JavaScript is required to for searching.
Skip Navigation Links
Exit Print View
Oracle Solaris Administration: Network Interfaces and Network Virtualization     Oracle Solaris 11 Express 11/10
search filter icon
search icon

Document Information


Part I Network Auto-Magic

1.  Introduction to NWAM

2.  NWAM Configuration and Administration (Overview)

3.  NWAM Profile Configuration (Tasks)

4.  NWAM Profile Administration (Tasks)

5.  About the NWAM Graphical User Interface

Part II Administering Single Interfaces

6.  Overview of the Networking Stack

7.  Datalink Configuration and Administration

8.  Configuring an IP Interface

9.  Configuring Wireless Interface Communications on Oracle Solaris

Part III Administering Interface Groups

10.  Administering Bridges

11.  Administering Link Aggregations

12.  Administering VLANs

13.  Introducing IPMP

What's New With IPMP

Deploying IPMP

Why You Should Use IPMP

When You Must Use IPMP

Comparing IPMP and Link Aggregation

Using Flexible Link Names on IPMP Configuration

How IPMP Works

IPMP Components in Oracle Solaris

Types of IPMP Interface Configurations

IPMP Addressing

IPv4 Test Addresses

IPv6 Test Addresses

Failure and Repair Detection in IPMP

Types of Failure Detection in IPMP

Link-Based Failure Detection

Probe-Based Failure Detection

NICs That Are Missing at Boot

Failure Detection and the Anonymous Group Feature

Detecting Physical Interface Repairs

The FAILBACK=no Mode

IPMP and Dynamic Reconfiguration

Attaching New NICs

Detaching NICs

Replacing NICs

IPMP Terminology and Concepts

14.  Administering IPMP

Part IV  Network Virtualization and Resource Management

15.  Introducing Network Virtualization and Resource Control (Overview)

16.  Planning for Network Virtualization and Resource Control

17.  Configuring Virtual Networks (Tasks)

18.  Using Link Protection in Virtualized Environments

19.  Managing Network Resources

20.  Monitoring Network Traffic and Resource Usage



IPMP Terminology and Concepts

This section introduces terms and concepts that are used throughout the IPMP chapters in this book.

active interface

Refers to an underlying interface that can be used by the system to send or receive data traffic. An interface is active if the following conditions are met:

  • At least one IP address is UP in the interface. See UP address.

  • The FAILED, INACTIVE, or OFFLINE flag is not set on the interface.

  • The interface has not been flagged as having a duplicate hardware address.

Compare to unusable interface, INACTIVE interface.

data address

Refers to an IP address that can be used as the source or destination address for data. Data addresses are part of an IPMP group and can be used to send and receive traffic on any interface in the group. Moreover, the set of data addresses in an IPMP group can be used continuously, provided that one interface in the group is functioning. In previous IPMP implementations, data addresses were hosted on the underlying interfaces of an IPMP group. In the current implementation, data addresses are hosted on the IPMP interface.


Refers to an IP address that cannot be used as the source address for data. Typically, IPMP test addresses, which have the NOFAILOVER flag, are also automatically marked as DEPRECATED by the system. However, any address can be marked DEPRECATED to prevent the address from being used as a source address.

dynamic reconfiguration

Refers to a feature that allows you to reconfigure a system while the system is running, with little or no impact on ongoing operations. Not all Sun platforms support DR. Some Sun platforms might only support DR of certain types of hardware. On platforms that support DR of NICs, IPMP can be used for uninterrupted network access to the system during DR.

For more information about how IPMP supports DR, refer to IPMP and Dynamic Reconfiguration.

explicit IPMP interface creation

Applies only to the current IPMP implementation. The term refers to the method of creating an IPMP interface by using the ifconfig ipmp command. Explicit IPMP interface creation is the preferred method for creating IPMP groups. This method allows the IPMP interface name and IPMP group name to be set by the administrator.

Compare to implicit IPMP interface creation.

FAILBACK=no mode

Refers to a setting of an underlying interface that minimizes rebinding of incoming addresses to interfaces by avoiding redistribution during interface repair. Specifically, when an interface repair is detected, the interface's FAILED flag is cleared. However, if the mode of the repaired interface is FAILBACK=no, then the INACTIVE flag is also set to prevent use of the interface, provided that a second functioning interface also exists. If the second interface in the IPMP group fails, then the INACTIVE interface is eligible to take over. While the concept of failback no longer applies in the current IPMP implementation, the name of this mode is preserved for administrative compatibility.

FAILED interface

Indicates an interface that the in.mpathd daemon has determined to be malfunctioning. The determination is achieved by either link-based or probe-based failure detection. The FAILED flag is set on any failed interface.

failure detection

Refers to the process of detecting when a physical interface or the path from an interface to an Internet layer device no longer works. Two forms of failure detection are implemented: link-based failure detection, and probe-based failure detection.

implicit IPMP interface creation

Refers to the method of creating an IPMP interface by using the ifconfig command to place an underlying interface in a nonexistent IPMP group. Implicit IPMP interface creation is supported for backward compatibility with the previous IPMP implementation. Thus, this method does not provide the ability to set the IPMP interface name or IPMP group name.

Compare to explicit IPMP interface creation.

INACTIVE interface

Refers to an interface that is functioning but is not being used according to administrative policy. The INACTIVE flag is set on any INACTIVE interface.

Compare to active interface, unusable interface.

IPMP anonymous group support

Indicates an IPMP feature in which the IPMP daemon tracks the status of all network interfaces in the system, regardless of whether they belong to an IPMP group. However, if the interfaces are not actually in an IPMP group, then the addresses on these interfaces are not available in case of interface failure.

IPMP group

Refers to a set of network interfaces that are treated as interchangeable by the system in order to improve network availability and utilization. Each IPMP group has a set of data addresses that the system can associate with any set of active interfaces in the group. Use of this set of data addresses maintains network availability and improves network utilization. The administrator can select which interfaces to place into an IPMP group. However, all interfaces in the same group must share a common set of properties, such as being attached to the same link and configured with the same set of protocols (for example, IPv4 and IPv6).

IPMP group interface

See IPMP interface.

IPMP group name

Refers to the name of an IPMP group, which can be assigned with the ifconfig group subcommand. All underlying interfaces with the same IPMP group name are defined as part of the same IPMP group. In the current implementation, IPMP group names are de-emphasized in favor of IPMP interface names. Administrators are encouraged to use the same name for both the IPMP interface and the group.

IPMP interface

Applies only to the current IPMP implementation. The term refers to the IP interface that represents a given IPMP group, any or all of the interface's underlying interfaces, and all of the data addresses. In the current IPMP implementation, the IPMP interface is the core component for administering an IPMP group, and is used in routing tables, ARP tables, firewall rules, and so forth.

IPMP interface name

Indicates the name of an IPMP interface. This document uses the naming convention of ipmpN. The system also uses the same naming convention in implicit IPMP interface creation. However, the administrator can choose any name by using explicit IPMP interface creation.

IPMP singleton

Refers to an IPMP configuration that is used by Sun Cluster software that allows a data address to also act as a test address. This configuration applies, for instance, when only one interface belongs to an IPMP group.

link-based failure detection

Specifies a passive form of failure detection, in which the link status of the network card is monitored to determine an interface's status. Link-based failure detection only tests whether the link is up. This type of failure detection is not supported by all network card drivers. Link-based failure detection requires no explicit configuration and provides instantaneous detection of link failures.

Compare to probe-based failure detection.

load spreading

Refers to the process of distributing inbound or outbound traffic over a set of interfaces. Unlike load balancing, load spreading does not guarantee that the load is evenly distributed. With load spreading, higher throughput is achieved. Load spreading occurs only when the network traffic is flowing to multiple destinations that use multiple connections.

Inbound load spreading indicates the process of distributing inbound traffic across the set of interfaces in an IPMP group. Inbound load spreading cannot be controlled directly with IPMP. The process is indirectly manipulated by the source address selection algorithm.

Outbound load spreading refers to the process of distributing outbound traffic across the set of interfaces in an IPMP group. Outbound load spreading is performed on a per-destination basis by the IP module, and is adjusted as necessary depending on the status and members of the interfaces in the IPMP group.


Applies only to the previous IPMP implementation. Refers to an address that is associated with an underlying interface and thus remains unavailable if the underlying interface fails. All NOFAILOVER addresses have the NOFAILOVER flag set. IPMP test addresses must be designated as NOFAILOVER, while IPMP data addresses must never be designated as NOFAILOVER. The concept of failover does not exist in the IPMP implementation. However, the term NOFAILOVER remains for administrative compatibility.

OFFLINE interface

Indicates an interface that has been administratively disabled from system use, usually in preparation for being removed from the system. Such interfaces have the OFFLINE flag set. The if_mpadm command can be used to switch an interface to an offline status.

physical interface

See: underlying interface


Refers to an ICMP packet, similar to the packets that are used by the ping command. This probe is used to test the send and receive paths of a given interface. Probe packets are sent by the in.mpathd daemon, if probe-based failure detection is enabled. A probe packet uses an IPMP test address as its source address.

probe-based failure detection

Indicates an active form of failure detection, in which probes are exchanged with probe targets to determine an interface's status. When enabled, probe-based failure detection tests the entire send and receive path of each interface. However, this type of detection requires the administrator to explicitly configure each interface with a test address.

Compare to link-based failure detection.

probe target

Refers to a system on the same link as an interface in an IPMP group. The target is selected by the in.mpathd daemon to help check the status of a given interface by using probe-based failure detection. The probe target can be any host on the link that is capable of sending and receiving ICMP probes. Probe targets are usually routers. Several probe targets are usually used to insulate the failure detection logic from failures of the probe targets themselves.

source address selection

Refers to the process of selecting a data address in the IPMP group as the source address for a particular packet. Source address selection is performed by the system whenever an application has not specifically selected a source address to use. Because each data address is associated with only one hardware address, source address selection indirectly controls inbound load spreading.

STANDBY interface

Indicates an interface that has been administratively configured to be used only when another interface in the group has failed. All STANDBY interfaces will have the STANDBY flag set.

target systems

See probe target.

test address

Refers to an IP address that must be used as the source or destination address for probes, and must not be used as a source or destination address for data traffic. Test addresses are associated with an underlying interface. If an underlying interface is configured with an UP test address, the in.mpathd daemon monitors this address by using probe-based failure detection. All test addresses must be designated as NOFAILOVER. These addresses are also automatically marked DEPRECATED by the system to ensure that they will not be considered as a possible source address for data packets.

underlying interface

Specifies an IP interface that is part of an IPMP group and is directly associated with an actual network device. For example, if ce0 and ce1 are placed into IPMP group ipmp0, then ce0 and ce1 comprise the underlying interfaces of ipmp0. In the previous implementation, IPMP groups consist solely of underlying interfaces. However, in the current implementation, these interfaces underlie the IPMP interface (for example, ipmp0) that represents the group, hence the name.

undo-offline operation

Refers to the act of administratively enabling a previously offlined interface to be used by the system. The if_mpadm command can be used to perform an undo-offline operation.

unusable interface

Refers to an underlying interface that cannot be used to send or receive data traffic at all in its current configuration. An unusable interface differs from an INACTIVE interface, that is not currently being used but can be used if an active interface in the group becomes unusable. An interface is unusable if one of the following conditions exists:

  • The interface has no UP address.

  • The FAILED or OFFLINE flag has been set for the interface.

  • The interface has been flagged has having the same hardware address as another interface in the group.

UP address

Refers to an address that has been made administratively available to the system by setting the UP flag. An address that is not UP is treated as not belonging to the system, and thus is never considered during source address selection.