IS-IS
Intermediate System to Intermediate System (IS-IS) is a link-state IGP that uses the Shortest Path First (SPF) algorithm to determine routes. Routing decisions are made using the link-state information. IS-IS evaluates topology changes and, if necessary, performs SPF recalculations.
Entities within IS-IS include networks, intermediate systems, and end systems. In IS-IS, a network is an Autonomous System (AS), or routing domain, with end systems and intermediate systems. A router is an intermediate system that sends, receives, and forwards Protocol Data Units (PDUs). End systems are network devices that send and receive PDUs.
End system and intermediate system protocols allow routers and nodes to identify each other. IS-IS sends out link-state updates periodically throughout the network, so each router can maintain current network topology information.
IS-IS supports large ASs by using a two-level hierarchy. A large AS can be administratively divided into smaller, more manageable areas. A system logically belongs to one area. Level 1 routing is performed within an area. Level 2 routing is performed between areas. You can configure routers as Level 1, Level 2, or both Level 1 and 2.
On SR Linux, you can configure a single named IS-IS instance per network instance. The following summarizes SR Linux support for IS-IS:
-
Level 1, Level 2, and Level 1/2 IS types
-
Configurable Network Entity Title (NET) per IS-IS instance
-
support for IPv4/v6 routing
-
ECMP with up to 64 next hops per destination
-
IS-IS export policies (redistribution of other types of routes into IS-IS)
-
authentication of LSP, CSNP, PSNP, and hello PDUs, using an authentication key or keychain specified as follows:
- per instance or per level for all PDU types
- per interface or per interface and level for Hello PDUs
-
authentication keychains with a single key per named keychain
-
Purge Originator ID TLV (RFC 6232)
-
options to ignore and suppress the attached bit
-
ability to set the overload bit immediately or after each subsequent restart of the IS-IS manager application and leave it on for a configurable duration each time
-
control over the link-state PDU (LSP) MTU size, with range from 490 bytes to 9490 bytes
-
configuration control over timers for LSP lifetime, LSP refresh interval, SPF calculation triggers, and LSP generation
-
hello padding (strict, loose, and adaptive modes)
-
graceful restart, acting in restarting router mode and helper router mode
-
Level 1 to Level 2 route summary
-
BFD for fast failure detection
-
configurable hello timer with multiple per interface and level
-
wide metrics (configurable per level)
-
configurable route preference for each route type, Level 1-internal, Level 1-external, Level 2-internal and Level 2-external
- detailed statistics for interfaces, adjacencies, and levels
The info detail command displays default values for an IS-IS instance on SR Linux as shown in the following example:
--{ * candidate shared default }--[ network-instance default protocols isis ]--
# info detail
instance i1 {
admin-state disable
level-capability L2
max-ecmp-paths 1
poi-tlv false
attached-bit {
ignore false
suppress false
}
overload {
advertise-interlevel false
advertise-external false
immediate {
set-bit false
max-metric false
}
on-boot {
set-bit false
max-metric false
}
}
timers {
lsp-lifetime 1200
lsp-refresh {
interval 600
half-lifetime true
}
spf {
initial-wait 1000
second-wait 1000
max-wait 10000
}
lsp-generation {
initial-wait 1000
second-wait 1000
max-wait 5000
}
}
transport {
lsp-mtu-size 1492
}
ipv4-unicast {
admin-state enable
}
ipv6-unicast {
admin-state enable
multi-topology false
}
graceful-restart {
helper-mode false
}
auto-cost {
}
authentication {
csnp-authentication false
psnp-authentication false
hello-authentication false
}
inter-level-propagation-policies {
level1-to-level2 {
}
}
}
Basic IS-IS configuration
To configure IS-IS, perform the following tasks:
Enable an IS-IS instance
If necessary, modify the level capability on the global IS-IS instance level
Define area addresses
Configure IS-IS interfaces
Enabling an IS-IS instance
On SR Linux, you can enable a single IS-IS instance within a network-instance. The following example enables an IS-IS instance within the default network-instance.
--{ * candidate shared default }--[ ]--
# info network-instance default protocols isis
network-instance default {
protocols {
isis {
instance i1 {
}
}
}
}
Configuring the router level
You can configure routers as Level 1, Level 2, or both Level 1 and 2. When IS-IS is enabled, the default level-capability value is Level 1/2. This means that the router operates with both Level 1 and Level 2 routing capabilities. To change the default value in order for the router to operate as a Level 1 router or a Level 2 router, you must explicitly modify the level value.
The level-capability value can be configured on the global IS-IS instance level and also on the interface level. The level-capability value determines which level values can be assigned on the router level or on an interface-basis.
In order for the router to operate as a Level 1 only router or as a Level 2 only router, you must explicitly specify the level-number value.
-
Specify Level 1 to route only within an area
-
Specify Level 2 to route to destinations outside an area, toward other eligible Level 2 routers
The following example configures the level capability for an IS-IS instance to Level 2.
--{ * candidate shared default }--[ network-instance default protocols ]--
# info isis
isis {
instance i1 {
level-capability L2
}
}
Configuring the Network Entity Title
On SR Linux, you can configure the network entity title (NET) per IS-IS instance. The NET is 8-20 octets long and consists of 3 parts: the area address (1-13 octets), the system ID (6 octets), and the n-selector (1 octet, must be 00)
The area address portion of the NET defines the IS-IS area to which the router belongs. At least one area address should be configured on each router participating in IS-IS.
The area address portion of the NET identifies a point of connection to the network, such as a router interface. The routers in an area manage routing tables about destinations within the area. The NET value is used to identify the IS-IS area to which the router belongs.
The NET value is divided into three parts. Only the Area ID portion is configurable.
1. Area ID — A variable length field between 1 and 13 bytes. This includes the Authority and Format Identifier (AFI) as the most significant byte and the area ID.
2. System ID — A 6-byte system identifier. This value is not configurable. The system ID is derived from the system or router ID.
3. Selector ID — A 1-byte selector identifier that must contain zeros when configuring a NET. This value is not configurable. The selector ID is always 00.
The following example configures a NET for an IS-IS instance:
--{ * candidate shared default }--[ network-instance default protocols ]--
# info isis
isis {
instance i1 {
net 49.0001.1921.6800.1002.00
}
}
}
Configuring global parameters
You can configure the commands and parameters on the global IS-IS instance level.
Commands and parameters configured on the global IS-IS instance level are inherited by the interface levels. Parameters specified in the interface and interface-level configurations take precedence over global configurations.
The following example shows the command usage to configure global-level IS-IS. The LSP PDU authentication setting references a keychain defined at the system level (see Protocol authentication).
--{ * candidate shared default }--[ network-instance default protocols ]--
# info isis
isis {
instance i1 {
level-capability L2
overload {
on-boot {
timeout 90
}
}
authentication {
lsp-authentication {
generate true
check-received strict
keychain isisglobal
}
}
}
}
Configuring interface parameters
There are no interfaces associated with IS-IS by default. An interface belongs to all areas configured on a router. Interfaces cannot belong to separate areas. There are no default interfaces applied to the router IS-IS instance. You must configure at least one IS-IS interface in order for IS-IS to work.
You can configure both the Level 1 parameters and the Level 2 parameters on an interface. The level-capability value determines which level values are used.
The following example configures interface parameters for an IS-IS instance:
--{ * candidate shared default }--[ network-instance default protocols isis ]--
# info instance i1
instance i1 {
interface ethernet-1/2.1 {
circuit-type point-to-point
ipv4-unicast {
admin-state enable
}
level 1 {
authentication {
hello-authentication {
generate true
check-received strict
keychain Hello
}
}
}
}
level 1 {
}
}
Configuring authentication keys
IS-IS supports authentication for PDUs using shared keys, which are changed at regular intervals using keys configured in a keychain. This authentication mechanism is described in Protocol authentication.
In addition to using shared keys, authentication for IS-IS Hello, CSNP, PSNP, and LSP PDUs can be done using directly configured keys. You can specify the key used for authenticating IS-IS PDUs associated with a specific IS-IS instance, received or transmitted on a specific interface, and associated with a specific level.
If a Hello PDU is received or transmitted on a specific interface, it is authenticated using the key configured for that interface. If no key exists for the interface, the key configured for the instance is used. For CSNP, PSNP. and LSP PDUs, authentication is performed using the key configured for the level. If no key exists for the level, the key configured for the instance is used.
To configure a key, you specify the secret key (auth-password) and
cryptographic algorithm to be used for generating the key.
The following example configures keys for an IS-IS instance, interface, and level.
--{ candidate shared default }--[ ]--
# info network-instance default protocols isis
network-instance default {
protocols {
isis {
instance i1 {
authentication {
key {
crypto-algorithm cleartext
auth-password $aes$9G3XrtckZzMg=$In9WuOvKPsTw6ehDX5YLgA==
}
}
interface ethernet-1/1.1 {
authentication {
key {
crypto-algorithm hmac-md5
auth-password $aes$97mfUA4Swx6I=$PfF02Mtu0gUXH5LwT/ltqQ==
}
}
}
level 1 {
authentication {
key {
crypto-algorithm hmac-sha-256
auth-password $aes$9YHCLtkxaLGw=$7HXuQHhR4wPXwifkGekFaQ==
}
}
}
}
}
}
}IS-IS graceful restart
IS-IS graceful restart is a mechanism to prevent routing protocol re-convergence during a control plane switchover, reset, or upgrade.
Without graceful restart, when an IS-IS router restarts, its IS-IS neighbors detect that the router has gone down and remove routes that pass through that neighbor, resulting in data loss even if the restarting router is able to maintain its forwarding table state.
Configuring graceful restart can prevent this data loss. To do this, the restarting router relies on neighbor routers, which have also been configured for graceful restart, to maintain forwarding state while the router restarts. These neighbor routers are known as helper routers.
-
restarting router mode when its own IS-IS manager application (isis_mgr) restarts, or the application is killed
-
helper router mode when it detects that an adjacent IS-IS router has restarted
Restarting router mode
When the router is configured to operate in restarting router node, if the SR Linux IS-IS manager application restarts or is killed, routes that are published by the IS-IS manager (associated with any IS-IS instance), and used for programming the FIB, remain programmed for a duration of time sufficient to allow a warm restart of the IS-IS manager application. The system starts sending periodic IIH PDUs (with the Restart Request flag set) on all IS-IS interfaces so that neighbors do not time out their adjacencies.
The helper router on each interface responds with an IIH PDU containing a Restart Acknowledgement (RA). The exchange of IIH PDUs on each interface causes adjacencies to reform. After each adjacency comes up, the helper router on the associated interface transmits a complete set of CSNPs to the restarting router.
Helper router mode
When the router is configured to operate in helper router node, SRLinux advertises TLV 211 in IIH PDUs sent on all IS-IS interfaces, which indicates that the IS-IS instance is able to help any adjacent neighbor that signals a restart.
When the router receives an IIH PDU with the RR flag set from an adjacent neighbor, the router maintains its adjacency with the neighbor, responds with an IIH PDU with the RA flag set, and sends a complete set of CSNP PDUs to the restarting router.
Configuring IS-IS graceful restart
You can configure SR Linux to operate as a restarting router (informing adjacent routers when the IS-IS manager application restarts or is killed) and as a helper router (indicating to adjacent routers that it can help those signaling a restart). By default, both restarting router mode and helper router mode are disabled.
Configure restarting router mode
To configure the router to operate in restarting router mode for IS-IS graceful restart, enable non-stop-forwarding for IS-IS. For example:
--{ candidate shared default }--[ ]--
# info network-instance green protocols isis
network-instance default {
protocols {
isis {
non-stop-forwarding {
admin-state enable
}
}
}
}Configure helper router mode
The following example enables the router to operate in helper router mode:
--{ candidate shared default }--[ ]--
# info network-instance green protocols isis instance i1 graceful-restart
network-instance default {
protocols {
isis {
instance i1 {
graceful-restart {
helper-mode true
acceptable-duration 120
}
}
}
}
}
The acceptable-duration parameter sets the amount of time in seconds that SR Linux advertises as the Remaining Time in the Restart TLV with the RA flag set when this router starts to help another router that has entered restart mode. By default, this is 60 seconds.
Displaying IS-IS information
Use the commands shown in this section to display the following information for an IS-IS instance running in a specified network instance:
-
interface information
-
adjacency information
-
IS-IS link state database information
IS-IS summary information
To display summary information for an IS-IS instance:
# show network-instance green_default protocols isis summary
------------------------------------------------------------------------------------------
Network instance "green_default", isis instance "default" is enable and up
Level Capability : L1L2
Export policy : None
------------------------------------------------------------------------------------------
System-id : 0050.0500.5005
NET : [ 49.0001.0050.0500.5005.00 ]
Area-id : [ 49.0001 ]
------------------------------------------------------------------------------------------
IPv4 routing is enable
IPv6 routing is enable using None
Max ECMP path : 1
------------------------------------------------------------------------------------------
Ldp Synchronization is Disabled
------------------------------------------------------------------------------------------
Overload
Current Status : not in overload
------------------------------------------------------------------------------------------
Metric
Reference bandwidth: NA
L1 metric style: wide
L2 metric style: wide
------------------------------------------------------------------------------------------
Graceful Restart
Helper Mode : disabled
Current Status : not helping any neighbors
------------------------------------------------------------------------------------------
Timers
LSP Lifetime : 1200
LSP Refresh : 600
SPF initial wait : 1000
SPF second wait : 1000
SPF max wait : 10000
LSP generation initial wait : 10
LSP generation second wait : 1000
LSP generation max wait : 5000
------------------------------------------------------------------------------------------
Route Preference
L1 internal : 15
L1 external : 160
L2 internal : 18
L2 external : 165
------------------------------------------------------------------------------------------
L1->L2 Summary Addresses Not configured
------------------------------------------------------------------------------------------
Instance Statistics
SPF run : 29
Last SPF : 2022-03-23T16:16:16.200Z
Partial SPF run : 16
Last Partial SPF : 2022-03-23T16:16:17.200Z
------------------------------------------------------------------------------------------
PDU Statistics
------------------------------------------------------------------------------------------
| pdu-name | received | processed | dropped | sent |
+==========+==================+==================+==================+==================+
| LSP | 460 | 457 | 3 | 528 |
| IIH | 308 | 281 | 27 | 497 |
| CSNP | 52 | 51 | 1 | 116 |
| PSNP | 30 | 30 | 0 | 3 |
------------------------------------------------------------------------------------------IS-IS interface information
To display interface information for an IS-IS instance:
# show network-instance green_default protocols isis interface
------------------------------------------------------------------------------------------
Network Instance : green_default
Instance : default
------------------------------------------------------------------------------------------
| Interface Name | Oper State | Level | Circuit id | Circuit type | Ipv4 Metric L1/L2 | Ipv6 Metric L1/L2
===============================================================================================
| ethernet-1/1.1 | up | L1L2 | 2 | point-to-point | 10/10 | 10/10
| ethernet-1/2.1 | up | L1L2 | 2 | broadcast | 10/10 | 10/10
| ethernet-1/3.1 | up | L1L2 | 3 | broadcast | 10/10 | 10/10
| ethernet-1/16. | up | L1L2 | 4 | broadcast | 10/10 | 10/10
| 1 | | | | | |
| lo0.1 | up | L1L2 | 5 | broadcast | 0/0 | 0/0
-----------------------------------------------------------------------------------------------IS-IS interface detail information
To display detail information for a specific IS-IS interface:
# show network-instance green_default protocols isis interface ethernet-1/1.1 detail
----------------------------------------------------------------------------------------
Network Instance : green_default
Instance : default
----------------------------------------------------------------------------------------
Interface-Name : ethernet-1/1.1
Status : IS-IS is admin enabled, oper up
Circuit : id 1 is broadcast and not passive
Hello Authentication Generate : True
Hello Authentication Check Received : Strict
Hello Padding : disable
Csnp Interval : 10
Lsp Pacing : 100
Ldp Sync State : disabled
Ldp Sync Duration : 3274
---------------------------------------------------------------------------------------
Level : 1
Status : enabled
Adjacencies : 1
Hello Authentication Generate : True
Hello Authentication Check Received : Strict
Priority : 64
Hello Interval : 9
Hello Multiplier : 3
Ipv4 Metric : 10
Ipv6 Metric : 10
----------------------------------------------------------------------------------------
Level : 2
Status : enabled
Adjacencies : 1
Hello Authentication Generate : True
Hello Authentication Check Received : Strict
Priority : 64
Hello Interval : 9
Hello Multiplier : 3
Ipv4 Metric : 10
Ipv6 Metric : 10
----------------------------------------------------------------------------------------IS-IS adjacency information
To display IS-IS adjacency information:
# show network-instance default protocols isis adjacency
--------------------------------------------------------------------------------------
Network-instance : default
IS-IS instance : global
--------------------------------------------------------------------------------------
System-Id Adj-Level Interface IPv4-Address State Uptime Rem-Hold
<hostname1> L1 ethernet-1/1.0 10.0.0.1 Up 0d 00:46:43 19s
<hostname1> L2 ethernet-1/1.0 10.0.0.1 Up 0d 00:46:43 19s
--------------------------------------------------------------------------------------
Adjacencies: 2
--------------------------------------------------------------------------------------IS-IS link state database information
To display information for the IS-IS link state database:
# show network-instance green_default protocols isis database
--------------------------------------------------------------------------------------
Network-instance : green_default
IS-IS instance : default
--------------------------------------------------------------------------------------
| Level Number | Lsp Id | Sequence | Checksum | Lifetime | Attributes |
+==============+======================+==========+==========+==========+============+
| 1 | 0010.0100.1001.00-00 | 0x33 | 0x1672 | 1167 | L1 L2 |
| 1 | 0020.0200.2002.00-00 | 0x35 | 0xd562 | 1014 | L1 L2 |
| 1 | 0030.0300.3003.00-00 | 0x38 | 0xf447 | 640 | L1 L2 |
| 1 | 0030.0300.3003.01-00 | 0x2f | 0x4db6 | 1005 | L1 L2 |
| 1 | 0030.0300.3003.02-00 | 0x2e | 0xd355 | 709 | L1 L2 |
| 1 | 0040.0400.4004.00-00 | 0x39 | 0x6f2a | 638 | L1 L2 |
| 1 | 0040.0400.4004.01-00 | 0x2f | 0xf0ef | 822 | L1 L2 |
| 1 | 0040.0400.4004.02-00 | 0x2f | 0xa5f8 | 999 | L1 L2 |
| 1 | 0050.0500.5005.00-00 | 0x38 | 0xfbbb | 905 | L1 L2 |
| 1 | 0050.0500.5005.01-00 | 0x31 | 0x3937 | 745 | L1 L2 |
| 1 | 0050.0500.5005.02-00 | 0x2f | 0xd19 | 657 | L1 L2 |
| 1 | 0060.0600.6006.00-00 | 0x37 | 0xf287 | 967 | L1 L2 |
| 1 | 0060.0600.6006.01-00 | 0x2f | 0xadfb | 753 | L1 L2 |
| 1 | 0060.0600.6006.02-00 | 0x2f | 0x5f95 | 819 | L1 L2 |
| 1 | 0070.0700.7007.00-00 | 0x33 | 0x48dd | 1058 | L1 L2 |
| 1 | 0070.0700.7007.01-00 | 0x2f | 0xadb2 | 1164 | L1 L2 |
| 1 | 0070.0700.7007.02-00 | 0x2e | 0xdf8e | 852 | L1 L2 |
| 2 | 0010.0100.1001.00-00 | 0x3e | 0xb92f | 1150 | L1 L2 |
| 2 | 0010.0100.1001.00-01 | 0x3c | 0x1875 | 818 | L1 L2 |
| 2 | 0020.0200.2002.00-00 | 0x41 | 0x540f | 1177 | L1 L2 |
| 2 | 0020.0200.2002.00-01 | 0x3f | 0x2db9 | 699 | L1 L2 |
| 2 | 0030.0300.3003.00-00 | 0x3c | 0x302f | 1058 | L1 L2 |
| 2 | 0030.0300.3003.00-01 | 0x3f | 0x5150 | 668 | L1 L2 |
| 2 | 0030.0300.3003.01-00 | 0x30 | 0xb518 | 915 | L1 L2 |
| 2 | 0030.0300.3003.02-00 | 0x2f | 0xe113 | 1035 | L1 L2 |
| 2 | 0040.0400.4004.00-00 | 0x3e | 0xa17b | 657 | L1 L2 |
| 2 | 0040.0400.4004.00-01 | 0x3e | 0x260b | 1018 | L1 L2 |
| 2 | 0040.0400.4004.01-00 | 0x30 | 0x1511 | 1066 | L1 L2 |
| 2 | 0040.0400.4004.02-00 | 0x2f | 0x27a7 | 1035 | L1 L2 |
| 2 | 0050.0500.5005.00-00 | 0x41 | 0x59d6 | 608 | L1 L2 |
| 2 | 0050.0500.5005.00-01 | 0x44 | 0xf165 | 1110 | L1 L2 |
| 2 | 0050.0500.5005.01-00 | 0x33 | 0x7709 | 584 | L1 L2 |
| 2 | 0050.0500.5005.02-00 | 0x31 | 0xa74 | 657 | L1 L2 |
| 2 | 0060.0600.6006.00-00 | 0x3d | 0xd5ed | 903 | L1 L2 |
| 2 | 0060.0600.6006.00-01 | 0x44 | 0xdc97 | 666 | L1 L2 |
| 2 | 0060.0600.6006.01-00 | 0x30 | 0x9024 | 1145 | L1 L2 |
| 2 | 0060.0600.6006.02-00 | 0x30 | 0xbc66 | 1020 | L1 L2 |
| 2 | 0070.0700.7007.00-00 | 0x3a | 0x81fd | 862 | L1 L2 |
| 2 | 0070.0700.7007.00-01 | 0x3e | 0xf82a | 765 | L1 L2 |
| 2 | 0070.0700.7007.01-00 | 0x30 | 0xbad5 | 658 | L1 L2 |
| 2 | 0070.0700.7007.02-00 | 0x2f | 0x1ecb | 675 | L1 L2 |
+--------------+----------------------+----------+----------+----------+------------+
LSP Count: 41
-------------------------------------------------------------------------------------# show network-instance green_default protocols isis database 1
--------------------------------------------------------------------------------------
Network-instance : green_default
IS-IS instance : default
--------------------------------------------------------------------------------------
| Level Number | Lsp Id | Sequence | Checksum | Lifetime | Attributes |
+==============+======================+==========+==========+==========+============+
| 1 | 0010.0100.1001.00-00 | 0x33 | 0x1672 | 1048 | L1 L2 |
| 1 | 0020.0200.2002.00-00 | 0x35 | 0xd562 | 894 | L1 L2 |
| 1 | 0030.0300.3003.00-00 | 0x39 | 0x7762 | 1179 | L1 L2 |
| 1 | 0030.0300.3003.01-00 | 0x2f | 0x4db6 | 886 | L1 L2 |
| 1 | 0030.0300.3003.02-00 | 0x2f | 0x4ace | 1188 | L1 L2 |
| 1 | 0040.0400.4004.00-00 | 0x3a | 0xcd9f | 1135 | L1 L2 |
| 1 | 0040.0400.4004.01-00 | 0x2f | 0xf0ef | 703 | L1 L2 |
| 1 | 0040.0400.4004.02-00 | 0x2f | 0xa5f8 | 879 | L1 L2 |
| 1 | 0050.0500.5005.00-00 | 0x38 | 0xfbbb | 785 | L1 L2 |
| 1 | 0050.0500.5005.01-00 | 0x31 | 0x3937 | 625 | L1 L2 |
| 1 | 0050.0500.5005.02-00 | 0x30 | 0xb86e | 1148 | L1 L2 |
| 1 | 0060.0600.6006.00-00 | 0x37 | 0xf287 | 847 | L1 L2 |
| 1 | 0060.0600.6006.01-00 | 0x2f | 0xadfb | 633 | L1 L2 |
| 1 | 0060.0600.6006.02-00 | 0x2f | 0x5f95 | 700 | L1 L2 |
| 1 | 0070.0700.7007.00-00 | 0x33 | 0x48dd | 938 | L1 L2 |
| 1 | 0070.0700.7007.01-00 | 0x2f | 0xadb2 | 1044 | L1 L2 |
| 1 | 0070.0700.7007.02-00 | 0x2e | 0xdf8e | 733 | L1 L2 |
+--------------+----------------------+----------+----------+----------+------------+
LSP Count: 17
--------------------------------------------------------------------------------------Clearing IS-IS information
To clear information for an IS-IS instance, use the tools commands below:
Clear statistics
To clear statistics for an IS-IS instance running in a specified network instance:
# tools network-instance default protocols isis instance i1 statistics clearClear link state database information
To clear link state database information for a level:
# tools network-instance default protocols isis instance i1 link-state-database clearClear IS-IS adjacency information
To clear IS-IS adjacency information for an interface:
# tools network-instance default protocols isis instance i1 interface ethernet-1/1.1 adjacencies clearIS-IS Weighted ECMP
Weighted IP ECMP, also known as UCMP (unequal cost multipath), allows the installation of a multipath route in the FIB, where the ECMP flow hashing distributes traffic by directing flows to each next-hop in a ratio proportional to the weight assigned to each next-hop. Multipath routes are the routes that have multiple next-hops.
Weighted ECMP (wECMP) distributes traffic unequally over multiple paths and uses available bandwidth more efficiently for better load balancing. For example, if you have four equal-cost paths to a destination, but one path has a lower bandwidth than the other three, you can use wECMP to assign more traffic to the higher bandwidth path and less traffic to the lower bandwidth paths.
The wECMP feature is available on 7250 IXR-6/10/6e/10e platforms and is supported for both IS-IS IPv4 and IPv6 routes.
Enabling weighted ECMP
You can enable the weighted ECMP per IS-IS instance. wECMP can be configured for an IS-IS instance under a default network instance or network instance of type ip-vrf . Enabling weighted ECMP, triggers weighted ECMP programming for all eligible multipath IS-IS routes associated with the instance. To program an IS-IS route as a multipath route, all the multipaths must have an equal total path cost to the destination.
The ECMP weights (load-balancing-weight) are normalized based on the number of hash buckets per next-hop group (NHG). The sum of all normalized datapath weights must not exceed the hash bucket limit.
You must configure the maximum number of ECMP hash buckets used for IS-IS weighted
ECMP routing on a per IS-IS instance basis. If the hash bucket limit is
T and the normalized datapath weight of each next-hop is
Ni then the sum of all
Ni in each mutipath set (NHG) cannot exceed
T. The system normalizes the weights used in weighted ECMP
according to the algorithm described in Normalizing datapath weights.
The 7250 IXR-6/10/6e/10e platforms have a 128 hash bucket limit. Datapath resources can be conserved by configuring a lower value for the bucket size. However, this reduces the granularity or fine distribution of the traffic. With a smaller bucket size, there are fewer next-hop options available for load balancing, which can limit the flexibility and precision of traffic distribution. You cannot reduce the hash bucket limit below the ECMP limit of the IS-IS instance.
The following example enables weighted ECMP per IS-IS instance:
Enabling weighted ECMP
--{ * candidate shared default }--[ ]--
# info network-instance default protocols isis
network-instance mgmt {
protocols {
isis {
instance test {
weighted-ecmp {
admin-state enable
max-ecmp-hash-buckets-per-next-hop-group 56
}
}
}
}
}- admin-state
When set to enable, triggers weighted ECMP programming for all eligible multipath IS-IS routes associated with the instance. The default is disable. This ensures backward compatibility with previous releases that only supported ECMP wherein the traffic was distributed in equal proportion.
- max-ecmp-hash-buckets-per-next-hop-group
Specify the maximum number of ECMP hash buckets per next-hop-group. Default bucket size is
128.Note: The configured value of max-ecmp-hash-buckets-per-next-hop-group must always be greater than or equal to max-ecmp-paths (network-instance.protocols.isis.instance.max-ecmp-paths).
Disabling weighted ECMP
--{ * candidate shared default }--[ ]--
# info network-instance default protocols isis
network-instance mgmt {
protocols {
isis {
instance test {
weighted-ecmp {
admin-state disable
max-ecmp-hash-buckets-per-next-hop-group 1
}
}
}
}
}load-balancing-weight)
configured.Configuring weighted load-balancing over interface next-hops
When the weighted ECMP feature is enabled, the IS-IS route linked to the routing instance can be programmed into the datapath. This allows for weighted load-balancing across the interface next-hops of the route.
In order for weighted ECMP to be supported across the interface next-hops of an IS-IS route the following conditions must be met:
-
All ECMP next-hops must be interface next-hops
-
All next-hop interfaces are configured with non-zero load-balancing weights.
To configure load-balancing weight, set the parameter
load-balancing-weight under
network-instance.protocols.isis.instance.interface.level.weighted-ecmp.
This parameter can be set to a static value, or you can choose to use the options
auto or none. The static value assigned must
be between 1 and 4294967295.
If all the
next-hop interfaces in the multipath set have a
load-balancing-weight, which is either automatically derived
from the port/LAG bandwidth using the auto option or configured
with a static value, then the route is programmed in the FIB with wECMP. If one or
more next-hop interfaces have zero weight (load-balancing-weight =
none), then the wECMP load-balancing falls back to classic ECMP
operation and equally distributes the traffic.
In the NHG for the IS-IS route, each next-hop is assigned a non-zero weight. In SR Linux, the wECMP programming capability is available even if some multipath next-hop interfaces are connected to different adjacent routers.
The following example configures weighted load-balancing over interface next-hops:Configuring weighted load-balancing over interface next-hops
--{ * candidate shared default }--[ ]--
# info network-instance default protocols isis
network-instance mgmt {
protocols {
isis {
instance test {
interface mgmt0.0 {
weighted-ecmp {
load-balancing-weight auto
}
}
}
}
}
}Here the load-balancing-weight value is automatically derived from the port/LAG bandwidth.
Normalizing datapath weights
The ECMP weights (load-balancing-weight) are normalized based on the number of hash buckets (max-ecmp-hash-buckets-per-next-hop-group) defined per NHG. The sum of all normalized datapath weight must not exceed the hash bucket limit.
- Calculate the G.C.D of all datapath weights.
- Divide all datapath weights by the G.C.D.
- If
the sum of normalized weights is less than the maximum number of buckets, weight
determination process is complete.
Or
- If the sum of normalized weights exceeds the bucket size,
- Assign each nexthop a minimum of one bucket.
- Distribute the remaining buckets (max_buckets - num_nexthops) to each nexthop based on the ratio of each nexthop's weight to the total weight (rounded down). There may be cases where no extra buckets are assigned.