Last Call Review of draft-ietf-6man-rfc1981bis-04
I am the assigned Gen-ART reviewer for this draft. The General Area
Review Team (Gen-ART) reviews all IETF documents being processed
by the IESG for the IETF Chair. Please treat these comments just
like any other last call comments.
For more information, please see the FAQ at
Reviewer: Stewart Bryant
Review Date: 9/Feb/2017
IETF LC End Date: 1/Mar/2017
IESG Telechat date: unknown
Summary: This draft is on the right track but has open issues, described in the review.
This review together with the lengthy discussion on the IETF list
suggest that this draft has a number of issues that need to be
addressed before publication.
I wonder if we would best serve both our future and our heritage
if we declared RFC1981 as historic, and either left the idea there,
or declared it as historic and wrote a new text from a clean start?
Nits points out a number of faults with the document, but the only
one of substance is:
The text has a lot of RFC2119 language, but no RFC2119 declaration.
The document could use a thorough RFC2119 scrub
The document lists the three original authors one with an affiliation
change, but no email addresses. Has this been agreed with the original
authors, and have arrangements been put in place for the RFC editor
to process auth48?
It is concerning that the draft does not talk in any detail about
how modern ECMP works, i.e. using the five tuple, and noting that
the PMTU may be different depending on the transport layer port numbers.
Given that a very large fraction of packets will traverse an MPLS
network at some point, I am surprised that there is no text talking
about the importance of providing support for this feature in the
MPLS domain. RFC3988 talks to this point, but is only experimental.
If flows [I-D.ietf-6man-rfc2460bis] are in use, an implementation
could use the flow id as the local representation of a path. Packets
sent to a particular destination but belonging to different flows may
use different paths, with the choice of path depending on the flow
id. This approach will result in the use of optimally sized packets
on a per-flow basis, providing finer granularity than PMTU values
maintained on a per-destination basis.
SB> How widely is flow-id supported in networks? I thought that the
SB> current position was that it was unreliable as an ECMP indicator
SB> and thus routers tended to glean information from the packet themselves.
Note: if the original packet contained a Routing header, the
Routing header should be used to determine the location of the
destination address within the original packet. If Segments Left
is equal to zero, the destination address is in the Destination
Address field in the IPv6 header. If Segments Left is greater
than zero, the destination address is the last address
(Address[n]) in the Routing header.
SB> So this has the effect that a traffic engineered packet and
SB> a non-traffic engineered packet will have the lower of the
SB> two PMTUs. This was all harmless when source routing was a curiosity
SB> as far as mainstream networking was concerned, but may be
SB> more of a problem as a result of the SPRING work.
5.3. Purging stale PMTU information
Internetwork topology is dynamic; routes change over time. While the
local representation of a path may remain constant, the actual
path(s) in use may change. Thus, PMTU information cached by a node
can become stale.
If the stale PMTU value is too large, this will be discovered almost
immediately once a large enough packet is sent on the path. No such
mechanism exists for realizing that a stale PMTU value is too small,
so an implementation should "age" cached values. When a PMTU value
has not been decreased for a while (on the order of 10 minutes), the
PMTU estimate should be set to the MTU of the first-hop link, and the
packetization layers should be notified of the change. This will
cause the complete Path MTU Discovery process to take place again.
SB> Should that be an RFC2119 SHOULD?
SB> The impact of this advice is going to be a disruption to what might
SB> be a critical service every 10 mins.
SB> Should there be some advice along the lines of noting the
SB> importance of service delivery as part of deciding whether to
SB> test for bigger PMTU vs improving efficiency?
IPv6 defines a standard mechanism for a node to discover the
PMTU of an arbitrary path.
SB> Do you mean "This document defines ....."? Otherwise this needs
SB> a reference.
An extension to Path MTU Discovery defined in this document can be
found in [RFC4821]. It defines a method for Packetization Layer Path
SB> Rather than have the reader figure out what "It" is, perhaps
Upon receipt of such a
message, the source node reduces its assumed PMTU for the path based
on the MTU of the constricting hop as reported in the Packet Too Big
SB> We should perhaps state up front that this procedure
SB> hunts for the worst case of the ECMP set associated with the
SB> ingress nodes PMTU classifier.
If a node receives a Packet Too Big message reporting a next-hop MTU
that is less than the IPv6 minimum link MTU, it should discard it.
SB> Should that be an RFC2119 SHOULD?
5.2. Storing PMTU information
Ideally, a PMTU value should be associated with a specific path
traversed by packets exchanged between the source and destination
nodes. However, in most cases a node will not have enough
information to completely and accurately identify such a path.
Rather, a node must associate a PMTU value with some local
representation of a path. It is left to the implementation to select
the local representation of a path.
SB> Is it worth noting the five tuple since that is how a lot of
SB> load balancers work?
The set of paths in use to a
particular destination is expected to be small, in many cases
consisting of a single path.
SB> I am not sure that remains true in modern networks.
One approach to implementing PMTU aging is to associate a timestamp
field with a PMTU value. This field is initialized to a "reserved"
value, indicating that the PMTU is equal to the MTU of the first hop
link. Whenever the PMTU is decreased in response to a Packet Too Big
message, the timestamp is set to the current time.
Once a minute, a timer-driven procedure runs through all cached PMTU
values, and for each PMTU whose timestamp is not "reserved" and is
older than the timeout interval:
- The PMTU estimate is set to the MTU of the first hop link.
- The timestamp is set to the "reserved" value.
- Packetization layers using this path are notified of the increase.
SB> Such detailed implementation advice is uncommon in modern RFCs. It has
SB> the disadvantage of de-facto standardizing something that should be left to
SB> the innovation of the implementer.
5.4. TCP layer actions
SB> TCP implementations have moved on a lot since this section was
SB> written. Is this still current best practise?
5.5. Issues for other transport protocols
Some transport protocols (such as ISO TP4 [ISOTP]) are not allowed to
repacketize when doing a retransmission.
SB> How much TP4 is there going over IPv6? Doesn't this example
SB> show the IETF as not being in the modern age?
upper layer a protocol layer immediately above IPv6.
Examples are transport protocols such as TCP and
UDP, control protocols such as ICMP, routing
protocols such as OSPF, and internet or lower-
layer protocols being "tunneled" over (i.e.,
encapsulated in) IPv6 such as IPX, AppleTalk, or
SB> Everything in the list above is in the well known list, except
SB> IPX, so technically it needs expansion. However it might be nice
SB> to use some modern example in common use.
link a communication facility or medium over which
nodes can communicate at the link layer, i.e.,
the layer immediately below IPv6. Examples are
Ethernets (simple or bridged); PPP links; X.25,
Frame Relay, or ATM networks; and internet (or
higher) layer "tunnels", such as tunnels over
IPv4 or IPv6 itself.
SB> Technically X.25 needs a reference, since it is not "well known"
path the set of links traversed by a packet between a
source node and a destination node.
SB> Is it a set of links, or a set of links and nodes?
the value of MMS_S, the "maximum send transport-message size".
SB> The modern convention is full-name(abbreviation)
The Sun Network File System (NFS) uses a Remote Procedure Call (RPC)
protocol [RPC] that, when used over UDP, in many cases will generate
payloads that must be fragmented even for the first-hop link. This
might improve performance in certain cases, but it is known to cause
reliability and performance problems, especially when the client and
server are separated by routers.
SB> Perhaps this should point to RFC7530 (the current NFS Spec), assuming
SB> the behaviour description is still correct.
The former can be accomplished by associating a flag with the path;
when a packet is sent on a path with this flag set, the IP layer does
not send packets larger than the IPv6 minimum link MTU.
SB> We do not normally give this level of implementation advice