- Home
- RFC 10069
RFC 10069: Managing Multiple Paths for a QUIC Connection
- Y. Liu, Ed.,
- Y. Ma,
- Q. De Coninck, Ed.,
- O. Bonaventure,
- C. Huitema,
- M. Kühlewind, Ed.
22 subscribed
Added to 17 Sets
Abstract
This document specifies a multipath extension for the QUIC protocol to enable the simultaneous usage of multiple paths for a single connection. It introduces explicit path identifiers to create, delete, and manage multiple paths. This document does not specify address discovery or management, nor how applications using QUIC schedule traffic over multiple paths.¶
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 7841.¶
Information about the current status of this document, any
errata, and how to provide feedback on it may be obtained at
https://
Copyright Notice
Copyright (c) 2026 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
(https://
1. Introduction Skip to TOC
This document specifies an extension to QUIC version 1 [QUIC-TRANSPORT] to enable the simultaneous usage of multiple paths for a single connection, using the same or different 4-tuples (of source/destination port numbers and source/destination IP addresses).¶
Connection migration as specified in Section 9 of [QUIC-TRANSPORT] directs a peer to switch sending through a new preferred path and, if successful, to release resources associated with the old path. The multipath extension specified in this document builds on this mechanism but introduces a path identifier, or path ID, to manage connection IDs and packet number spaces per path, enabling the use of multiple paths simultaneously. As such, a path in this extension is defined by its path ID. Note that it therefore is possible to have multiple paths / path IDs for the same 4-tuple.¶
The connection ID of a packet binds the packet to a path ID and therefore to a packet number space. That means each connection ID is associated with exactly one path ID but multiple connection IDs are usually issued for each path ID. The same path ID is used in both directions, starting with 0 for the initial path. Path IDs are monotonically increasing and cannot be reused.¶
This extension uses multiple packet number spaces, specifically one for each path. Each path ID-specific packet number space starts at packet number 0. As such, each path maintains distinct packet number states for sending and receiving packets, as in [QUIC-TRANSPORT]. Using multiple packet number spaces enables direct use of the loss detection and congestion control mechanisms defined in [QUIC-RECOVERY] on a per-path basis. However, use of multiple packet number spaces requires non-zero connection IDs in order to identify the path and the respective packet number space, as well as a modified Authenticated Encryption with Associated Data (AEAD) calculation including the path ID (see Section 2.4).¶
As such, this extension specifies a departure from the specification of path management in Section 9 of [QUIC-TRANSPORT] and therefore requires a new transport parameter, as specified in Section 2.1, to indicate support of the multipath extension specified in this document.¶
Further, this document specifies the needed path management mechanisms for path initiation in Section 3.1, handling of per-path connection IDs in Section 3.2, signaling of preferred path usage in Section 3.3, and explicit removal of paths that have been abandoned in Section 3.4. Note that in this extension, a QUIC server does not initiate the creation of a path, but it has to validate a new path created by a client.¶
This extension does not cover address discovery and management. The extension specified in this document can be used as is if address management and the actual decision to set up or tear down paths are handled by the application. This document does not prevent future extensions from defining mechanisms to cope with the remaining scenarios.¶
Further, this document does not specify scheduling algorithms that define how multiple, simultaneously open paths are used to send packets. As these differ depending on application requirements, only some basic implementation guidance is discussed in Section 5. This extension can be used with different scheduling algorithms that, e.g., can range from support for failover to simultaneous use of the aggregated capacity across all open paths. There are currently no IETF specifications that define scheduling algorithms for simultaneously (i.e., concurrently) using multiple paths.¶
Specifically, while failover between Wi-Fi and mobile networks is a well-known multipath use case, it only temporarily uses two paths at the same time to avoid transmission pauses. Simultaneous path usage generally, however, needs more consideration than specified in this document to avoid negative performance impacts, e.g., when stream data is distributed over multiple paths with different delays.¶
The operational considerations for QUIC are addressed in [RFC9312]. They also apply to QUIC connections using the extensions defined in this document. An additional complexity is that applications might use a combination of monitored and non-monitored paths, but that complexity already exists when using path migration as defined in [QUIC-TRANSPORT].¶
1.1. Conventions and Definitions Skip to TOC
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document uses the terminology defined in [QUIC-TRANSPORT]. When this document uses the term "path", it refers to the notion of "network path" used in [QUIC-TRANSPORT].¶
The packet diagrams in this document use the conventions defined
in Section 1.3 of [QUIC-TRANSPORT], including the notation (i) to denote
variable-
2. Connection Lifecycle and Packet Protection Skip to TOC
This document defines a new transport parameter initial_
The transport parameter negotiation enables incremental deployment of implementations of the multipath extension.¶
2.1. initial_max_path_id Transport Parameter
Skip to TOC
The new transport parameter is defined as follows:¶
- initial_
max_ path_ id (parameter ID 0x3e): - Parameter value is a
variable-
length integer specifying the maximum path ID an endpoint is willing to maintain at connection initiation. This value MUST NOT exceed 232-1, which is the maximum allowed value for the path ID due to restrictions on the nonce calculation (see Section 2.4).¶
The initial_
Setting initial_
If an initial_
When advertising the initial_
Cipher suites with a nonce shorter than 12 bytes cannot be used together with
the multipath extension. If such a cipher suite is selected and the use of the
multipath extension is supported, endpoints MUST abort the handshake with
an error of type TRANSPORT_
The initial_
2.2. Relation to Other Transport Parameters Skip to TOC
When the QUIC multipath extension is used, the active_
If an endpoint receives a disable_
If the server uses the preferred_
2.3. Handling ACK and PATH_ACK in 0-RTT and 1-RTT
Skip to TOC
The PATH_
After the handshake concludes with support for the multipath extension,
endpoints SHOULD use PATH_
2.4. Nonce Calculation after Handshake Completion Skip to TOC
Section 5.3 of [QUIC-TLS] specifies AEAD usage, particularly the use of a nonce, N, formed by combining the packet protection Initialization Vector (IV) with the packet number. When multiple packet number spaces are used, the packet number alone would not guarantee the uniqueness of the nonce. Therefore, the nonce N is calculated for 1-RTT packets if the multipath extension is used by combining the packet protection IV with the packet number and with the 32 bits of the path ID. In order to guarantee the uniqueness of the nonce, the path ID is limited to a max value of 232-1, as specified in Section 2.1.¶
To calculate the nonce, a 96-bit Path-
PPN {
Path ID (32),
Zeroes (2) = 0b00,
Packet Number (62)
}The IV length is equal to the nonce length. If the IV is larger than 96 bits, the path-
IV(12); N(12) = IV xor PPN;
For example, assuming the IV value is 0x6b26114
IV: 6b26114b9cba2b63a9e8dd4f ⊕ PPN: 00000003000000000000d431 ------------------------------------ Nonce: 6b2611489cba2b63a9e8097e
2.5. Key Phase Update Process Skip to TOC
The Key Phase bit update process is specified in Section 6 of [QUIC-TLS]. The general principles of key update are not changed in this specification. Following [QUIC-TLS], the Key Phase bit is used to indicate which packet protection keys are used to protect the packet. The Key Phase bit is toggled to signal each subsequent key update.¶
Because of network delays, packets protected with the older key might arrive later than the packets protected with the new key. However, receivers can solely rely on the Key Phase bit to determine the corresponding packet protection key, assuming that there is a sufficiently large interval between two consecutive key updates (Section 6.5 of [QUIC-TLS]).¶
When this specification is used, endpoints SHOULD wait for at least three times the largest Probe Timeout (PTO) (see Section 6.2 of [QUIC-RECOVERY]) among all the paths before initiating a new key update after receiving an acknowledgment that confirms the receipt of the previous key update. This interval is different from that in [QUIC-TLS], which used three times the PTO of the single path.¶
Since packets that arrive after their decryption key has been discarded will be dropped, the choice of three times the largest PTO is a trade-off: Longer delays reduce the probability of losing packets, but keeping old keys longer can negatively impact the security of the protocol. The use of three times the largest PTO aims to minimize packet lost for all paths and therefore limits the impact on performance.¶
Following Section 5.4 of [QUIC-TLS], the Key Phase bit is protected, so sending multiple packets with the Key Phase bit flipping at the same time should not cause activity across different paths to be linkable by an observer.¶
2.6. Connection Closure Skip to TOC
CONNECTION_
3. Path Management Skip to TOC
After completing the handshake indicating multipath support, endpoints can start using multiple paths. In accordance with Section 9 of [QUIC-TRANSPORT], this extension only enables clients to open a new path. The client can open a new path when both endpoints have issued available connection IDs for at least one unused, common path ID, as the same path ID is used in both directions.¶
This document specifies path initiation (see Section 3.1), issuing and retirement of per-path connection IDs (see Section 3.2), path status management (see Section 3.3), and path closure (see Section 3.4). However, this document does not specify when a client decides to initiate or close a path or how multiple open paths are used for sending.¶
For path management, this extension specifies the following frames in Section 4:¶
-
PATH_
ABANDON (see Section 4.2)¶ -
PATH_
STATUS_ BACKUP (see Section 4.3)¶ -
PATH_
STATUS_ AVAILABLE (see Section 4.3)¶ -
PATH_
NEW_ CONNECTION_ ID (see Section 4.4)¶ -
PATH_
RETIRE_ CONNECTION_ ID (see Section 4.5)¶ -
MAX_
PATH_ ID (see Section 4.6)¶ -
PATHS_
BLOCKED (see Section 4.7)¶ -
PATH_
CIDS_ BLOCKED (see Section 4.7)¶
3.1. Path Initiation and Validation Skip to TOC
To open a new path, an endpoint MUST use a new connection ID associated with an unused path ID. An endpoint MUST use a connection ID associated to the same path ID as used in the packet received by the endpoint when it intends to send packets on the same path.¶
A client that wants to use a new path MUST validate the peer's address as described in Section 8.2 of [QUIC-TRANSPORT], unless it has previously validated the 4-tuple used for that path.¶
After receiving packets from the
client on a new path, if the server decides to use the new path,
the server MUST validate the peer's address before sending any data
as described in (Section 8.2 of [QUIC-TRANSPORT]),
unless it has previously validated the 4-tuple used for that path.
Until the client's address is
validated, the anti-
If an endpoint sends a PATH_
The server might receive packets for a yet unused path ID that do not
contain a PATH_
Each endpoint MUST also validate that a minimum QUIC packet MTU of 1200 bytes is supported on the path. This can be done during initial path validation or separately later if the amplification limit prevents it initially, as specified in Section 8.2.1 of [QUIC-TRANSPORT].¶
An endpoint that receives packets on a new path and does not want to establish
this path is expected to close the path by sending a PATH_
An endpoint that has no active connection ID for this path or
lacks other resources to immediately configure a new path could
delay sending the PATH_
PATH_
If validation succeeds, the client can continue to use the path. If validation fails, the client MUST NOT use the path and can remove any status associated to the path initiation attempt. As the used path ID is consumed either way, the endpoint MUST explicitly close the path, as specified in Section 3.4.¶
3.1.1. Path Establishment Example Skip to TOC
In the example below, it is assumed that both endpoints have
indicated an initial_
Client Server
(Provide new CIDs for path 1 on an existing path 0)
1-RTT[X]: DCID=S0, PATH_NEW_CONNECTION_ID[C1, Seq=0, PathID=1] -->
<-- 1-RTT[Y]: DCID=C0,
PATH_NEW_CONNECTION_ID[S1, Seq=0, PathID=1],
PATH_ACK[PathID=0, PN=X]
<-- 1-RTT[Y+1]: DCID=C0, PATH_NEW_CONNECTION_ID[S2, Seq=0,
PathID=2]
...
(start sending packets on a new path using path ID 1)
1-RTT[0]: DCID=S1, PATH_CHALLENGE[X] -->
<-- 1-RTT[0]: DCID=C1, PATH_RESPONSE[X], PATH_CHALLENGE[Y],
PATH_ACK[PathID=1, PN=0]
1-RTT[1]: DCID=S1, PATH_RESPONSE[Y],
PATH_ACK[PathID=1, PN=0], ... -->In Figure 4, the endpoints first exchange
new available connection IDs with the PATH_
Before the client opens a new path by sending a packet on that path
with a PATH_
3.1.2. Relation to Probing and Migration Skip to TOC
Section 9.1 of [QUIC-TRANSPORT] introduces the concept of
"probing" and "non-probing" frames. A packet that contains at least
one "non-probing" frame is a "non-probing" packet.
Migration, as specified in Section 9.2 of [QUIC-TRANSPORT], is initiated
by sending packets containing non-probing frames on a new (validated) path;
however, it uses the same path ID as on the old path.
When the multipath
extension is negotiated, the reception of any packet (no matter if
"probing" or "non-probing") on a new path with a new, so far unused
path ID does not impact the path status of any existing path. Therefore, any frame can be sent on a new path with a new path ID at any time
as long as the anti-
An endpoint could receive a packet with a connection ID associated to an active path ID where the packet's 4-tuple does not match the 4-tuple currently used with that path ID. This MUST be treated as path migration, as specified in Section 9.3 of [QUIC-TRANSPORT], with the constraint that all connection IDs used during path migration MUST be associated with the current path ID of the path being migrated.¶
3.1.3. Address Validation Token Skip to TOC
As specified in Section 9.3 of [QUIC-TRANSPORT], the server is expected to send a new address validation token to a client following the successful validation of a new client address. The client will receive several tokens. When considering using a token for subsequent connections, it might be difficult for the client to pick the "right" token among multiple tokens obtained in a previous connection. The client is likely to fall back to the strategy specified in Section 8.1.3 of [QUIC-TRANSPORT], i.e., picking the last received token. To avoid issues when clients make the "wrong" choice, a server SHOULD issue tokens that are capable of validating any of the previously validated addresses. Including more addresses increases the probability that the token will be useful in the future but at the cost of a larger token. Further guidance on token usage can be found in Section 8.1.3 of [QUIC-TRANSPORT].¶
3.2. Handling Connection IDs Skip to TOC
When the multipath extension is used,
endpoints have to use the PATH_
Endpoints SHOULD also use PATH_
3.2.1. Issuing New Connection IDs Skip to TOC
In order to let the peer open new paths, it is RECOMMENDED to proactively issue at least one Connection ID for each unused path ID up to the minimum of the peer's and the local maximum path ID limits.¶
If an endpoint does not want to issue connection IDs for all unused path IDs for any reason, it SHOULD NOT introduce discontinuity in the issuing of path IDs, as path initiation requires available connection IDs for the same path ID on both sides. For instance, if the maximum path ID limit is 2 and the endpoint wants to provide connection IDs for only one path ID inside range [1, 2], it should select path ID 1 (and not path ID 2).¶
Similarly, endpoints SHOULD consume path IDs in a continuous way, i.e., when creating paths. However, endpoints cannot expect to receive new connection IDs or path initiation attempts with in-order use of path IDs due to out-of-order delivery or path validation failure.¶
Each endpoint maintains the set of connection IDs received from its peer for each path, any of which it can use when sending packets on that path; see also Section 5.1 of [QUIC-TRANSPORT]. Usually, it is desired to provide at least one additional connection ID for all used paths to allow for (unintentional) migration events (Section 9.5 of [QUIC-TRANSPORT]).¶
As further specified in Section 5.1 of [QUIC-TRANSPORT], connection IDs cannot be issued more than once on the same connection and therefore are unique for the scope of the connection, regardless of the associated path ID.¶
Endpoints MUST NOT issue new connection IDs with path IDs greater than
the Maximum Path Identifier field in MAX_
When an endpoint cannot open a path because there are no unused path IDs,
it SHOULD either send a MAX_
3.2.2. Rotating and Retiring Connection IDs Skip to TOC
Section 5.1.2 of [QUIC-TRANSPORT] indicates that an endpoint can change the connection ID it uses to another available one at any time during the connection. For the extension specified in this document, endpoints MUST only rotate to another connection ID associated with the same path ID. Use of a connection ID associated with another path ID will be considered as an attempt to open a new path instead.¶
An endpoint is supposed to retire any connection ID that is not being used,
and the server is expected to provide
replacements, as specified in Section 5.1.2 of [QUIC-TRANSPORT].
As such, when receiving a PATH_
While it is expected that the peer provides at least one unused connection ID
for all active paths using the PATH_
Retirement of connection IDs will not retire the path ID that corresponds to the connection ID or any other path resources as the packet number space is associated to the path ID.¶
The peer that sends the PATH_
3.3. Path Status Management Skip to TOC
An endpoint can send PATH_
Each peer indicates its preference of path usage independently of the other peer. That means that peers could have different usage preferences for the same path. Depending on the data sender's decisions, this might lead to usage of paths that have been indicated as "backup" by the peer or non-usage of some locally available paths.¶
PATH_
PATH_
Similarly, if no frame indicating a path usage preference was received for a certain path, the preference of the peer is unknown and the sender needs to decide based on its own local logic if the path should be used.¶
If an endpoint starts using a backup path
because it has detected issues on the paths marked as "available", it is RECOMMENDED
to update its own path state signaling such that the peer avoids using the broken path.
An endpoint that detects a path breakage can also explicitly close the path
by sending a PATH_
The PATH_
3.4. Path Close Skip to TOC
At any time in the connection, each endpoint can decide to
abandon a path, for example, following changes in local
connectivity or local preferences.
An endpoint that wants to abandon a path MUST explicitly
close the path by sending a PATH_
Endpoints that send a PATH_
PATH_
When an endpoint receives a PATH_
It is also possible that an endpoint will receive a PATH_
If a peer sends a PATH_
After a path is abandoned, the path ID MUST NOT be reused for new paths, as the path ID is part of the nonce calculation (see Section 2.4).¶
If a PATH_
Note that other explicit closing mechanisms of [QUIC-TRANSPORT] still
apply on the whole connection. In particular, the reception of either a
CONNECTION_
3.4.1. Path Closure Example Skip to TOC
In the example below, the client wants to close the path with path ID 0.
It sends the PATH_
Client Server
(client tells server to abandon a path with path ID 0)
1-RTT[X]: DCID=S1 PATH_ABANDON[path ID=0]->
(server tells client to abandon a path)
<-1-RTT[Y]: DCID=C1 PATH_ABANDON[path ID=0],
PATH_ACK[PATH ID=1, PN=X]
1-RTT[U]: DCID=S1 PATH_ACK[path ID=1, PN=Y] ->Note that if the PATH_
3.4.2. Avoiding Spurious Stateless Resets Skip to TOC
Due to network delays, packets sent on an abandoned path can
arrive well after the connection IDs have been retired.
If not recognized as bound to the local
connection, such a packet triggers the peer to send a Stateless Reset
packet. The requirement to retain knowledge of connection ID and about
the packet number space for 3 PTOs
after receiving a PATH_
Section 10.3 of [QUIC-TRANSPORT] specified that the Stateless Reset Tokens associated with retired connection IDs cannot be used to identify Stateless Reset packets. The immediate retirement of connection IDs received from the peer for an abandoned path guarantees that spurious Stateless Reset packets sent by the peer will not cause the closure of the QUIC connection.¶
3.4.3. Handling PATH_ACK for Abandoned Paths
Skip to TOC
When an endpoint sends a PATH_
PATH_
When an endpoint finally deletes all state associated with the path,
the packets sent over the path and not yet acknowledged MUST be considered lost.
PATH_
4. New Frames Skip to TOC
All frames defined in this document MUST only be sent in 1-RTT packets.
If an endpoint receives a multipath-
Receipt of multipath-
If an endpoint receives a multipath-
4.1. PATH_ACK Frame
Skip to TOC
The PATH_
PATH_
PATH_ACK Frame {
Type (i) = 0x3e..0x3f,
Path Identifier (i),
Largest Acknowledged (i),
ACK Delay (i),
ACK Range Count (i),
First ACK Range (i),
ACK Range (..) ...,
[ECN Counts (..)],
}Compared to the ACK frame specified in [QUIC-TRANSPORT], the following field is added:¶
- Path Identifier:
-
The path ID associated with the packet number space of the 0-RTT and 1-RTT packets that are acknowledged by the PATH_
ACK frame.¶
4.2. PATH_ABANDON Frame
Skip to TOC
The PATH_
PATH_
PATH_ABANDON Frame {
Type (i) = 0x3e75,
Path Identifier (i),
Error Code (i),
}PATH_
- Path Identifier:
-
The path ID associated to the to-
be- abandoned path.¶ - Error Code:
-
A variable-
length integer that indicates the reason for abandoning this path. NO_ ERROR (0x0) indicates that the path is being abandoned without any error being encountered. Other error codes can be found in Section 4.2.1.¶
PATH_
Use of the PATH_
4.2.1. Error Codes Skip to TOC
QUIC transport error codes are 62-bit unsigned integers
(see Section 20.1 of [QUIC-TRANSPORT]). In addition to
NO_
- APPLIC
ATION_ ABANDON_ PATH (0x3e): -
The endpoint is abandoning the path at the request of the application.¶
- PATH_
RESOURCE_ LIMIT_ REACHED (0x3e75): -
The endpoint is abandoning the path because it cannot allocate sufficient resources to maintain it.¶
- PATH_
UNSTABLE_ OR_ POOR (0x3e76): -
The endpoint is abandoning the path because the used interface is observed to be unstable or performance is considered poor. This condition can occur, e.g., due to frequent handover events during high-speed mobility or due to a weak wireless signal.¶
- NO_
CID_ AVAILABLE_ FOR_ PATH (0x3e77): -
The endpoint is abandoning the path due to the lack of a connection ID for this path. This might occur when the peer initiates a new path but has not provided a corresponding connection ID for the path ID (or the packet containing the connection IDs has not arrived yet).¶
4.3. PATH_STATUS_AVAILABLE and PATH_STATUS_BACKUP Frames
Skip to TOC
PATH_
PATH_
PATH_STATUS_AVAILABLE Frame {
Type (i) = 0x3e77,
Path Identifier (i),
Path Status Sequence Number (i),
}PATH_
PATH_
PATH_STATUS_BACKUP Frame {
Type (i) = 0x3e76
Path Identifier (i),
Path Status Sequence Number (i),
}Both PATH_
- Path Identifier:
-
The path ID that the status update corresponds to. All path IDs below the maximum path ID limit can be indicated, even if the path is not in active use yet.¶
- Path Status Sequence Number:
-
A variable-
length integer specifying the per-path sequence number assigned for this frame.¶
The sequence number space is common to the two frame types,
and monotonically increasing values MUST be used when sending PATH_
Frames might be received out of order. A peer MUST ignore an incoming
PATH_
The requirement of monotonically increasing sequence numbers
is per path. Receivers could very well receive the
same sequence number for PATH_
PATH_
A PATH_
4.4. PATH_NEW_CONNECTION_ID Frame
Skip to TOC
The PATH_
PATH_
PATH_NEW_CONNECTION_ID Frame {
Type (i) = 0x3e78,
Path Identifier (i),
Sequence Number (i),
Retire Prior To (i),
Length (8),
Connection ID (8..160),
Stateless Reset Token (128),
}Compared to the NEW_
- Path Identifier:
-
The path ID associated with the connection ID. This means the provided connection ID can only be used on the corresponding path.¶
Note that, other than for the NEW_
The Retire Prior To field indicates which connection IDs should be retired among those that share the path ID in the Path Identifier field. Connection IDs associated with different path IDs are not affected.¶
Note that the NEW_
The last paragraph of Section 5.1.2 of [QUIC-TRANSPORT] specifies how to
verify the Retire Prior To field of an incoming NEW_
4.5. PATH_RETIRE_CONNECTION_ID Frame
Skip to TOC
The PATH_
PATH_
PATH_RETIRE_CONNECTION_ID Frame {
Type (i) = 0x3e79,
Path Identifier (i),
Sequence Number (i),
}Compared to the RETIRE_
- Path Identifier:
-
The path ID associated with the connection ID to retire.¶
Note that the RETIRE_
As the PATH_
The processing of an incoming RETIRE_
4.6. MAX_PATH_ID Frame
Skip to TOC
A MAX_
MAX_
MAX_PATH_ID Frame {
Type (i) = 0x3e7a,
Maximum Path Identifier (i),
}MAX_
- Maximum Path Identifier:
-
The maximum path ID that the sending endpoint is willing to accept. This value MUST NOT exceed 232-1, which is the maximum allowed value for the path ID due to restrictions on the nonce calculation (see Section 2.4). The Maximum Path Identifier value MUST NOT be lower than the value advertised in the initial_
max_ path_ id transport parameter.¶
Receipt of an invalid Maximum Path Identifier value MUST be treated as a
connection error of type PROTOCOL_
Loss or reordering can cause an endpoint to receive a MAX_
MAX_
4.7. PATHS_BLOCKED and PATH_CIDS_BLOCKED Frames
Skip to TOC
A sender can send a PATHS_
A sender can send a PATH_
Note that PATHS_
PATHS_
PATHS_BLOCKED Frame {
Type (i) = 0x3e7b,
Maximum Path Identifier (i),
}PATHS_
- Maximum Path Identifier:
-
A variable-
length integer indicating the maximum path ID that was allowed at the time the frame was sent. If the received value is lower than the currently allowed maximum value, this frame can be ignored.¶
PATH_
PATH_CIDS_BLOCKED Frame {
Type (i) = 0x3e7c,
Path Identifier (i),
Next Sequence Number (i),
}PATH_
- Path Identifier:
-
Identifier of the path for which unused connection IDs are not available.¶
- Next Sequence Number:
-
The next sequence number that is expected to be issued for a connection ID for this path by the peer.¶
Receipt of a value of Maximum Path Identifier or Path Identifier that is higher than
the local maximum value MUST be treated as a connection error of type PROTOCOL_
Receipt of a value of Next Sequence Number that is higher than
the sequence number of the next expected to be issued connection ID for this path
MUST be treated as a connection error of type PROTOCOL_
PATHS_
5. Implementation Considerations Skip to TOC
This section provides informational guidance for implementors.¶
5.1. Connection ID Changes, Migration, and NAT Rebindings Skip to TOC
With the multipath extension, each path uses a separate packet number space. This is a major difference from [QUIC-TRANSPORT], which only defines three number spaces (Initial, Handshake, and Application data packets).¶
For any given path, connection ID rotation, NAT rebinding, or client-
It is generally preferable to use multipath mechanisms, such as creating a new path and later abandoning the old path rather than doing migration of a single path as specified in [QUIC-TRANSPORT]. This enables a smoother handover and allows a more controlled migration handling at the server side. However, migration of a single path cannot be avoided in case of NAT rebinding or if the server requests migration to a "preferred address" during the handshake.¶
Section 9.3 of [QUIC-TRANSPORT] allows an endpoint to skip validation of a peer address if that address has been seen recently. However, when the multipath extension is used and an endpoint has multiple addresses that could lead to switching between different paths, it should rather maintain multiple open paths instead.¶
Servers observing a 4-tuple change will perform path validation (see Section 9 of [QUIC-TRANSPORT]). If path validation process succeeds, the endpoints reset the path's congestion controller and round-trip time estimator according to Section 9.4 of [QUIC-TRANSPORT].¶
5.2. Using Multiple Paths on the Same 4-tuple Skip to TOC
It is possible to create paths that refer to the same 4-tuple. For example, endpoints might want to create paths that use different Differentiated Service [RFC2475] markings. This could be done in conjunction with scheduling algorithms that match streams to paths so that, for example, data frames for low priority streams are sent over low priority paths. Since these paths use different path IDs, they can be managed independently to suit the needs of the application. (The application would need to manage how client and server use differentiated services on a path. This is not specified in this document.)¶
There might be cases in which paths are created with different 4-tuples but end up using the same 4-tuples as a consequence of path migrations. Consider the following example where all paths use the same source and destination ports:¶
-
Client starts path 1 from address 192.0.2.1 to server address 198.51.100.1.¶
-
Client starts path 2 from address 192.0.2.2 to server address 198.51.100.1.¶
-
Both paths are used for a while.¶
-
Server sends packet from address 198.51.100.1 to client address 192.0.2.1 with Connection ID indicating path ID 2.¶
-
Client receives the packet, recognizes a path migration, and updates the source address of path 2 to 192.0.2.1.¶
(For simplicity, this example uses IPv4 addresses but would work similarly for IPv6 addresses.)¶
Such unintentional use of the same 4-tuple on different paths ought to be rare. When they happen, the two paths would be redundant, and the endpoint could want to close one of them.¶
5.3. Congestion Control Skip to TOC
When the QUIC multipath extension is used, senders manage per-path congestion status as required in Section 9.4 of [QUIC-TRANSPORT]. However, in [QUIC-TRANSPORT], only one active path is assumed and, as such, the requirement is to reset the congestion control status on path migration. With the multipath extension, multiple paths can be used simultaneously. Therefore, separate congestion control state is maintained for each path. This means a sender is not allowed to send more data on a given path than congestion control for that path indicates.¶
When a Multipath QUIC connection uses two or more paths, there is no guarantee that these paths are fully disjoint. When two (or more paths) share the same bottleneck, using a standard congestion control scheme could result in an unfair distribution of the bandwidth with the multipath connection getting more bandwidth than competing single paths connections. Multipath TCP uses the Linked Increases Algorithm (LIA) congestion control scheme specified in [RFC6356] to solve this problem. This scheme can immediately be adapted to Multipath QUIC. Other coupled congestion control schemes have been proposed for Multipath TCP, such as [OLIA]. Designers of congestion control algorithms specialized for Multipath QUIC are advised to follow BCP 133; see Section 7.10 of [RFC9743].¶
Section 5.1.2 of [QUIC-TRANSPORT] indicates that an endpoint can change the connection ID it uses to another available one at any time during the connection. As such, a change of the Connection ID without any change in the address does not indicate a path change and the endpoint can keep the same congestion control and RTT measurement state.¶
5.4. Computing Path RTT Skip to TOC
PATH_
Congestion control state as defined in [QUIC-RECOVERY] is kept per path ID. However, depending on which path acknowledgments are sent, the actual RTT of a path cannot be calculated or might not be the right value to be used.¶
Instead of using the real RTT of a path, it is recommended to consider the sum of two one-way delays: the delay on the packet sending path and the delay on the return path chosen for the acknowledgments. When different paths have different characteristics, the delays can vary widely. For example, consider a multipath transmission using both a terrestrial path with a latency of 50ms in each direction and a geostationary satellite path with a latency of 300ms in each direction. The sum of the two one-way delays will depend on the combination of paths used for the packet transmission and the acknowledgment transmission, as shown in Table 1.¶
| ACK Path \ Data Path | Terrestrial | Satellite |
|---|---|---|
| Terrestrial | 100ms | 350ms |
| Satellite | 350ms | 600ms |
The computed values reflect both the state of the network path and the
scheduling decisions of the acknowledgment sender. If we
assume that the PATH_
The simplest implementation is to use the delays measured when receiving new packet acknowledgments
to compute smoothed_
Further, congestion control functions that rely on delay estimates need to consider cases where acknowledgments are sent over multiple paths with different delays explicitly.¶
5.5. Packet Scheduling Skip to TOC
The transmission of packets containing data is limited by the arrival of data from the application and by congestion control. Generally, QUIC packets that increase the number of bytes in flight can only be sent when the congestion window for the selected path allows it.¶
Most frames, including control frames (PATH_
This implies that an endpoint might send and receive PATH_
Since packets that only carry PATH_
5.6. Retransmissions Skip to TOC
Simultaneous use of multiple paths enables different retransmission strategies to cope with losses such as a) retransmitting lost frames over the same path, b) retransmitting lost frames on a different or dedicated path, and c) duplicate lost frames on several paths (not recommended for general purpose use due to the network overhead). While this document does not preclude a specific strategy, more detailed specification is out of scope.¶
As noted in Section 2.2 of [QUIC-TRANSPORT], STREAM frame boundaries are not expected to be preserved when data is retransmitted. Especially when STREAM frames have to be retransmitted over a different path with a smaller MTU limit, smaller STREAM frames might need to be sent instead.¶
5.7. PTO Expiration Skip to TOC
An implementation should follow the mechanism specified in [QUIC-RECOVERY] for detecting packet loss on each individual path. A special case happens when the PTO timer expires. According to [QUIC-RECOVERY], no packet will be declared lost until either the packet sender receives a new acknowledgment for this path or the path itself is finally declared broken. This cautious process minimizes the risk of spurious retransmissions, but it might cause significant delivery delay for the frames contained in these "lost packets".¶
Endpoints could take advantage of the multipath extension and retransmit the content of the delayed packets on other available paths if the congestion control window on these paths allows.¶
5.8. Paths Having Different PMTU Sizes Skip to TOC
An implementation should take care to handle different Path MTU (PMTU) sizes across multiple paths. As specified in Section 14.3 of [QUIC-TRANSPORT], the Datagram Packetization Layer PMTU Discovery (DPLPMTUD) Maximum Packet Size (MPS) is maintained for each combination of local and remote IP addresses. Note that with the multipath extension, multiple paths could use the same 4-tuple but might have different MPS due to other factors (see Section 5.2).¶
One simple option, if the PMTUs are similar, is to apply the minimum PMTU of all paths to each path, which could also help to simplify retransmission processing.¶
5.9. Idle Timeout and Keep-Alives Skip to TOC
[QUIC-TRANSPORT] defines an idle timeout for closing the connection, which applies in case of multipath usage if no packet is received on any path for the duration of the idle timeout.¶
This document does not specify per-path idle timeouts. An endpoint
can decide to close a path at any time, whether the path is in active
use or not. For example, an endpoint might wait to send
the initial PATH_
If a path is not actively used for a while, it might not be usable anymore, e.g., due to middlebox timeouts. To avoid such path breakage, endpoints can send ack-eliciting packets, such as packets containing PING frames (Section 19.2 of [QUIC-TRANSPORT]) on that path to keep it alive. However, this specification does not recommend sending keep-alives, as it can create unnecessary overhead, especially if there are other actively used paths.¶
Section 5.3 of [QUIC-TRANSPORT] defines an optional keep-alive process. This process can be applied to each path separately depending on application needs. Some applications could decide to not keep any not-actively used path alive or keep only one additional path or multiple paths (e.g., for more redundancy) alive. As discussed in Section 10.1.2 of [QUIC-TRANSPORT], the keep-alive interval needs to incorporate timeouts in middleboxes on the path.¶
If a path was not actively used for a while and no keep-alives have been sent, an endpoint can probe it before switching to active use if there are still other paths that are currently usable.¶
6. IANA Considerations Skip to TOC
This document defines a new transport parameter to enable simultaneous use of multiple paths within one QUIC connection. Further, it specifies new frame types for path management and new error codes when a path is abandoned.¶
IANA has allocated short values with
permanent as the status, IETF as the change controller, and
the QUIC WG as the contact to the respective registries under
<https://
The following parameter defined in Table 2 has been added to the "QUIC Transport Parameters" registry.¶
| Value | Parameter Name | Specification |
|---|---|---|
| 0x3e | initial_ |
Section 2.1 |
The following frame types defined in Table 3 have been added to the "QUIC Frame Types" registry.¶
| Value | Frame Type Name | Specification |
|---|---|---|
| 0x3e-0x3f | PATH_ |
Section 4.1 |
| 0x3e75 | PATH_ |
Section 4.2 |
| 0x3e76 | PATH_ |
Section 4.3 |
| 0x3e77 | PATH_ |
Section 4.3 |
| 0x3e78 | PATH_ |
Section 4.4 |
| 0x3e79 | PATH_ |
Section 4.5 |
| 0x3e7a | MAX_ |
Section 4.6 |
| 0x3e7b | PATHS_ |
Section 4.7 |
| 0x3e7c | PATH_ |
Section 4.7 |
The following transport error codes defined in Table 4 have been added to the "QUIC Transport Error Codes" registry.¶
| Value | Code | Description | Specification |
|---|---|---|---|
| 0x3e | APPLIC |
Path abandoned at the application's request | Section 4.2.1 |
| 0x3e75 | PATH_ |
Path abandoned due to resource limitations in the transport | Section 4.2.1 |
| 0x3e76 | PATH_ |
Path abandoned due to unstable interfaces or poor performance | Section 4.2.1 |
| 0x3e77 | NO_ |
Path abandoned due to no available connection IDs for the path | Section 4.2.1 |
7. Security Considerations Skip to TOC
The multipath extension retains all security properties of [QUIC-TRANSPORT] and [QUIC-TLS] but requires some additional consideration regarding:¶
-
potential additional resource usage for per-path connection IDs and multiple concurrent path contexts,¶
-
a potentially increased amplification risk for denial-
of- service attacks if multiple paths are used simultaneously, and¶ -
changes to the nonce calculation due to the use of multiple packet number spaces.¶
7.1. Memory Allocation for Per-Path Resources Skip to TOC
The maximum path ID limit in initial_
To avoid unnecessary resource usage that could be exploited in a resource exhaustion attack, endpoints SHOULD allocate additional path resources, such as, e.g., for packet number handling, only after path validation has successfully completed.¶
7.2. Denial of Service with Multiple Paths Skip to TOC
Path validation as specified in Section 8.2 of [QUIC-TRANSPORT]
for migration is used
unchanged for path initiation in this extension.
Further, the multipath extension allows for the creation of multiple paths, which means
that in addition to the security considerations
on source address spoofing outlined in Section 21.5.4 of [QUIC-TRANSPORT],
there is a risk of amplified denial-
[QUIC-TRANSPORT] only allows the use of one path and the number of concurrent path validation attempts is limited by the number of issued connection IDs. This extension allows for multiple open paths that could in theory be migrated all at the same time, still limited by the number of issued connection IDs for each path. Further, multiple paths could be initialized simultaneously, limited by the maximum number of allowed paths.¶
Therefore, endpoints are advised to keep the maximum number of paths small and might consider additional measures to limit the number of concurrent path validation processes, e.g., by pacing them out or limiting the number of path initiation attempts over a certain time period.¶
However, the use of multipath does not change the
anti-
7.3. Cryptographic Handshake and AEAD Nonce Skip to TOC
The multipath extension as specified in this document is only enabled after a successful handshake when both endpoints indicate support for this extension. All new frames defined in this extension are only used in 1-RTT packets.¶
As the handshake is not changed by this extension, the transport security mechanisms as specified in [QUIC-TLS], such as encryption key exchange and peer authentication, remain unchanged. As such, the security considerations in [QUIC-TLS] apply unaltered.¶
The limits as discussed on Appendix B of [QUIC-TLS] apply to the total number of packets sent on all paths, not each path separately.¶
This specification changes the AEAD nonce calculation by including the path ID as part of the calculation (see Section 2.4). To ensure unique nonces, path IDs are limited to 32 bits and cannot be reused for another path of the same connection.¶
8. References Skip to TOC
8.1. Normative References Skip to TOC
- [QUIC-
RECOVERY] -
Iyengar, J., Ed. and I. Swett, Ed., "QUIC Loss Detection and Congestion Control", RFC 9002, DOI 10.17487
/RFC9002 , , <https://www >..rfc- editor .org /info /rfc9002 - [QUIC-TLS]
-
Thomson, M., Ed. and S. Turner, Ed., "Using TLS to Secure QUIC", RFC 9001, DOI 10.17487
/RFC9001 , , <https://www >..rfc- editor .org /info /rfc9001 - [QUIC-
TRANSPORT] -
Iyengar, J., Ed. and M. Thomson, Ed., "QUIC: A UDP-Based Multiplexed and Secure Transport", RFC 9000, DOI 10.17487
/RFC9000 , , <https://www >..rfc- editor .org /info /rfc9000 - [RFC2119]
-
Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487
/RFC2119 , , <https://www >..rfc- editor .org /info /rfc2119 - [RFC8174]
-
Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487
/RFC8174 , , <https://www >..rfc- editor .org /info /rfc8174
8.2. Informative References Skip to TOC
- [OLIA]
-
Khalili, R., Gast, N., Popovic, M., Upadhyay, U., and J. Le Boudec, "MPTCP is not pareto-
optimal: performance issues and a possible solution" , CoNEXT '12: Proceedings of the 8th international conference on Emerging networking experiments and technologies, pp. 1-12, DOI 10.1145/2413176 , , <https://.2413178 doi >..org /10 .1145 /2413176 .2413178 - [RFC2475]
-
Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., and W. Weiss, "An Architecture for Differentiated Services", RFC 2475, DOI 10.17487
/RFC2475 , , <https://www >..rfc- editor .org /info /rfc2475 - [RFC6356]
-
Raiciu, C., Handley, M., and D. Wischik, "Coupled Congestion Control for Multipath Transport Protocols", RFC 6356, DOI 10.17487
/RFC6356 , , <https://www >..rfc- editor .org /info /rfc6356 - [RFC9312]
-
Kühlewind, M. and B. Trammell, "Manageability of the QUIC Transport Protocol", RFC 9312, DOI 10.17487
/RFC9312 , , <https://www >..rfc- editor .org /info /rfc9312 - [RFC9743]
-
Duke, M., Ed. and G. Fairhurst, Ed., "Specifying New Congestion Control Algorithms", BCP 133, RFC 9743, DOI 10.17487
/RFC9743 , , <https://www >..rfc- editor .org /info /rfc9743
Acknowledgments Skip to TOC
This document is a collaboration of authors that combines work from three proposals. Further authors of one of the original proposals are Qing An and Zhenyu Li.¶
Thanks to Marten Seemann, Kazuho Oku, Martin Thomson, Magnus Westerlund, Mike Bishop, Lucas Pardue, Michael Eriksson, Yu Zhu, Gorry Fairhurst, Tilmann Zäschke, and Tommy Pauly for their thorough reviews and valuable contributions.¶
Thanks for IETF Last Call reviews to Carsten Bormann (artart), Antoine Fressancourt (intdir), Adrian Farrel (opsdir), and Meral Shirazipour (genart).¶