use the command...

Remarks

Enter system view

system-view

—

Enable DHCP-triggered authentication

dot1x dhcp-launch

Required

By default, DHCP-triggered authentication is disabled.

1.4.4 Configuring Guest VLAN

Follow these steps to configure guest VLAN:

To do...		Use the command...	Remarks
Enter system view		system-view	—
Configure port access method		dot1x port-method portbased	Required The default port access method is MAC-address-based. That is, the macbased keyword is used by default.
Enable the guest VLAN function	In system view	dot1x guest-vlan vlan-id [ interface interface-list ]	Required By default, the guest VLAN function is disabled.
	In port view	interface interface-type interface-number
		dot1x guest-vlan vlan-id
		quit

Caution:

l The guest VLAN function is available only when the switch operates in the port-based authentication mode.

l Only one guest VLAN can be configured for each switch.

l The guest VLAN function cannot be implemented if you configure the dot1x dhcp-launch command on the switch to enable DHCP-triggered authentication. This is because the switch does not send authentication packets in that case.

1.4.5 Configuring 802.1x Re-Authentication

Follow these steps to enable 802.1x re-authentication:

To do...		Use the command...	Remarks
Enter system view		system-view	—
Enable 802.1x re-authentication on port(s)	In system view	dot1x re-authenticate [ interface interface-list ]	Required By default, 802.1x re-authentication is disabled on a port.
Enable 802.1x re-authentication on port(s)	In port view	dot1x re-authenticate

& Note:

l To enable 802.1x re-authentication on a port, you must first enable 802.1x globally and on the port.

l An authentication server running CAMS authenticates the username and password during re-authentication of a user in the EAP authentication mode but does not in PAP or CHAP authentication mode.

1.4.6 Configuring the 802.1x Re-Authentication Timer

After 802.1x re-authentication is enabled on the switch, the switch determines the re-authentication interval in one of the following two ways:

1) The switch uses the value of the Session-timeout attribute field of the Access-Accept packet sent by the RADIUS server as the re-authentication interval.

2) The switch uses the value configured with the dot1x timer reauth-period command as the re-authentication interval for access users.

Note the following:

During re-authentication, the switch always uses the latest re-authentication interval configured, no matter which of the above-mentioned two ways is used to determine the re-authentication interval. For example, if you configure a re-authentication interval on the switch and the switch receives an Access-Accept packet whose Termination-Action attribute field is 1, the switch will ultimately use the value of the Session-timeout attribute field as the re-authentication interval.

The following introduces how to configure the 802.1x re-authentication timer on the switch.

Follow these steps to configure the re-authentication interval:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Configure a re-authentication interval	dot1x timer reauth-period reauth-period-value	Optional By default, the re-authentication interval is 3,600 seconds.

To do...

Use the command...

Remarks

Enter system view

system-view

—

Configure a re-authentication interval

dot1x timer reauth-period reauth-period-value

Optional

By default, the re-authentication interval is 3,600 seconds.

1.5 Displaying and Maintaining 802.1x Configuration

To do...	Use the command...	Remarks
Display the configuration, session, and statistics information about 802.1x	display dot1x [ sessions \| statistics ] [ interface interface-list ]	Available in any view
Clear 802.1x-related statistics information	reset dot1x statistics [ interface interface-list ]	Available in user view

1.6 Configuration Example

1.6.1 802.1x Configuration Example

I. Network requirements

l Authenticate users on all ports to control their accesses to the Internet. The switch operates in MAC-based access control mode.

l All supplicant systems that pass the authentication belong to the default domain named “aabbcc.net”. The domain can accommodate up to 30 users. As for authentication, a supplicant system is authenticated locally if the RADIUS server fails. And as for accounting, a supplicant system is disconnected by force if the RADIUS server fails. The name of an authenticated supplicant system is not suffixed with the domain name. A connection is terminated if the total size of the data passes through it during a period of 20 minutes is less than 2,000 bytes.

l The switch is connected to a server comprising of two RADIUS servers whose IP addresses are 10.11.1.1 and 10.11.1.2. The RADIUS server with an IP address of 10.11.1.1 operates as the primary authentication server and the secondary accounting server. The other operates as the secondary authentication server and primary accounting server. The password for the switch and the authentication RADIUS servers to exchange message is “name”. And the password for the switch and the accounting RADIUS servers to exchange message is “money”. The switch sends another packet to the RADIUS servers again if it sends a packet to the RADIUS server and does not receive response for 5 seconds, with the maximum number of retries of 5. And the switch sends a real-time accounting packet to the RADIUS servers once in every 15 minutes. A user name is sent to the RADIUS servers with the domain name truncated.

l The user name and password for local 802.1x authentication are “localuser” and “localpass” (in plain text) respectively. The idle disconnecting function is enabled.

II. Network diagram

Figure 1-12 Network diagram for AAA configuration with 802.1x and RADIUS enabled

III. Configuration procedure

& Note:

Following configuration covers the major AAA/RADIUS configuration commands. Refer to AAA Operation for the information about these commands. Configuration on the client and the RADIUS servers is omitted.

# Enable 802.1x globally.

<Sysname> system-view

System View: return to User View with Ctrl+Z.

[Sysname] dot1x

# Enable 802.1x on Ethernet 1/0/1.

[Sysname] dot1x interface Ethernet 1/0/1

# Set the access control method to MAC-based (This operation can be omitted, as MAC-based is the default).

[Sysname] dot1x port-method macbased interface Ethernet 1/0/1

# Create a RADIUS scheme named “radius1” and enter RADIUS scheme view.

[Sysname] radius scheme radius1

# Assign IP addresses to the primary authentication and accounting RADIUS servers.

[Sysname-radius-radius1] primary authentication 10.11.1.1

[Sysname-radius-radius1] primary accounting 10.11.1.2

# Assign IP addresses to the secondary authentication and accounting RADIUS server.

[Sysname-radius-radius1] secondary authentication 10.11.1.2

[Sysname-radius-radius1] secondary accounting 10.11.1.1

# Set the password for the switch and the authentication RADIUS servers to exchange messages.

[Sysname-radius-radius1] key authentication name

# Set the password for the switch and the accounting RADIUS servers to exchange messages.

[Sysname-radius-radius1] key accounting money

# Set the interval and the number of the retries for the switch to send packets to the RADIUS servers.

[Sysname-radius-radius1] timer 5

[Sysname-radius-radius1] retry 5

# Set the timer for the switch to send real-time accounting packets to the RADIUS servers.

[Sysname-radius-radius1] timer realtime-accounting 15

# Configure to send the user name to the RADIUS server with the domain name truncated.

[Sysname-radius-radius1] user-name-format without-domain

[Sysname-radius-radius1] quit

# Create the domain named “aabbcc.net” and enter its view.

[Sysname] domain aabbcc.net

# Specify to adopt radius1 as the RADIUS scheme of the user domain. If RADIUS server is invalid, specify to adopt the local authentication scheme.

[Sysname-isp-aabbcc.net] scheme radius-scheme radius1 local

# Specify the maximum number of users the user domain can accommodate to 30.

[Sysname-isp-aabbcc.net] access-limit enable 30

# Enable the idle disconnecting function and set the related parameters.

[Sysname-isp-aabbcc.net] idle-cut enable 20 2000

[Sysname-isp-aabbcc.net] quit

# Set the default user domain to aabbcc.net.

[Sysname] domain default enable aabbcc.net

# Create a local access user account.

[Sysname] local-user localuser

[Sysname-luser-localuser] service-type lan-access

[Sysname-luser-localuser] password simple localpass

Chapter 2 Quick EAD Deployment Configuration

When configuring quick EAD deployment, go to these sections for information you are interested in:

l Introduction to Quick EAD Deployment

l Configuring Quick EAD Deployment

l Displaying and Maintaining Quick EAD Deployment

l Quick EAD Deployment Configuration Example

l Troubleshooting

2.1 Introduction to Quick EAD Deployment

2.1.1 Quick EAD Deployment Overview

As an integrated solution, an Endpoint Admission Defense (EAD) solution can improve the overall defense power of a network. In real applications, however, deploying EAD clients proves to be time consuming and inconvenient.

To address the issue, the H3C S3100-52P provides the forcible deployment of EAD clients with 802.1x authentication, easing the work of EAD client deployment.

2.1.2 Operation of Quick EAD Deployment

Quick EAD deployment is achieved with the two functions: restricted access and HTTP redirection.

I. Restricted access

Before passing 802.1x authentication, a user is restricted (through ACLs) to a specific range of IP addresses or a specific server. Services like EAD client upgrading/download and dynamic address assignment are available on the specific server.

II. HTTP redirection

In the HTTP redirection approach, when the terminal users that have not passed 802.1x authentication access the Internet through Internet Explorer, they are redirected to a predefined URL for EAD client download.

The two functions ensure that all the users without an EAD client have downloaded and installed one from the specified server themselves before they can access the Internet, thus decreasing the complexity and effort that EAD client deployment may involve.

& Note:

The quick EAD deployment feature takes effect only when the access control mode of an 802.1x-enabled port is set to auto.

2.2 Configuring Quick EAD Deployment

2.2.1 Configuration Prerequisites

l Enable 802.1x on the switch.

l Set the access mode to auto for 802.1x-enabled ports.

2.2.2 Configuration Procedure

I. Configuring a free IP range

A free IP range is an IP range that users can access before passing 802.1x authentication.

Follow these steps to configure a free IP range:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Configure the URL for HTTP redirection	dot1x url url-string	Required
Configure a free IP range	dot1x free-ip ip-address { mask-address \| mask-length }	Required By default, no free IP range is configured.

Caution:

l You must configure the URL for HTTP redirection before configuring a free IP range. A URL must start with http:// and the segment where the URL resides must be in the free IP range. Otherwise, the redirection function cannot take effect.

l You must disable the DHCP-triggered authentication function of 802.1x before configuring a free IP range.

l With dot1x enabled but quick EAD deployment disabled, users cannot access the DHCP server if they fail 802.1x authentication. With quick EAD deployment enabled, users can obtain IP addresses dynamically before passing authentication if the IP address of the DHCP server is in the free IP range.

l The quick EAD deployment function applies to only ports with the access control mode set to auto through the dot1x port-control command.

l At present, 802.1x is the only access approach that supports quick EAD deployment.

l Currently, the quick EAD deployment function does not support port security. The configured free IP range cannot take effect if you enable port security.

II. Setting the ACL timeout period

The quick EAD deployment function depends on ACLs in restricting access of users failing authentication. Each online user that has not passed authentication occupies a certain amount of ACL resources. After a user passes authentication, the occupied ACL resources will be released. When a large number of users log in but cannot pass authentication, the switch may run out of ACL resources, preventing other users from logging in. A timer called ACL timer is designed to solve this problem.

You can control the usage of ACL resources by setting the ACL timer. The ACL timer starts once a user gets online. If the user has not passed authentication when the ACL timer expires, the occupied ACL resources are released for other users to use. When a tremendous of access requests are present, you can decrease the timeout period of the ACL timer appropriately for higher utilization of ACL resources.

Follow these steps to configure the ACL timer:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Set the ACL timer	dot1x timer acl-timeout acl-timeout-value	Required By default, the ACL timeout period is 30 minutes.

To do...

Use the command...

Remarks

Enter system view

system-view

—

Set the ACL timer

dot1x timer acl-timeout acl-timeout-value

Required

By default, the ACL timeout period is 30 minutes.

2.2.3 Displaying and Maintaining Quick EAD Deployment

To do...	Use the command...	Remarks
Display configuration information about quick EAD deployment	display dot1x [ sessions \| statistics ] [ interface interface-list ]	Available in any view

2.3 Quick EAD Deployment Configuration Example

I. Network requirements

A user connects to the switch directly. The switch connects to the Web server and the Internet. The user will be redirected to the Web server to download the authentication client and upgrade software when accessing the Internet through IE before passing authentication. After passing authentication, the user can access the Internet.

II. Network diagram

Figure 2-1 Network diagram for quick EAD deployment

III. Configuration procedure

& Note:

Before enabling quick EAD deployment, be sure that:

l The Web server is configured properly.

l The default gateway of the user’s PC is configured as the IP address of the connected VLAN interface on the switch.

# Configure the URL for HTTP redirection.

<Sysname> system-view

[Sysname] dot1x url http://192.168.0.111

# Configure a free IP range.

[Sysname] dot1x free-ip 192.168.0.111 24

# Set the ACL timer to 10 minutes.

[Sysname] dot1x timer acl-timeout 10

# Enable dot1x globally.

[Sysname] dot1x

# Enable dot1x for Ethernet 1/0/1.

[Sysname] dot1x interface Ethernet 1/0/1

2.4 Troubleshooting

Symptom: A user cannot be redirected to the specified URL server, no matter what URL the user enters in the IE address bar.

Solution:

l If a user enters an IP address in a format other than the dotted decimal notation, the user may not be redirected. This is related with the operating system used on the PC. In this case, the PC considers the IP address string a name and tries to resolve the name. If the resolution fails, the PC will access a specific website. Generally, this address is not in dotted decimal notation. As a result, the PC cannot receive any ARP response and therefore cannot be redirected. To solve this problem, the user needs to enter an IP address that is not in the free IP range in dotted decimal notation.

l If a user enters an address in the free IP range, the user cannot be redirected. This is because the switch considers that the user wants to access a host in the free IP range, unconcerned about whether this PC exists or not. To solve this problem, the user needs to enter an address not in the free IP range.

l Check that you have configured an IP address in the free IP range for the Web server and a correct URL for redirection, and that the server provides Web services properly.

Chapter 3 HABP Configuration

When configuring HABP, go to these sections for information you are interested in:

l Introduction to HABP

l HABP Server Configuration

l HABP Client Configuration

l Displaying and Maintaining HABP Configuration

3.1 Introduction to HABP

When a switch is configured with the 802.1x function, 802.1x will authenticate and authorize 802.1x-enabled ports and allow only the authorized ports to forward packets. In case a port fails 802.1x authentication and authorization, service packets from and to that port will be blocked, making it impossible to manage the switch attached to the port. The Huawei Authentication Bypass Protocol (HABP) aims at solving this problem.

An HABP packet carries the MAC addresses of the attached switches with it. It can bypass the 802.1x authentications when traveling between HABP-enabled switches, through which management devices can obtain the MAC addresses of the attached switches and thus the management of the attached switches is feasible.

HABP is built on the client-server model. Typically, the HABP server sends HABP requests to the client periodically to collect the MAC address(es) of the attached switch(es). The client responds to the requests, and forwards the HABP requests to the attached switch(es). The HABP server usually runs on the administrative device while the HABP client runs on the attached switches.

For ease of switch management, it is recommended that you enable HABP for 802.1x-enabled switches.

3.2 HABP Server Configuration

With the HABP server launched, a management device sends HABP request packets regularly to the attached switches to collect their MAC addresses. You need also to configure the interval on the management device for an HABP server to send HABP request packets.

Follow these steps to configure an HABP server:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable HABP	habp enable	Optional By default, HABP is enabled.
Configure the current switch to be an HABP server	habp server vlan vlan-id	Required By default, a switch operates as an HABP client after you enable HABP on the switch. If you want to use the switch as a management switch, you need to configure the switch to be an HABP server.
Configure the interval to send HABP request packets.	habp timer interval	Optional The default interval for an HABP server to send HABP request packets is 20 seconds.

3.3 HABP Client Configuration

HABP clients reside on switches attached to HABP servers. After you enable HABP for a switch, the switch operates as an HABP client by default. So you only need to enable HABP on a switch to make it an HABP client.

Follow these steps to configure an HABP client:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable HABP	habp enable	Optional HABP is enabled by default. And a switch operates as an HABP client after you enable HABP for it.

To do...

Use the command...

Remarks

Enter system view

system-view

—

Enable HABP

habp enable

Optional

HABP is enabled by default. And a switch operates as an HABP client after you enable HABP for it.

3.4 Displaying and Maintaining HABP Configuration

To do...	Use the command...	Remarks
Display HABP configuration and status	display habp	Available in any view
Display the MAC address table maintained by HABP	display habp table	Available in any view
Display statistics on HABP packets	display habp traffic	Available in any view

Chapter 4 System Guard Configuration

When configuring System Guard, go to these sections for information you are interested in:

l System Guard Overview

l Configuring System Guard

l Displaying and Maintaining System Guard Configuration

4.1 System Guard Overview

4.1.1 Guard Against IP Attacks

System-guard operates to inspect the IP packets over 10-second intervals for the CPU for suspicious source IP addresses. Once the packets from such an IP address hit the predefined threshold, System Guard does one of the following:

l The switch logs out the host (hereafter referred to as infected host) by automatically applying an ACL rule and waits a certain period of time before resuming forwarding packets for that host.

l If the packets from the infected host need processing by the CPU, the switch decreases the precedence of such packets and discards the packets already delivered to the CPU.

4.1.2 Guard Against TCN Attacks

System Guard monitors the rate at which TCN/TC packets are received on the ports. If a port receives an excessive number of TCN/TC packets within a given period of time, the switch sends only one TCN/TC packet in every 10 seconds to the CPU and discards the rest TCN/TC packets, while outputting trap and log information.

4.1.3 Layer 3 Error Control

With the Layer 3 error control feature enabled, the switch delivers all Layer 3 packets that the switch considers to be error packets to the CPU.

4.2 Configuring System Guard

4.2.1 Configuring System Guard Against IP Attacks

Configuration of System Guard against IP attacks includes these tasks:

l Enabling System Guard against IP attacks

l Setting the maximum number of infected hosts that can be concurrently monitored

l Configuring parameters related to MAC address learning

Follow these steps to configure System Guard against IP attacks:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable System Guard against IP attacks	system-guard ip enable	Required Disabled by default
Set the maximum number of infected hosts that can be concurrently monitored	system-guard ip detect-maxnum number	Optional 30 by default
Set the maximum number of addresses that the system can learn, the maximum number of times an address can be hit before an action is taken and the address isolation time (presented in the number of multiples of MAC address aging time)	system-guard ip detect-threshold ip-record-threshold record-times-threshold isolate-time	Optional By default, ip-record-threshold is 30; record-times-threshold is 1, and isolate-time is 3.

& Note:

The correlations among the arguments of the system-guard ip detect-threshold command can be clearly described with this example: If you set ip-record-threshold, record-times-threshold and isolate-time to 30, 1 and 3 respectively, when the system detects successively three times that over 50 IP packets (destined for an address other that an IP address of the switch) from a source IP address are received within a period of 10 seconds, the system considers that it is being attacked — the system sorts out that source IP address and waits a period of 5 times the MAC address aging time before learning the destination IP address(es) of packets from that source IP address again.

4.2.2 Configuring System Guard Against TCN Attacks

Configuration of System Guard against TCN attacks includes these tasks:

l Enabling System Guard against TCN attacks

l Setting the threshold of TCN/TC packet receiving rate

Follow these steps to configure System Guard against TCN attacks:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable System Guard against TCN attacks	system-guard tcn enable	Required Disabled by default
Set the threshold of TCN/TC packet receiving rate	system-guard tcn rate-threshold rate-threshold	Optional 1 pps by default

& Note:

As the system monitoring cycle is 10 seconds, the system sends trap and log information if more than 10 TCN/TC packets are received within 10 seconds by default. If the TCN/TC packet receiving rate is lower than the set threshold within a 10-second monitoring cycle, the system will not send trap or log information in the next 10-second monitoring cycle.

4.2.3 Enabling Layer 3 Error Control

Follow these steps to enable Layer 3 error control:

To do...	Use the command...	Remarks
Enter system view	system-view	—
Enable Layer 3 error control	system-guard l3err enable	Required Enabled by default

To do...

Use the command...

Remarks

Enter system view

system-view

—

Enable Layer 3 error control

system-guard l3err enable

Required

Enabled by default

4.3 Displaying and Maintaining System Guard Configuration

To do...	Use the command...	Remarks
Display the monitoring result and parameter settings of System Guard against IP attacks	display system-guard ip state	Available in any view
Display the information about IP packets received by the CPU	display system-guard ip-record
Display the status of Layer 3 error control	display system-guard l3err state
Display the status of TCN	display system-guard tcn state

H3C S3100-52P Operation Manual-Release 1602(V1.01)

Chapter 1 802.1x Configuration

I. PAE

II. Controlled port and uncontrolled port

III. The valid direction of a controlled port

IV. The way a port is controlled

I. The format of an EAPoL packet

II. The format of an EAP packet

III. Newly added fields for EAP authentication

I. EAP relay mode

II. EAP terminating mode

I. Checking the supplicant system

II. Checking the client version

III. The guest VLAN function

IV. Enabling 802.1x re-authentication

1.2 Introduction to 802.1x Configuration

1.3.1 Configuration Prerequisites

1.4.1 Configuring Proxy Checking

1.4.4 Configuring Guest VLAN

1.4.5 Configuring 802.1x Re-Authentication

1.4.6 Configuring the 802.1x Re-Authentication Timer

1.6 Configuration Example

I. Network requirements

II. Network diagram

III. Configuration procedure

2.1 Introduction to Quick EAD Deployment

2.1.2 Operation of Quick EAD Deployment

I. Restricted access

II. HTTP redirection

2.2 Configuring Quick EAD Deployment

I. Configuring a free IP range

II. Setting the ACL timeout period

2.2.3 Displaying and Maintaining Quick EAD Deployment

I. Network requirements

II. Network diagram

III. Configuration procedure

Cloud & AI

InterConnect

Intelligent Terminal Products

Product Support Services

Technical Service Solutions

Resource Center

Policy

Online Help

Become a Partner

Partner Resources

Partner Business Management

Company Information

News & Events

Contact Us