Internet Engineering Task Force (IETF) D. Eastlake 3rd
Request for Comments: 6327 Huawei
Updates: 6325 R. Perlman
Category: Standards Track Intel Labs
ISSN: 2070-1721 A. Ghanwani
Brocade
D. Dutt
Cisco Systems
V. Manral
Hewlett Packard Co.
July 2011
Routing Bridges (RBridges): Adjacency
Abstract
The IETF TRILL (TRansparent Interconnection of Lots of Links)
protocol provides optimal pair-wise data forwarding without
configuration, safe forwarding even during periods of temporary
loops, and support for multipathing of both unicast and multicast
traffic. TRILL accomplishes this by using IS-IS (Intermediate System
to Intermediate System) link state routing and by encapsulating
traffic using a header that includes a hop count. Devices that
implement TRILL are called Routing Bridges (RBridges).
TRILL supports multi-access LAN (Local Area Network) links that can
have multiple end stations and RBridges attached. This document
describes four aspects of the TRILL LAN Hello protocol used on such
links, particularly adjacency, designated RBridge selection, and MTU
(Maximum Transmission Unit) and pseudonode procedures, with state
machines. There is no change for IS-IS point-to-point Hellos used on
links configured as point-to-point in TRILL.
Status of This Memo
This is an Internet Standards Track document.
This document is a product of the Internet Engineering Task Force
(IETF). It represents the consensus of the IETF community. It has
received public review and has been approved for publication by the
Internet Engineering Steering Group (IESG). Further information on
Internet Standards is available in Section 2 of RFC 5741.
Information about the current status of this document, any errata,
and how to provide feedback on it may be obtained at
http://www.rfc-editor.org/info/rfc6327.
Copyright Notice
Copyright (c) 2011 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Table of Contents
1. Introduction ....................................................4
1.1. Content and Precedence .....................................4
1.2. Terminology and Acronyms ...................................5
2. The TRILL Hello Environment and Purposes ........................5
2.1. Incrementally Replacing 802.1Q-2005 Bridges ................5
2.2. Handling Native Frames .....................................6
2.3. Zero or Minimal Configuration ..............................7
2.4. MTU Robustness .............................................7
2.5. Purposes of the TRILL Hello Protocol .......................8
3. Adjacency State Machinery .......................................9
3.1. TRILL LAN Hellos, MTU Test, and VLANs ......................9
3.2. Adjacency Table Entries and States ........................10
3.3. Adjacency and Hello Events ................................11
3.4. Adjacency State Diagram and Table .........................13
3.5. Multiple Parallel Links ...................................14
3.6. Insufficient Space in Adjacency Table .....................15
4. RBridge LAN Ports and DRB State ................................15
4.1. Port Table Entries and DRB Election State .................16
4.2. DRB Election Events .......................................16
4.2.1. DRB Election Details ...............................17
4.2.2. Change in DRB ......................................18
4.2.3. Change in Designated VLAN ..........................18
4.3. State Table and Diagram ...................................18
5. MTU Matching ...................................................20
6. Pseudonodes ....................................................21
7. TRILL Hello Reception and Transmission .........................21
7.1. Transmitting TRILL Hellos .................................22
7.2. Receiving TRILL Hellos ....................................23
8. Multiple Ports on the Same Link ................................24
9. Security Considerations ........................................24
10. References ....................................................24
10.1. Normative References .....................................24
10.2. Informative References ...................................25
11. Acknowledgements ..............................................25
1. Introduction
The IETF TRILL (TRansparent Interconnection of Lots of Links)
protocol [RFC6325] provides optimal pair-wise data frame forwarding
without configuration, safe forwarding even during periods of
temporary loops, and support for multipathing of both unicast and
multicast traffic. TRILL accomplishes this by using [IS-IS]
(Intermediate System to Intermediate System) link state routing and
encapsulating traffic using a header that includes a hop count. The
design supports VLANs (Virtual Local Area Networks) and optimization
of the distribution of multi-destination frames based on VLANs and
IP-derived multicast groups. Devices that implement TRILL are called
RBridges (Routing Bridges).
The purpose of this document is to improve the quality of the
description of four aspects of the TRILL LAN (Local Area Network)
Hello protocol that RBridges use on broadcast (LAN) links. It
includes reference implementation details. Alternative
implementations that interoperate on the wire are permitted. There
is no change for IS-IS point-to-point Hellos used on links configured
as point-to-point in TRILL.
The scope of this document is limited to the following aspects of the
TRILL LAN Hello protocol:
- Adjacency formation
- DRB (Designated RBridge aka DIS (Designated Intermediate
System)) election
- Rules for 2-way and MTU (Maximum Transmission Unit) matching for
advertisements
- Creation and use of pseudonodes
For other aspects of the TRILL base protocol, see [RFC6325].
1.1. Content and Precedence
Section 2 below explains the rationale for the differences between
the TRILL LAN Hello protocol and the Layer 3 IS-IS LAN Hello protocol
[IS-IS] [RFC1195] in light of the environment for which the TRILL
protocol is designed. It also describes the purposes of the TRILL
LAN Hello protocol.
Section 3 describes the adjacency state machine and its states and
relevant events.
Section 4 describes the Designated RBridge (DRB) election state
machine for RBridge ports and its states and relevant events.
Section 5 describes MTU testing and matching on a TRILL link.
Section 6 discusses pseudonode creation and use.
Section 7 provides more details on the reception and transmission of
TRILL LAN Hellos.
Section 8 discusses multiple ports from one RBridge on the same link.
In case of conflict between this document and [RFC6325], this
document prevails.
1.2. Terminology and Acronyms
This document uses the acronyms defined in [RFC6325] supplemented by
the following additional acronym:
SNPA - Subnetwork Point of Attachment
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in [RFC2119].
2. The TRILL-Hello Environment and Purposes
[IS-IS] has subnetwork-independent functions and subnetwork-dependent
functions. Currently, Layer 3 use of IS-IS supports two types of
subnetworks: (1) point-to-point link subnetworks between routers and
(2) general broadcast (LAN) subnetworks. Because of the differences
between the environment of Layer 3 routers and the environment of
TRILL RBridges, instead of the broadcast (LAN) subnetwork-dependent
functions encountered at Layer 3, which are specified in [IS-IS]
Section 8.4, the TRILL protocol uses modified subnetwork-dependent
functions for a LAN subnetwork. The environmental differences are
described in Sections 2.1 through 2.4, followed by a summation, in
Section 2.5, of the purposes of the TRILL LAN Hello protocol.
2.1. Incrementally Replacing 802.1Q-2005 Bridges
RBridges can incrementally replace IEEE [802.1Q-2005] bridges. Thus,
RBridges need to provide similar services, including delivery of
frames only to links in the frame's VLAN and priority queuing of
frames, to the extent that multiple queues are implemented at any
particular RBridge port.
RBridge ports are IEEE [802.1Q-2005] ports in terms of their frame
VLAN and priority configuration and processing as described in
Section 2.6 of [RFC6325]. When a frame is received through an
RBridge port, like a frame received through any [802.1Q-2005] port,
it has an associated VLAN ID and frame priority. When a frame is
presented to an [802.1Q-2005] port for queuing and transmission, it
must be accompanied by a VLAN ID and frame priority. However,
whether the frame, if actually transmitted, will be VLAN tagged is
determined by whether or not the port is configured to "strip VLAN
tags". Furthermore, in the general case, a broadcast (LAN) link
between RBridges can be a VLAN-capable bridged LAN that may be
configured to partition VLANs.
Because devices that restrict VLAN connectivity, such as bridged LANs
or provider bridging equipment, can be part of the link between
RBridges, TRILL Data and TRILL IS-IS frames between RBridges use the
link's Designated VLAN. The Designated VLAN is dictated for a link
by the elected Designated RBridge (equivalent to the Designated
Intermediate System at Layer 3). Because TRILL Data frames flow
between RBridges on a link only in the link's Designated VLAN,
adjacency for routing calculations is based only on connectivity
characteristics in that VLAN.
2.2. Handling Native Frames
Ordinary Layer 3 data packets are already "tamed" when they are
originated by an end station: they include a hop count and Layer 3
source and destination address fields. Furthermore, for ordinary
data packets, there is no requirement to preserve their outer Layer 2
addressing and, at least if the packets are unicast, they are
addressed to their first hop router. In contrast, RBridges running
TRILL must accept, transport, and deliver untamed "native" frames (as
defined in Section 1.4 of [RFC6325]). Native frames lack a TRILL hop
count field. Native frames also have Layer 2 addresses that indicate
their source and are used as the basis for their forwarding. These
Layer 2 addresses must be preserved for delivery to the native
frame's Layer 2 destination. One resulting difference is that
RBridge ports providing native frame service must receive in
promiscuous MAC (Media Access Control) address mode, while Layer 3
router ports typically receive in a regularly selective MAC address
mode.
TRILL handles this by having, on the link where an end station
originated a native frame, one RBridge "ingress" such a locally
originated native frame by adding a TRILL Header that includes a hop
count, thus converting it to a TRILL Data frame. This augmented
frame is then routed to one RBridge on the link having the
destination end station for the frame (or one RBridge on each such
link if it is a multi-destination frame). Such final RBridges
perform an "egress" function, removing the TRILL Header and
delivering the original frame to its destination(s). (For the
purposes of TRILL, a Layer 3 router is an end station.)
Care must be taken to avoid a loop that would involve egressing a
native frame and then re-ingressing it because, while it is in native
form, it would not be protected by a hop count. Such a loop could
involve multiplication of the number of frames each time around and
would likely saturate all links involved within milliseconds. For
TRILL, safety against such loops for a link is more important than
data connectivity on that link.
The primary TRILL defense mechanism against such loops, which is
mandatory, is to assure that, as far as practically possible, there
is only a single RBridge on each link that is in charge of ingressing
and egressing native frames from and to that link. This is the
Designated RBridge that is elected using TRILL LAN Hellos as further
described in Sections 2.5 and 4 below.
Because bridged LANs between RBridges can be configured in complex
ways (e.g., so that some VLANs pass frames unidirectionally) and loop
safety is important, there are additional TRILL defenses against
loops that are beyond the scope of this document. Specifically,
these defend against the occurrence of looping traffic that is in
native format for part of the loop. These additional defenses have
no effect on adjacency states or the receipt or forwarding of TRILL
Data frames; they only affect native frame ingress and egress.
2.3. Zero or Minimal Configuration
RBridges are expected to provide service with zero configuration,
except for services such as non-default VLAN or priority that require
configuration when offered by [802.1Q-2005] bridges. This differs
from Layer 3 routing where routers typically need to be configured as
to the subnetworks connected to each port, etc., to provide service.
2.4. MTU Robustness
TRILL IS-IS needs to be robust against links with reasonably
restricted MTUs, including links that accommodate only classic
Ethernet frames, despite the addition of reasonable headers such as
VLAN tags. This is particularly true for TRILL LAN Hellos so as to
assure that a unique DRB is elected.
TRILL will also be used inside data centers where it is not uncommon
for all or most of the links and switches to support frames
substantially larger than the classic Ethernet maximum. For example,
they may have an MTU adequate to comfortably handle Fiber Channel
over Ethernet frames, for which T11 recommends a 2,500-byte MTU
[FCoE]. It would be beneficial for an RBridge campus with such a
large MTU to be able to safely make use of it.
These needs are met by limiting the size of TRILL LAN Hellos and by
the use of MTU testing as described below.
2.5. Purposes of the TRILL-Hello Protocol
There are three purposes for the TRILL-Hello protocol as listed below
along with a reference to the section of this document in which each
is discussed:
a) To determine which RBridge neighbors have acceptable connectivity
to be reported as part of the topology (Section 3)
b) To elect a unique Designated RBridge on the link (Section 4)
c) To determine the MTU with which it is possible to communicate with
each RBridge neighbor (Section 5)
In Layer 3 IS-IS, all three of these functions are combined. Hellos
may be padded to the maximum length (see [RFC3719], Section 6) so
that a router neighbor is not even discovered if it is impossible to
communicate with it using maximum-sized packets. Also, even if
Hellos from a neighbor R2 are received by R1, if connectivity to R2
is not 2-way (i.e., R2 does not list R1 in R2's Hello), then R1 does
not consider R2 as a Designated Router candidate. Because of this
logic, it is possible at Layer 3 for multiple Designated Routers to
be elected on a LAN, with each representing the LAN as a pseudonode.
It appears to the topology as if the LAN is now two or more separate
LANs. Although this is surprising, it does not disrupt Layer 3
IS-IS.
In contrast, this behavior is not acceptable for TRILL, since in
TRILL it is important that all RBridges on the link know about each
other, and choose a single RBridge to be the DRB and to control the
native frame ingress and egress on that link. Otherwise, multiple
RBridges might encapsulate/decapsulate the same native frame, forming
loops that are not protected by the hop count in the TRILL header as
discussed above.
So, the TRILL-Hello protocol is best understood by focusing on each
of these functions separately.
One other issue with TRILL LAN Hellos is to ensure that subsets of
the information can appear in any single message, and be processable,
in the spirit of IS-IS Link State PDUs (LSPs) and Complete Sequence
Number PDUs (CSNPs). TRILL-Hello frames, even though they are not
padded, can become very large. An example where this might be the
case is when some sort of backbone technology interconnects hundreds
of TRILL sites over what would appear to TRILL to be a giant
Ethernet, where the RBridges connected to that cloud will perceive
that backbone to be a single link with hundreds of neighbors. Thus,
the TRILL Hello uses a different Neighbor TLV [RFC6326] that lists
neighbors seen for a range of MAC (SNPA) addresses.
3. Adjacency State Machinery
Each RBridge port has associated with it a port state, as discussed
in Section 4, and a table of zero or more adjacencies as discussed in
this section. The states such adjacencies can have, the events that
cause state changes, the actions associated with those state changes,
and a state table and diagram are given below.
3.1. TRILL LAN Hellos, MTU Test, and VLANs
The determination of LSP-reported adjacencies on links that are not
configured as point-to-point is made using TRILL LAN Hellos (see also
Section 7) and an optional MTU test. Appropriate TRILL LAN Hello
exchange and the satisfaction of the MTU test, if the MTU test is
enabled (see Section 5), is required for there to be an adjacency
that will be reported in an LSP of the RBridge in question.
Because bridges acting as glue on the LAN might be configured in such
a way that some VLANs are partitioned, it is necessary for RBridges
to transmit Hellos with multiple VLAN tags. The conceptually
simplest solution may have been to have all RBridges transmit up to
4,094 times as many Hellos, one with each legal VLAN ID enabled at
each port, but this would obviously have deleterious performance
implications. So, the TRILL protocol specifies that if RB1 knows it
is not the DRB, it transmits its Hellos on only a limited set of
VLANs, and only an RBridge that believes itself to be the DRB on a
port "sprays" its TRILL Hellos on all of its enabled VLANs at a port
(with the ability to configure to send on only a subset of those).
The details are given in [RFC6325], Section 4.4.3.
If the MAC (SNPA) address of more than one RBridge port on a link are
the same, all but one of such ports are put in the Suspended state
(see Section 4) and do not participate in the link except to monitor
whether they should stay suspended.
All TRILL LAN Hellos issued by an RBridge on a particular port MUST
have the same source MAC address, priority, desired Designated VLAN,
and Port ID, regardless of the VLAN in which the Hello is sent. Of
course, the priority and desired Designated VLAN can change on
occasion, but then the new value must similarly be used in all TRILL
Hellos on the port, regardless of VLAN.
3.2. Adjacency Table Entries and States
Each adjacency is in one of the following four states:
Down:
This is a virtual state for convenience in creating state
diagrams and tables. It indicates that the adjacency is non-
existent, and there is no entry in the adjacency table for it.
Detect:
An adjacent neighbor has been detected either (1) not on the
Designated VLAN or (2) on the Designated VLAN, but neither
2-way connectivity nor the MTU of such connectivity has been
confirmed.
2-Way:
2-way connectivity to the neighbor has been found on the
Designated VLAN but MTU testing is enabled and has not yet
confirmed that the connectivity meets the campus minimum MTU
requirement.
Report:
There is 2-way connectivity to the neighbor on the Designated
VLAN and either MTU testing has confirmed that the connectivity
meets the campus minimum MTU requirement or MTU testing is not
enabled. This connectivity will be reported in an LSP (with
appropriate provision for the link pseudonode, if any, as
described in Section 6).
For an adjacency in any of the three non-down states (Detect, 2-Way,
or Report), there will be an adjacency table entry. That entry will
give the state of the adjacency and will also include the information
listed below.
o The address of the neighbor (that is, its SNPA address, usually
a 48-bit MAC address), and the Port ID and the System ID in the
received Hellos. Together, these three quantities uniquely
identify the adjacency.
o Exactly two Hello holding timers, each consisting of a 16-bit
unsigned integer number of seconds: a Designated VLAN holding
timer and a non-Designated VLAN holding timer.
o The 7-bit unsigned priority of the neighbor to be the DRB.
o The VLAN that the neighbor RBridge wants to be the Designated
VLAN on the link, called the desired Designated VLAN.
3.3. Adjacency and Hello Events
The following events can change the state of an adjacency:
A0. Receive a TRILL Hello whose source MAC address (SNPA) is
equal to that of the port on which it is received. This is a
special event that is handled as described immediately after
this list of events. It does not appear in the state
transition table or diagram.
A1. Receive a TRILL Hello (other than an A0 event) on the
Designated VLAN with a TRILL Neighbor TLV that explicitly
lists the receiver's (SNPA) address.
A2. Receive a TRILL Hello (other than an A0 event) that either
(1) is not on the Designated VLAN (any TRILL Neighbor TLV in
such a Hello is ignored) or (2) is on the Designated VLAN but
does not contain a TRILL Neighbor TLV covering an address
range that includes the receiver's (SNPA) address.
A3. Receive a TRILL Hello (other than an A0 event) on the
Designated VLAN with one or more TRILL Neighbor TLVs covering
an address range that includes the receiver's (SNPA) address
-- and none of which lists the receiver.
A4. The expiration of one or both Hello holding timers results in
them both being expired.
A5. The Designated VLAN Hello holding timer expires, but the non-
Designated VLAN Hello holding timer still has time left until
it expires.
A6. MTU test successful.
A7. MTU test was successful but now fails.
A8. The RBridge port goes operationally down.
For the special A0 event, the Hello is examined to determine if it is
higher priority to be the DRB than the port on which it is received
as described in Section 4.2.1. If the Hello is of lower priority
than the receiving port, it is discarded with no further action. If
it is of higher priority than the receiving port, then any
adjacencies for that port are discarded (transitioned to the Down
state), and the port is suspended as described in Section 4.2.
The receipt of a TRILL LAN Hello with a source MAC (SNPA) address
different from that of the receiving port (that is, the occurrence of
events A1, A2, or A3), causes the following actions (except where the
Hello would create a new adjacency table entry, the table is full, or
the Hello is too low priority to displace an existing entry as
described in Section 3.6). The Designated VLAN used in these actions
is the Designated VLAN dictated by the DRB determined without taking
the received TRILL LAN Hello into account (see Section 4).
o If the receipt of the Hellos creates a new adjacency table
entry, the neighbor RBridge MAC (SNPA) address, Port ID, and
System ID are set from the Hello.
o The appropriate Hello holding timer for the adjacency,
depending on whether or not the Hello was received on the
Designated VLAN, is set to the Holding Time field of the Hello.
If the receipt of the Hello is creating a new adjacency table
entry, the other timer is set to expired.
o The priority of the neighbor RBridge to be the DRB is set to
the priority field of the Hello.
o The VLAN that the neighbor RBridge wants to be the Designated
VLAN on the link is set from the Hello.
o If the creation of a new adjacency table entry or the priority
update above changes the results of the DRB election on the
link, the appropriate RBridge port event (D2 or D3) occurs,
after the above actions, as described in Section 4.2.
o If there is no change in the DRB, but the neighbor Hello is
from the DRB and has a changed Designated VLAN from the
previous Hello received from the DRB, the result is a change in
Designated VLAN for the link as specified in Section 4.2.3.
An event A4 resulting in both Hello Holding timers for an adjacency
being expired and the adjacency going Down may also result in an
event D3 as described in Section 4.2.
Concerning events A6 and A7, if MTU testing is not enabled, A6 is
considered to occur immediately upon the adjacency entering the 2-Way
state, and A7 cannot occur.
See further TRILL LAN Hello receipt details in Section 7.
3.4. Adjacency State Diagram and Table
The table below shows the transitions between the states defined
above based on the events defined above:
| Event | Down | Detect | 2-Way | Report |
+-------+--------+--------+--------+--------+
| A1 | 2-Way | 2-Way | 2-Way | Report |
| A2 | Detect | Detect | 2-Way | Report |
| A3 | Detect | Detect | Detect | Detect |
| A4 | N/A | Down | Down | Down |
| A5 | N/A | Detect | Detect | Detect |
| A6 | N/A | N/A | Report | Report |
| A7 | N/A | N/A | 2-Way | 2-Way |
| A8 | Down | Down | Down | Down |
N/A indicates that the event to the left is Not Applicable in the
state at the top of the column. These events affect only a single
adjacency. The special A0 event transitions all adjacencies to Down,
as explained immediately after the list of adjacency events above.
The diagram below presents the same information as that in the state
table:
+---------------+
| Down |<--------+
+---------------+ |
| | ^ | |
A2,A3| |A8| |A1 |
| +--+ | |
| +-----------|---+
V | |
+----------------+ A4,A8 | |
+----->| Detect |------->| |
| +----------------+ | |
| | | ^ | |
| A1| |A2,A3,A5 | | |
| | +---------+ | |
| | | |
| | +------------|---+
| | | |
| V V |
|A3,A5 +----------------+ A4,A8 |
|<-----| 2-Way |------->|
| +----------------+ |
| | ^ | ^ |
| A6| | |A1,A2,A7| |
| | | +--------+ |
| | | |
| | |A7 |
| V | |
|A3,A5 +-------------+ A4,A8 |
|<-----| Report |---------->|
+-------------+
| ^
|A1,A2,A6 |
+---------+
3.5. Multiple Parallel Links
There can be multiple parallel adjacencies between neighbor RBridges
that are visible to TRILL. (Multiple low-level links that have been
bonded together by technologies such as link aggregation [802.1AX]
appear to TRILL as a single link over which only a single TRILL
adjacency could be established.)
Any such links that have pseudonodes (see Section 6) are
distinguished in the topology; such adjacencies, if they are in the
Report state, appear in LSPs as per Section 6. However, there can be
multiple parallel adjacencies without pseudonodes because they are
point-to-point adjacencies or LAN adjacencies for which a pseudonode
is not being created. Such parallel, non-pseudonode adjacencies in
the Report state appear in LSPs as a single adjacency. The cost of
such an adjacency MAY be adjusted downwards to account for the
parallel paths. Multipathing across such parallel connections can be
freely done for unicast TRILL Data traffic on a per-flow basis but is
restricted for multi-destination traffic, as described in Section
4.5.2 (point 3) and Appendix C of [RFC6325].
3.6. Insufficient Space in Adjacency Table
If the receipt of a TRILL LAN Hello would create a new adjacency
table entry (that is, would transition an adjacency out of the Down
state), there may be no space for the new entry. In that case, the
DRB election priority (see Section 4.2.1) of the new entry that would
be created is compared with that priority for the existing entries.
If the new entry is higher priority than the lowest priority existing
entry, it replaces the lowest priority existing entry, which is
transitioned to the Down state.
4. RBridge LAN Ports and DRB State
The information at an RBridge associated with each of its LAN ports
includes the following:
o Enablement bit, which defaults to enabled.
o SNPA address (usually a 48-bit MAC address) of the port.
o Port ID, used in TRILL Hellos sent on the port.
o The Holding Time, used in TRILL Hellos sent on the port.
o The Priority to be the DRB, used in TRILL Hellos sent on the
port.
o The DRB status of the port, determined as specified below.
o A 16-bit unsigned Suspension timer, measured in seconds.
o The desired Designated VLAN. The VLAN this RBridge wants to be
the Designated VLAN for the link out this port, used in TRILL
Hellos sent on the port.
o A table of zero or more adjacencies (see Section 3).
4.1. Port Table Entries and DRB Election State
The TRILL equivalent of the DIS (Designated Intermediate System) on a
link is the DRB or Designated RBridge. The DRB election state
machinery is described below.
Each RBridge port is in one of the following four DRB states:
Down:
The port is operationally down. It might be administratively
disabled or down at the link layer. In this state, there will
be no adjacency table entries for the port, and no TRILL Hellos
or other IS-IS PDUs or TRILL Data frames are accepted or
transmitted.
Suspended:
Operation of the port is suspended because there is a higher
priority port on the link with the same MAC (SNPA) address.
This is the same as the down state with the exception that
TRILL Hellos are accepted for the sole purpose of determining
whether to change the value of the Suspension timer for the
port as described below.
DRB:
The port is the DRB and can receive and transmit TRILL Data
frames.
Not DRB:
The port is deferring to another port on the link, which it
believes is the DRB, but can still receive and transmit TRILL
Data frames.
4.2. DRB Election Events
The following events can change the DRB state of a port:
D1. Expiration of the suspension timer while the port is in the
Suspended state or the enablement of the port.
D2. Adjacency table for the port changes, and there are now
entries for one or more other RBridge ports on the link that
appear to be higher priority to be the DRB than the local
port.
D3. The port is not Down or Suspended, and the adjacency table
for the port changes, so there are now no entries for other
RBridge ports on the link that appear to be higher priority
to be the DRB than the local port.
D4. Receipt of a TRILL Hello with the same MAC address (SNPA) as
the receiving port and higher priority to be the DRB as
described for event A0.
D5. The port becomes operationally down.
Event D1 is considered to occur on RBridge boot if the port is
administratively and link-layer enabled.
Event D4 causes the port to enter the Suspended state and all
adjacencies for the port to be discarded (transitioned to the Down
state). If the port was in some state other than Suspended, the
suspension timer is set to the Holding Time in the Hello that causes
event D4. If it was in the Suspended state, the suspension timer is
set to the maximum of its current value and the Holding Time in the
Hello that causes event D4.
4.2.1. DRB Election Details
Events D2 and D3 constitute losing and winning the DRB election at
the port, respectively.
The candidates for election are the local RBridge and all RBridges
with which there is an adjacency on the port in an adjacency state
other than Down state. The winner is the RBridge with highest
priority to be the DRB, as determined from the 7-bit priority field
in that RBridge's Hellos received and the local port's priority to be
the DRB field, with MAC (SNPA) address as a tiebreaker, Port ID as a
secondary tiebreaker, and System ID as a tertiary tiebreaker. These
fields are compared as unsigned integers with the larger magnitude
being considered higher priority.
Resort to the secondary and tertiary tiebreakers should only be
necessary in rare circumstances when multiple ports have the same
priority and MAC (SNPA) address and some of them are not yet
suspended. For example, RB1, that has low priority to be the DRB on
the link, could receive Hellos from two other ports on the link that
have the same MAC address as each other and are higher priority to be
the DRB. One of these two ports with the same MAC address will be
suspended, cease sending Hellos, and the Hello from it received by
RB1 will eventually time out. But, in the meantime, RB1 can use the
tiebreakers to determine which port is the DRB and thus which port's
Hello to believe for such purposes as setting the Designated VLAN on
the link.
4.2.2. Change in DRB
Events D2 and D3 result from a change in the apparent DRB on the
link. Unnecessary DRB changes should be avoided, especially on links
offering native frame service, as a DRB change will generally cause a
transient interruption to native frame service.
If a change in the DRB on the link changes the Designated VLAN on the
link, the actions specified in Section 4.2.3 are taken.
If an RBridge changes in either direction between being the
Designated RBridge and not being the Designated RBridge at a port,
this will generally change the VLANs on which Hellos are sent by that
RBridge on that port as specified in Section 4.4.3 of [RFC6325].
4.2.3. Change in Designated VLAN
Unnecessary changes in the Designated VLAN on a link should be
avoided because a change in the Designated VLAN can cause a transient
interruption to TRILL Data forwarding on the link. When practical,
all RBridge ports on a link should be configured with the same
desired Designated VLAN so that, in case the winner of the DRB
election changes, for any reason, the Designated VLAN will remain the
same.
If an RBridge detects a change in Designated VLAN on a link, then,
for all adjacency table entries for a port to that link, the RBridge
takes the following steps in the order given:
o The non-Designated VLAN Hello Holding timer is set to the
maximum of its time to expiration and the current time to
expiration of the Designated VLAN Hello Holding timer.
o The Designated VLAN Hello Holding timer is then set to expired
(if necessary), and an event A5 occurs for the adjacency (see
Section 3.3).
If the Designated VLAN for a link changes, this will generally change
the VLANs on which Hellos are sent by an RBridge port on that link as
specified in Section 4.4.3 of [RFC6325].
4.3. State Table and Diagram
The table below shows the transitions between the DRB states defined
above based on the events defined above:
| Event | Down | Suspend | DRB | Not DRB |
+-------+--------+---------+---------+---------+
| D1 | DRB | DRB | N/A | N/A |
| D2 | N/A | N/A | Not DRB | Not DRB |
| D3 | N/A | N/A | DRB | DRB |
| D4 | N/A | Suspend | Suspend | Suspend |
| D5 | Down | Down | Down | Down |
N/A indicates that the event to the left is Not Applicable in the
state at the top of the column.
The diagram below presents the same information as in the state
table:
+-------------+
| Down |<--------------+
+-+---+-------+ ^ |
| | ^ | |
D1| |D5 | | |
| +---+ |D5 |
| | |
| +--------+----+ |
| | Suspended |<---|---+
| +-+-----+-----+ | |
| D1| ^ | ^ | |
| | | |D4 | | |
| | | +---+ | |
| | | | |
| | |D4 | |
V V | | |
+---------------+-+ D5 | |
| DRB |---------->| |
+--------+--+-----+ | |
^ | | ^ | |
| D2| |D3| | |
| | +--+ | |
| | D4 | |
|D3 | +-----------------|---+
| V | |
+----+-------+-+ D5 |
| Not DRB |-------------->|
+----+---------+
| ^
|D2 |
+----+
5. MTU Matching
The purpose of MTU testing is to ensure that the links used in the
campus topology can pass TRILL IS-IS and Data frames at the RBridge
campus MTU.
An RBridge, RB1, determines the desired campus link MTU by
calculating the minimum of its originatingL1LSPBufferSize and the
originatingL1LSPBufferSize of other RBridges in the campus, as
advertised in the link state database, but not less than 1,470 bytes.
Although originatingL1LSPBufferSize in Layer 3 [IS-IS] is limited to
the range 512 to 1,492 bytes inclusive, in TRILL it is limited to the
range 1,470 to 65,535 bytes inclusive.
Although MTU testing is optional, it is mandatory for an RBridge to
respond to an MTU-probe PDU with an MTU-ack PDU [RFC6325] [RFC6326].
Use of multicast or unicast for MTU-probe and MTU-ack is an
implementation choice. However, the burden on the link is generally
minimized by multicasting MTU-probes when a response from all other
RBridges on the link is desired, such as when initializing or re-
confirming MTU, unicasting MTU-probes when a response from a single
RBridge is desired, such as one that has just been detected on the
link, and unicasting all MTU-ack frames.
RB1 can test the MTU size to RB2 as described in Section 4.3.2 of
[RFC6325]. For this purpose, MTU testing is only done in the
Designated VLAN. An adjacency that fails the MTU test at the campus
MTU will not enter the Report state or, if the adjacency is in that
state, it leaves that state. Thus, an adjacency failing the MTU test
will not be reported by the RBridge performing the test. Since
inclusion in least-cost route computation requires the adjacency to
be reported by both ends, as long as the MTU failure is noticed by
the RBridge at either end of the adjacency, it will not be so used.
If it tests MTU, RB1 reports the largest size for which the MTU test
succeeds or a flag indicating that it fails at the campus MTU. This
report always appears with the neighbor in RB1's TRILL Neighbor TLV.
RB1 MAY also report this with the adjacency in an Extended
Reachability TLV in RB1's LSP. RB1 MAY choose to test MTU sizes
greater than the desired campus MTU as well as the desired campus
MTU.
Most types of TRILL IS-IS frames, such as LSPs, can make use of the
campus MTU. The exceptions are TRILL Hellos, which must be kept
small for loop safety, and the MTU PDUs, whose size must be adjusted
appropriately for the tests being performed.
6. Pseudonodes
The Designated RBridge (DRB), determined as described above, controls
whether a pseudonode will be used on a link.
If the DRB sets the bypass pseudonode bit in its TRILL LAN Hellos,
the RBridges on the link (including the DRB) just directly report all
their adjacencies on the LAN that are in the Report state. If the
DRB does not set the bypass pseudonode bit in its TRILL Hellos, then
(1) the DRB reports in its LSP its adjacency to the pseudonode, (2)
the DRB sends LSPs on behalf of the pseudonode in which it reports
adjacency to all other RBridges on the link where it sees that
adjacency in the Report state, and (3) all other RBridges on the link
report their adjacency to the pseudonode if they see their adjacency
to the DRB as being in the Report state and do not report any other
adjacencies on the link. Setting the bypass pseudonode bit has no
effect on how LSPs are flooded on a link. It only affects what LSPs
are generated.
It is anticipated that many links between RBridges will actually be
point-to-point, in which case using a pseudonode merely adds to the
complexity. For example, if RB1 and RB2 are the only RBridges on the
link, and RB1 is DRB, then if RB1 creates a pseudonode that is used,
there are 3 LSPs: for, say, RB1.25 (the pseudonode), RB1, and RB2,
where RB1.25 reports connectivity to RB1 and RB2, and RB1 and RB2
each just say they are connected to RB1.25. Whereas if DRB RB1 sets
the bypass pseudonode bit in its Hellos, then there will be only 2
LSPs: RB1 and RB2 each reporting connectivity to each other.
A DRB SHOULD set the bypass pseudonode bit in its Hellos if it has
not seen at least two simultaneous adjacencies in the Report state
since it last rebooted or was reset by network management.
7. TRILL-Hello Reception and Transmission
This section provides further details on the receipt and transmission
of TRILL LAN Hellos.
TRILL LAN Hellos, like all TRILL IS-IS frames, are primarily
distinguished from Layer 3 IS-IS frames by being sent to the
All-IS-IS-RBridges multicast address (01-80-C2-00-00-41). TRILL
IS-IS frames also have the L2-IS-IS Ethertype (0x22F4) and are
Ethertype encoded.
Although future extensions to TRILL may include use of Level 2 IS-IS,
[RFC6325] specifies TRILL using a single Level 1 Area with Area
Address zero (see Section 4.2 of [RFC6326]).
IS-IS Layer 3 routers are frequently connected to other Layer 3
routers that are part of a different routing domain. In that case,
the externalDomain flag (see [IS-IS]) is normally set for the port
through which such a connection is made. The setting of this flag to
"true" causes no IS-IS PDUs to be sent out the port and any IS-IS
PDUs received to be discarded, including Hellos. RBridges operate in
a different environment where all neighbor RBridges merge into a
single campus. For loop safety, RBridges do not implement the
externalDomain flag or implement it with the fixed value "false".
They send and receive TRILL LAN Hellos on every port that is not
disabled or configured as point-to-point.
7.1. Transmitting TRILL Hellos
TRILL LAN Hellos are sent with the same timing as Layer 3 IS-IS LAN
Hellos [IS-IS]; however, no Hellos are sent if a port is in the
Suspended or Down states.
TRILL-Hello PDUs SHOULD NOT be padded and MUST NOT be sent exceeding
1,470 octets; however, a received TRILL Hello longer than 1,470
octets is processed normally.
TRILL-Hello PDU headers MUST conform to the following:
o Maximum Area Addresses equal to 1.
o Circuit Type equal to 1.
Each TRILL Hello MUST contain an Area Addresses TLV listing only the
single Area zero, and an MT Port Capabilities TLV containing a VLAN-
FLAGS sub-TLV [RFC6326]. If a Protocols Supported TLV is present, it
MUST list the TRILL NLPID (0xC0).
The TRILL Neighbor TLV sent in a Hello MUST show the neighbor
information, as sensed by the transmitting RBridge, for the VLAN on
which the Hello is sent. Since implementations conformant to this
document maintain such information on a per-VLAN basis only for the
Designated VLAN, such implementations only send the TRILL Neighbor
TLV in TRILL Hellos on the Designated VLAN.
It is RECOMMENDED that, if there is sufficient room, a TRILL Neighbor
TLV or TLVs, as described in Section 4.4.2.1 of [RFC6325], covering
the entire range of MAC addresses and listing all adjacencies with a
non-zero Designated VLAN Hello Holding time, or an empty list of
neighbors if there are no such adjacencies, be in TRILL Hellos sent
on the Designated VLAN. If this is not possible, then TRILL Neighbor
TLV's covering sub-ranges of MAC addresses should be sent so that the
entire range is covered reasonably promptly. Delays in sending TRILL
Neighbor TLVs will delay the advancement of adjacencies to the Report
state and the discovery of some link failures. Rapid (for example,
sub-second) detection of link or node failures is best addressed with
a protocol designed for that purpose, such as Bidirectional
Forwarding Detection (BFD) [RFC5880], use of which with TRILL will be
specified in a separate document.
To ensure that any RBridge RB2 can definitively determine whether RB1
can hear RB2, RB1's neighbor list MUST eventually cover every
possible range of IDs, that is, within a period that depends on RB1's
policy and not necessarily within any specific period such as its
Holding Time. In other words, if X1 is the smallest ID reported in
one of RB1's neighbor lists, and the "smallest" flag is not set, then
X1 MUST appear in a different neighbor list as well, as the largest
ID reported in that fragment. Or lists may overlap, as long as there
is no gap, such that some range, say between Xi and Xj, never appears
in any list.
A TRILL Hello MAY also contain any TLV permitted in a Layer 3 IS-IS
Hello. TLVs that are unsupported/unknown are ignored.
7.2. Receiving TRILL Hellos
Assuming a frame has the All-IS-IS-RBridges multicast address and
L2-IS-IS Ethertype, it will be examined to see if it appears to be an
IS-IS PDU. If so, and it appears to be a LAN Hello PDU, the
following tests are performed.
o If the Circuit Type field is not 1, the PDU is discarded.
o If the PDU does not contain an Area Address TLV or it contains
an Area Address TLV that is not the single Area Address zero,
it is discarded.
o If the Hello includes a Protocols Supported TLV that does not
list the TRILL NLPID (0xC0), it is discarded. It is acceptable
if there is no Protocols Supported TLV present.
o If the Hello does not contain an MT Port Capabilities TLV
containing a VLAN-FLAGS sub-TLV [RFC6326], it is discarded.
o If the maximumAreaAddresses field of the PDU is not 1, it is
discarded.
o If IS-IS authentication is in use on the link and the PDU
either has no Authentication TLV or validation of that
Authentication TLV fails, it is discarded.
If none of the rules in the list above has been satisfied, and the
frame is parseable, it is assumed to be a well-formed TRILL Hello
received on the link. It is treated as an event A0, A1, A2, or A3
based on the criteria listed in Section 3.3.
8. Multiple Ports on the Same Link
It is possible for an RBridge RB1 to have multiple ports on the same
link that are not in the Suspended state. It is important for RB1 to
recognize which of its ports are on the same link. RB1 can detect
this condition based on receiving TRILL LAN Hello messages with the
same LAN ID on multiple ports.
The DRB election is port-based (see Section 4) and only the Hellos
from the elected port can perform certain functions such as dictating
the Designated VLAN or whether a pseudonode will be used; however,
the election also designates the RBridge with that port as DRB for
the link. An RBridge may choose to load split some tasks among its
ports on the link if it has more than one and it is safe to do so as
described in Section 4.4.4 of [RFC6325].
9. Security Considerations
This memo provides improved documentation of some aspects of the
TRILL base protocol standard, particularly four aspects of the TRILL
LAN Hello protocol. It does not change the security considerations
of the TRILL base protocol. See Section 6 of [RFC6325].
10. References
10.1. Normative References
[IS-IS] ISO/IEC 10589:2002, Second Edition, "Intermediate
System to Intermediate System Intra-Domain Routing
Exchange Protocol for use in Conjunction with the
Protocol for Providing the Connectionless-mode Network
Service (ISO 8473)", 2002.
[RFC1195] Callon, R., "Use of OSI IS-IS for routing in TCP/IP and
dual environments", RFC 1195, December 1990.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC6325] Perlman, R., D. Eastlake, D. Dutt, S. Gai, and A.
Ghanwani, "RBridges: Base Protocol Specification", RFC
6325, July 2011.
[RFC6326] Eastlake, D., Banerjee, A., Dutt, D., Perlman, R., and
A. Ghanwani, "TRILL Use of IS-IS", RFC 6326, July 2011.
10.2. Informative References
[802.1AX] "IEEE Standard for Local and metropolitan area networks
/ Link Aggregation", 802.1AX-2008, 1 January 2008.
[802.1Q-2005] "IEEE Standard for Local and metropolitan area networks
/ Virtual Bridged Local Area Networks", 802.1Q-2005, 19
May 2006.
[FCoE] From www.t11.org discussion of "FCoE Max Size"
generated from T11/09-251v1, 04/27/2009, "FCoE frame or
FCoE PDU".
[RFC3719] Parker, J., Ed., "Recommendations for Interoperable
Networks using Intermediate System to Intermediate
System (IS-IS)", February 2004.
[RFC5880] Katz, D. and D. Ward, "Bidirectional Forwarding
Detection (BFD)", RFC 5880, June 2010.
11. Acknowledgements
The authors of [RFC6325], those listed in the Acknowledgements
section of [RFC6325], and the contributions of Jari Arkko, Ayan
Banerjee, Les Ginsberg, Sujay Gupta, David Harrington, Pete McCann,
Erik Nordmark, and Mike Shand, to this document, are hereby
acknowledged.
Authors' Addresses
Donald E. Eastlake, 3rd
Huawei Technologies
155 Beaver Street
Milford, MA 01757 USA
Phone: +1-508-333-2270
EMail: d3e3e3@gmail.com
Radia Perlman
Intel Labs
2200 Mission College Blvd.
Santa Clara, CA 95054-1549 USA
Phone: +1-408-765-8080
EMail: Radia@alum.mit.edu
Anoop Ghanwani
Brocade
130 Holger Way
San Jose, CA 95134 USA
Phone: +1-408-333-7149
EMail: anoop@alumni.duke.edu
Dinesh G. Dutt
Cisco Systems
170 Tasman Drive
San Jose, CA 95134-1706 USA
Phone: +1-408-527-0955
EMail: ddutt@cisco.com
Vishwas Manral
Hewlett Packard Co.
19111 Pruneridge Ave,
Cupertino, CA 95014 USA
Phone: +1-408-447-1497
EMail: vishwas.manral@hp.com
|
Comment about this RFC, ask questions, or add new information about this topic: