Internet DRAFT - draft-miao-rtgwg-hpccplus
draft-miao-rtgwg-hpccplus
Network Working Group R. Miao
Internet-Draft H. Liu
Intended status: Experimental Alibaba Group
Expires: September 8, 2022 R. Pan
J. Lee
C. Kim
Intel Corporation
B. Gafni
Y. Shpigelman
Mellanox Technologies, Inc.
J. Tantsura
Microsoft Corporation
March 7, 2021
HPCC++: Enhanced High Precision Congestion Control
draft-miao-rtgwg-hpccplus-00
Abstract
Congestion control (CC) is the key to achieving ultra-low latency,
high bandwidth and network stability in high-speed networks.
However, the existing high-speed CC schemes have inherent limitations
for reaching these goals.
In this document, we describe HPCC++ (High Precision Congestion
Control), a new high-speed CC mechanism which achieves the three
goals simultaneously. HPCC++ leverages inband telemetry to obtain
precise link load information and controls traffic precisely. By
addressing challenges such as delayed signaling during congestion and
overreaction to the congestion signaling using inband and granular
telemetry, HPCC++ can quickly converge to utilize all the available
bandwidth while avoiding congestion, and can maintain near-zero in-
network queues for ultra-low latency. HPCC++ is also fair and easy
to deploy in hardware, implementable with commodity NICs and
switches.
Status of This Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at https://datatracker.ietf.org/drafts/current/.
Miao, et al. Expires September 8, 2022 [Page 1]
Internet-Draft HPCC++ March 2021
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on September 8, 2022.
Copyright Notice
Copyright (c) 2021 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://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 . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. System Overview . . . . . . . . . . . . . . . . . . . . . . . 4
4. HPCC++ Algorithm . . . . . . . . . . . . . . . . . . . . . . 5
4.1. Notations . . . . . . . . . . . . . . . . . . . . . . . . 6
4.2. Design Functions and Procedures . . . . . . . . . . . . . 6
5. Configuration Parameters . . . . . . . . . . . . . . . . . . 8
6. Design enhancement and implementation . . . . . . . . . . . . 9
6.1. Inband telemetry padding at the network switches . . . . 9
6.1.1. Inband telemetry on IFA2.0 . . . . . . . . . . . . . 9
6.1.2. Inband telemetry on IOAM . . . . . . . . . . . . . . 9
6.1.3. Inband telemetry on P4 . . . . . . . . . . . . . . . 9
6.2. Congestion Notification . . . . . . . . . . . . . . . . . 10
6.2.1. Forward direction Congestion detection . . . . . . . 11
6.2.2. Reverse direction . . . . . . . . . . . . . . . . . . 11
6.3. Congestion control at NICs . . . . . . . . . . . . . . . 12
6.3.1. Sender-based HPCC . . . . . . . . . . . . . . . . . . 12
6.3.2. Receiver-based HPCC . . . . . . . . . . . . . . . . . 13
7. Reference Implementation . . . . . . . . . . . . . . . . . . 14
7.1. Implementation on RDMA RoCEv2 . . . . . . . . . . . . . . 14
7.2. Implementation on TCP . . . . . . . . . . . . . . . . . . 15
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15
9. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 15
9.1. Internet Deployment . . . . . . . . . . . . . . . . . . . 15
Miao, et al. Expires September 8, 2022 [Page 2]
Internet-Draft HPCC++ March 2021
9.2. Switch-assisted congestion control . . . . . . . . . . . 16
9.3. Work with QoS queuing . . . . . . . . . . . . . . . . . . 16
9.4. Path migration . . . . . . . . . . . . . . . . . . . . . 17
10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 17
11. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 17
12. References . . . . . . . . . . . . . . . . . . . . . . . . . 17
12.1. Normative References . . . . . . . . . . . . . . . . . . 17
12.2. Informative References . . . . . . . . . . . . . . . . . 18
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19
1. Introduction
The link speed in data center networks has grown from 1Gbps to
100Gbps in the past decade, and this growth is continuing. Ultralow
latency and high bandwidth, which are demanded by more and more
applications, are two critical requirements in today's and future
high-speed networks.
Given that traditional software-based network stacks in hosts can no
longer sustain the critical latency and bandwidth requirements as
described in [Zhu-SIGCOMM2015], offloading network stacks into
hardware is an inevitable direction in high-speed networks. As an
example, large-scale networks with RDMA (remote direct memory access)
often uses hardware-offloading solutions. In some cases, the RDMA
networks still face fundamental challenges to reconcile low latency,
high bandwidth utilization, and high stability.
This document describes a new congestion control mechanism, HPCC++
(Enhanced High Precision Congestion Control), for large-scale, high-
speed networks. The key idea behind HPCC++ is to leverage the
precise link load information from signaled through inband telemetry
to compute accurate flow rate updates. Unlike existing approaches
that often require a large number of iterations to find the proper
flow rates, HPCC++ requires only one rate update step in most cases.
Using precise information from inband telemetry enables HPCC++ to
address the limitations in current congestion control schemes.
First, HPCC++ senders can quickly ramp up flow rates for high
utilization and ramp down flow rates for congestion avoidance.
Second, HPCC++ senders can quickly adjust the flow rates to keep each
link's output rate slightly lower than the link's capacity,
preventing queues from being built-up as well as preserving high link
utilization. Finally, since sending rates are computed precisely
based on direct measurements at switches, HPCC++ requires merely
three independent parameters that are used to tune fairness and
efficiency.
The base form of HPCC++ is the original HPCC algorithm and its full
description can be found in [SIGCOMM-HPCC]. While the original
Miao, et al. Expires September 8, 2022 [Page 3]
Internet-Draft HPCC++ March 2021
design lays the foundation for inband telemetry based precision
congestion control, HPCC++ is an enhanced version which takes into
account system constraints and aims to reduce the design overhead and
further improves the performance. Section 6 describes these detailed
proposed design enhancements and guidelines.
This document describes the architecture changes in switches and end-
hosts to support the needed tranmission of inband telemetry and its
consumption, that imporves the efficiency in handling network
congestion.
2. Terminology
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.
3. System Overview
Figure 1 shows the end-to-end system that HPCC++ operates in. During
the traverse of the packet from the sender to the receiver, each
switch along the path inserts inband telemetry that reports the
current state of the packet's egress port, including timestamp (ts),
queue length (qLen), transmitted bytes (txBytes), and the link
bandwidth capacity (B), together with switch_ID and port_ID. When
the receiver gets the packet, it may copy all the inband telemetry
recorded from the network to the ACK message it sends back to the
sender, and then the sender decides how to adjust its flow rate each
time it receives an ACK with network load information.
Alternatively, the receiver may calculate the flow rate based on the
inband telemetry information and feedback the calculated rate back to
the sender. The notification packets would include delayed ack
information as well.
Note that there also exist network nodes along the reverse
(potentially uncongested) path that the feedback reports traverse.
Those network nodes are not shown in the figure for sake of brevity.
Miao, et al. Expires September 8, 2022 [Page 4]
Internet-Draft HPCC++ March 2021
+---------+ pkt +-------+ pkt+tlm +-------+ pkt+tlm +----------+
| Data |-------->| |-------->| |-------->| Data |
| Sender |=========|Switch1|=========|Switch2|=========| Receiver |
+---------+ Link-0 +-------+ Link-1 +-------+ Link-2 +----------+
/|\ |
| |
+---------------------------------------------------------+
Notification Packets/ACKs
Figure 1: System Overview (tlm=inband telemtry)
o Data sender: responsible for controlling inflight bytes. HPCC++
is a window-based congestion control scheme that controls the
number of inflight bytes. The inflight bytes mean the amount of
data that have been sent, but not acknowledged by the sender yet.
Controlling inflight bytes has an important advantage compared to
controlling rates. In the absence of congestion, the inflight
bytes and rate are interchangeable with equation inflight = rate *
T where T is the base propagation RTT. The rate can be calculated
locally or obtained from the notification packet. The sender may
further use the data pacing mechanism, potentially implemented in
hardware, to limit the rate accordingly.
o Network nodes: responsible of inserting the inband telemetry
information to the data packet. The inband telemetry information
reports the current load of the packet's egress port, including
timestamp (ts), queue length (qLen), transmitted bytes (txBytes),
and link bandwidth capacity (B). Besides, the inband telemetry
contains switch_ID and port_ID to identify a link.
o Data receiver: responsible for either reflecting back the inband
telemetry information in the data packet or calculating the proper
flow rate based on network congestion information in inband
telemetry and sending notification packets back to the sender.
4. HPCC++ Algorithm
HPCC++ is a window-based congestion control algorithm. The key
design choice of HPCC++ is to rely on network nodes to provide fine-
grained load information, such as queue size and accumulated tx/rx
traffic to compute precise flow rates. This has two major benefits:
(i) HPCC++ can quickly converge to proper flow rates to highly
utilize bandwidth while avoiding congestion; and (ii) HPCC++ can
consistently maintain a close-to-zero queue for low latency.
This section introduces the list of notations and describes the core
congestion control algorithm.
Miao, et al. Expires September 8, 2022 [Page 5]
Internet-Draft HPCC++ March 2021
4.1. Notations
This section summarizes the list of variables and parameters used in
the HPCC++ algorithm. Figure 3 also includes the default values for
choosing the algorithm parameters either to represent a typical
setting in practical applications or based on theoretical and
simulation studies.
+--------------+-------------------------------------------------+
| Notation | Variable Name |
+--------------+-------------------------------------------------+
| W_i | Window for flow i |
| Wc_i | Reference window for flow i |
| B_j | Bandwidth for Link j |
| I_j | Estimated inflight bytes for Link j |
| U_j | Normalized inflight bytes for Link j |
| qlen | Telemetry info: link j queue length |
| txRate | Telemetry info: link j output rate |
| ts | Telemetry info: timestamp |
| txBytes | Telemetry info: link j total transmitted bytes |
| | associated with timestamp ts |
+--------------+-------------------------------------------------+
Figure 2: List of variables.
+--------------+----------------------------------+----------------+
| Notation | Parameter Name | Default Value |
+--------------+----------------------------------+----------------+
| T | Known baseline RTT | 5us |
| eta | Target link utilization | 95% |
| maxStage | Maximum stages for additive | |
| | increases | 5 |
| N | Maximum number of flows | ... |
| W_ai | Additive increase amount | ... |
+--------------+----------------------------------+----------------+
Figure 3: List of algorithm parameters and their default values.
4.2. Design Functions and Procedures
The HPCC++ algorithm can be outlined as below:
Miao, et al. Expires September 8, 2022 [Page 6]
Internet-Draft HPCC++ March 2021
1: Function MeasureInflight(ack)
2: u = 0;
3: for each link i on the path do
4: ack.L[i].txBytes-L[i].txBytes
txRate = ----------------------------- ;
ack.L[i].ts-L[i].ts
5: min(ack.L[i].qlen,L[i].qlen) txRate
u' = ----------------------------- + ---------- ;
ack.L[i].B*T ack.L[i].B
6: if u' > u then
7: u = u'; tau = ack.L[i].ts - L[i].ts;
8: tau = min(tau, T);
9: U = (1 - tau/T)*U + tau/T*u;
10: return U;
11: Function ComputeWind(U, updateWc)
12: if U >= eta or incStage >= maxStagee then
13: Wc
W = ----- + W_ai;
U/eta
14: if updateWc then
15: incStagee = 0; Wc = W ;
16: else
17: W = Wc + W_ai ;
18: if updateWc then
19: incStage++; Wc = W ;
20: return W
21: Procedure NewAck(ack)
22: if ack.seq > lastUpdateSeq then
23: W = ComputeWind(MeasureInflight(ack), True);
24: lastUpdateSeq = snd_nxt;
25: else
26: W = ComputeWind(MeasureInflight(ack), False);
27: R = W/T; L = ack.L;
The above illustrates the overall process of CC at the sender side
for a single flow. Each newly received ACK message triggers the
procedure NewACK at Line 21. At Line 22, the variable lastUpdateSeq
is used to remember the first packet sent with a new W c , and the
sequence number in the incoming ACK should be larger than
lastUpdateSeq to trigger a new sync betweenW c andW (Line 14-15 and
18-19). The sender also remembers the pacing rate and current inband
telemetry information at Line 27. The sender computes a new window
size W at Line 23 or Line 26, depending on whether to update W c ,
with function MeasureInflight and ComputeWind. Function
MeasureInflight estimates normalized inflight bytes with Eqn (2) at
Line 5. First, it computes txRate of each link from the current and
Miao, et al. Expires September 8, 2022 [Page 7]
Internet-Draft HPCC++ March 2021
last accumulated transferred bytes txBytes and timestamp ts (Line 4).
It also uses the minimum of the current and last qlen to filter out
noises in qlen (Line 5). The loop from Line 3 to 7 selects maxi(Ui)
in Eqn. (3). Instead of directly using maxi(Ui), we use an EWMA
(Exponentially Weighted Moving Average) to filter the noises from
timer inaccuracy and transient queues. (Line 9). Function
ComputeWind combines multiplicative increase/ decrease (MI/MD) and
additive increase (AI) to balance the reaction speed and fairness.
If a sender finds it should increase the window size, it first tries
AI for maxStage times with the stepWAI (Line 17). If it still finds
room to increase after maxStage times of AI or the normalized
inflight bytes is above, it calls Eqn (4) once to quickly ramp up or
ramp down the window size (Line 12-13).
5. Configuration Parameters
HPCC++ has three easy-to-set parameters: eta, maxStagee, and W_ai.
eta controls a simple tradeoff between utilization and transient
queue length (due to the temporary collision of packets caused by
their random arrivals, so we set it to 95% by default, which only
loses 5% bandwidth but achieves almost zero queue. maxStage controls
a simple tradeoff between steady state stability and the speed to
reclaim free bandwidth. We find maxStage = 5 is conservatively large
for stability, while the speed of reclaiming free bandwidth is still
much faster than traditional additive increase, especially in high
bandwidth networks. W_ai controls the tradeoff between the maximum
number of concurrent flows on a link that can sustain near-zero
queues and the speed of convergence to fairness. Note that none of
the three parameters are reliability-critical.
HPCC++'s design brings advantages to short-lived flows, by allowing
flows starting at line-rate and the separation of utilization
convergence and fairness convergence. HPCC++ achieves fast
utilization convergence to mitigate congestion in almost one round-
trip time, while allows flows to gradually converge to fairness.
This design feature of HPCC++ is especially helpful for the workload
of datacenter applications, where flows are usually short and
latency-sensitive. Normally we set a very small W_ai to support a
large number of concurrent flows on a link, because slower fairness
is not critical. A rule of thumb is to set W_ai = W_init*(1-eta) / N
where N is the expected or receiver reported maximum number of
concurrent flows on a link. The intuition is that the total additive
increase every round (N*W_ai ) should not exceed the bandwidth
headroom, and thus no queue forms. Even if the actual number of
concurrent flows on a link exceeds N, the CC is still stable and
achieves full utilization, but just cannot maintain zero queues.
Miao, et al. Expires September 8, 2022 [Page 8]
Internet-Draft HPCC++ March 2021
6. Design enhancement and implementation
There are three compoments HPCC++ needs to implement: telementry
padding, congestion notification, and rate update.
6.1. Inband telemetry padding at the network switches
HPCC++ only relies on packets to share information across senders,
receivers, and switches. The switch should capture inband telemetry
information that includes link load (txBytes, qlen, ts) and link spec
(switch_ID, port_ID, B) at the egress port. Note, each switch should
record all those information at the single snapshot to achieve a
precise link load estimate. Inside a data center, the path length is
often no more than 5 hops. The overhead of the inband telemetry
padding for HPCC++ is considered to be low.
As long the above algorithm is met, HPCC++ is open to a variety of
inband telemetry format standards, which are orthogonal to the HPCC++
algorithm. Although this document does not mandate a particular
inband telemetry header format or encapsulation, we provide concrete
implementation specifications using strandard inband telemetry
protocols, including IFA [I-D.ietf-kumar-ippm-ifa], IETF IOAM
[I-D.ietf-ippm-ioam-data], and P4.org INT [P4-INT]. In fact, the
emerging inband telemetry protocols inform the evolution for a
broader range of protocols and network functions, where this document
leverages the trend to propose the architecture change to support in-
network functions like congestion control with high efficiency.
6.1.1. Inband telemetry on IFA2.0
For more details, please refer to IFA [I-D.ietf-kumar-ippm-ifa]
6.1.2. Inband telemetry on IOAM
Please refer to IETF IOAM [I-D.ietf-ippm-ioam-data]
6.1.3. Inband telemetry on P4
Miao, et al. Expires September 8, 2022 [Page 9]
Internet-Draft HPCC++ March 2021
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| nHop | pathID | Padding |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Speed | Timestamp |txBytes|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| txBytes(lower) | Queue Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| 2nd Hop |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| 2nd Hop(lower) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Options | Padding |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 4: Example P4.org INT header
Figure 4 shows the packet format of the INT padding after UDP header.
The field nHop is the hop count of the packet's path. The field
pathID is the XOR of all the switch IDs (which are 12 bits) along the
path. The sender sets nHop and pathID to 0. Each switch along the
path adds nHop by 1, and XORs its own switch ID to the pathID. The
sender uses pathID to judge whether the path of the flow has been
changed. If so, it throws away the existing status records of the
flow and builds up new records. Each switch has an 8-byte field to
record the status of the egress port of the packet when the packet is
emitted. B is a enum type which indicates the speed type of the
port(e.g. 40Gbps, 100Gbps, etc.). Timestamp (24 bits) is when the
packet is emitted from its egress port, txBytes (20 bits) is the
accumulative total bytes sent from the egress port, and Queue length
(16 bits) is the current queue length of the egress port.
6.2. Congestion Notification
HPCC++ uses congestion notification to fetch network congestion
information from switches for proper rate updates at end-hosts.
Although the basic algorithm described in Section 4 is to add inband
telemetry information into every data packet for optimal performance,
HPCC++ supports flexible implementation choices to work seamly with
transport protocol stacks. We consider congestion nofication choices
in both forward and reverse directions of the traffic.
Miao, et al. Expires September 8, 2022 [Page 10]
Internet-Draft HPCC++ March 2021
6.2.1. Forward direction Congestion detection
Forward direction is the traffic direction of data packets that
experience bandwidth contention and possible network congestion. The
function of congestion notification in forward direction is to fetch
inband telemetry from switches. HPCC++ defines two approaches of
doing this.
1. Inband with data packet.
This is basic algorithm setting described in Section 4, where the
end-host inserts inband telemetry header into data packets. Switches
along the path detect the inband telemetry header and correspondingly
add inband telemetry information into data packet to react to
congestion as soon as the very first packet observing the network
congestion. This is especially helpful to reduce the risk of severe
congestion in incast scenarios at the first round-trip time. In
addition, original HPCC's algorithm introduction of Wc is for the
purpose of solving the over-reaction issue from using this per-packet
response. Different with in Section 4, end-host can choice uses
every data packet or only a subset of data packets to reduce the
overhead. To insert telemetry header, differet telemetry protocols
have specific settings for IFA, IETF IOAM, and P4.org INT as
following.
2. Probe packet.
Switches touching every data packet for inband telemetry inserting
may lead to security or performance concerns, HPCC++ supports the
``out-of-band'' approach that uses special-generated probe packets at
end-hosts to fetch inband telemetry from switches. Thereby, the
probe packets should take the same routing path and QoS queueing with
the data packets. End-hosts can generate probe packets less
frequently and we recommend once per round trip time. In addition,
the end-host issues probe packets only when it has data packet in the
flight.
6.2.2. Reverse direction
Reverse direction is the receiver conveying inband telemetry back to
traffic sender for rate updates. Similar to forward direction, there
are also inband and out-of-band approaches.
1. Inband with ACK packet.
HPCC++ supports to use the ACK packet in transport protocols to
convey the inband telemetry. TCP generates ACK packet once per every
Miao, et al. Expires September 8, 2022 [Page 11]
Internet-Draft HPCC++ March 2021
data packet or per a few data packets. With ACK packet, the receive
sends accumulated inband telemetry back to sender for rate updates.
2. Notification packet.
Using ACK packet for inband telemetry notification requires transport
stack modification and sometimes leads to delay in notification when
certain delayed acknowledged mechanism is used. Hence, HPCC++ allows
the receiver to use special-generated notification packets to deliver
inband telemetry. The nofication packet is generated per each probe
packet or data packet with inband telemetry.
6.3. Congestion control at NICs
6.3.1. Sender-based HPCC
Figure 5 shows HPCC++ implementation on a NIC. The NIC provides an
HPCC++ module that resides on the data path of the NIC, HPCC++
modules realize both sender and receiver roles.
+------------------------------------------------------------------+
| +---------+ window update +-----------+ PktSend +-----------+ |
| | |-------------->| Scheduler |-------> |Tx pipeline|---+->
| | | rate update +-----------+ +-----------+ |
| | HPCC++ | ^ |
| | | inband telemetry| |
| | module | | |
| | | +-----+-----+ |
| | |<----------------------------------- |Rx pipeline| <-+--
| +---------+ telemetry response event +-----------+ |
+------------------------------------------------------------------+
Figure 5: Overview of NIC Implementation
1. Sender side flow
The HPCC++ module running the HPCC CC algorithm in the sender side
for every flow in the NIC. Flow can be defined by some transport
parameters including 5-tuples, destination QP (queue pair), etc. It
receives inband telemetry response events per flow which are
generated from the RX pipeline, adjusts the sending window and rate,
and update the scheduler on the rate and window of the flow.
The scheduler contains a pacing mechanism that determine the flow
rate by the value it got from the algorithm. It also maintains the
current sending window size for active flows. If the pacing
Miao, et al. Expires September 8, 2022 [Page 12]
Internet-Draft HPCC++ March 2021
mechanism and the flow's sending window permits, the scheduler
invokes for the flow a PktSend command to TX pipeline.
The TX pipeline implements packet processing. Once it receives the
PktSend event with flow ID from the scheduler, it generates the
corresponding packet and delivers to the Network. If a sent packet
should collect telemetry on its way the TX pipeline may add
indications/headers that triggers the network elements to add
telemetry data according to the inband telemetry protocol in use.
The telemetry can be collected by the data packet or by dedicated
prob packets generated in the TX pipeline.
The RX pipe parses the incoming packets from the network and
identifies whether telemetry is embedded in the parsed packet. On
receiving a telemetry response packet, the RX pipeline extracts the
network status from the packet and passes it to the HPCC++ module for
processing. A telemetry response packet can be an ACK containing
inband telemetry, or a dedicated telemetry response prob packet.
2. Receiver side flow
On receiving a packet containing inband telemetry, the RX pipeline
extracts the network status, and the flow parameters from the packet
and passes it to the TX pipeline. The packet can be a data packet
containing inband telemetry, or a dedicated telemetry request prob
packet. The Tx pipeline may process and edit the telemetry data, and
then sends back to the sender the data using either an ACK packet of
the flow or a dedicated telemetry response packet.
6.3.2. Receiver-based HPCC
Note that the window/rate calculation can be implemented at either
the data sender or the data receiver. If the ACK packets already
exist for reliability purpose, the inband telemetry information can
be echoed back to the sender via ACK self-clocking. Not all ACK
packets need to carry the inband telemetry information. To reduce
the Packet Per Second (PPS) overhead, the receiver may examine the
inband telemetry information and adopt the technique of delayed ACKs
that only sends out an ACK for a few of received packets. In order
to reduce PPS even further, one may implement the algorithm at the
receiver and feedback the calculated window in the ACK packet once
every RTT.
The receiver-based algorithm, Rx-HPCC, is based on int.L, which is
the inband telemetry information in the packet header. The receiver
performs the same functions except using int.L instead of ack.L. The
new function NewINT(int.L) is to replace NewACK(int.L)
Miao, et al. Expires September 8, 2022 [Page 13]
Internet-Draft HPCC++ March 2021
28: Procedure NewINT(int.L)
29: if now > (lastUpdateTime + T) then
30: W = ComputeWind(MeasureInflight(int), True);
31: send_ack(W)
32: lastUpdateTime = now;
33: else
34: W = ComputeWind(MeasureInflight(int), False);
Here, since the receiver does not know the starting sequence number
of a burst, it simply records the lastUpdateTime. If time T has
passed since lastUpdateTime, the algorithm would recalcuate Wc as in
Line 30 and send out the ACK packet which would include W
information. Otherwise, it would just update W information locally.
This would reduce the amount of traffic that needs to be feedback to
the data sender.
Note that the receiver can also measure the number of outstanding
flows, N, if the last hop is the congestion point and use this
information to dynamically adjust W_ai to achieve better fairness.
The improvement would allow flows to quickly converge to fairness
without causing large swings under heavy load.
7. Reference Implementation
HPCC++ can be adopted as the CC algorithm by a wide range of
transport protocols such as TCP and UDP, as well as others that may
run on top of them, such as iWARP, RoCE etc. It requires to have the
window limit and congestion feedback through ACK self-clocking, which
naturally conforms to the paradigm of TCP design. With that, HPCC++
introduces a scheme to measure the total inflight bytes for more
precise congestion control. To run in UDP, some modifications need
to be done to enforce the window limit and collect congestion
feedback via probing packets, which is incremental.
7.1. Implementation on RDMA RoCEv2
We describe reference implementation on RDMA RoCEv2. This is an
implementation for ``Sender-based HPCC++'' (see section 6.3.1.) using
dedicated probe packets to collect the telemetry. HPCC++ module in
the sender triggers the sending of ``telemetry request packet'' for a
given flow. The NIC then sends the probe packet. The packet will
have the same IP and UDP headers as the data packets of the given
flow. Such packet is expected to be sent every RTT, see section 6
for more details. On receiving of telemetry request packet, the NIC
extracts the telemetry from all the links along the path from the
sender. HPCC++ module chooses the link with the highest inflight
bytes and sends its telemetry (queue length, timestamp and tx bytes)
back to the receiver on top of dedicated ``telemetry response
Miao, et al. Expires September 8, 2022 [Page 14]
Internet-Draft HPCC++ March 2021
packet''. On receiving of telemetry response packet, the NIC
extracts the telemetry and pass it to the HPCC++ module which using
this info to implement the rate update scheme.
7.2. Implementation on TCP
Taking the benefit of precise congestion control for TCP is a natural
next step. Since TCP segmentation at TX side (e.g., TSO) and
coalescing at RX side (e.g., GRO) happen at the NIC HW or low-layer
of TCP/IP stack, carrying per-pkt inband telemetry info between the
TCP congestion control engine and network fabric has to work with the
TSO and GRO. Instead, one way to adopt HPCC++ for TCP is using the
special probe and notification packets to retrieve inband telemetry
information. The sender generates a probe packet when it is actively
sending data. The probe packet has the same 5-tuples (source and
destination addresses, source and destination ports and protocol
number) with the data packets and the inband telemetry header. The
switches along the path identify the probe packet by its inband
telemetry header and insert the inband telemetry. Once received the
probe packet with inband telemetry, the receiver replies with a
response packet piggybacking the inband telemetry to the sender.
Note, both probe and response packets use a special DSCP number so
that it can bypass the TSO and GRO in each side.
8. IANA Considerations
This document makes no request of IANA.
9. Discussion
9.1. Internet Deployment
Although the discussion above mainly focuses on the data center
environment, HPCC++ can be adopted at Internet at large. There are
several security considerations one should be aware of.
There may rise privacy concern when the telemetry information is
conveyed across Autonomous Systems (ASes) and back to end-users. The
link load information captured in telemetry can potentially reveal
the provider's network capacity, route utilization, scheduling
policy, etc. Those usually are considered to be sensitive data of
the network providers. Hence, certain action may take to anonymize
the telemetry data and only convey the relative ratio in rate
adaptation across ASes without revealing the actual network load.
Another consideration is the security of receiving telemetry
information. The rate adaptation mechanism in HPCC++ relies on
feedback from the network. As such, it is vulnerable to attacks
Miao, et al. Expires September 8, 2022 [Page 15]
Internet-Draft HPCC++ March 2021
where feedback messages are hijacked, replaced, or intentionally
injected with misleading information resulting in denial of service,
similar to those that can affect TCP. It is therefore RECOMMENDED
that the notification feedback message is at least integrity checked.
In addition, [I-D.ietf-avtcore-cc-feedback-message] discusses the
potential risk of a receiver providing misleading congestion feedback
information and the mechanisms for mitigating such risks.
9.2. Switch-assisted congestion control
HPCC++ falls in the general category of switch-assisted congestion
control. However, HPCC++ includes a few unique design choices that
are different from other switch-assisted approaches.
o First, HPCC++ implements a primal-mode algorithm that requires
only the ``write-to-packet'' operation from switches, which has
already been supported by telemetry protocols like INT [P4-INT] or
IOAM [I-D.ietf-ippm-ioam-data]. Please note that this is very
different from dual-mode algorithms such as XCP
[Katabi-SIGCOMM2002] and RCP [Dukkipati-RCP], where switches take
an actively role in determining flows' rates.
o Second, HPCC++ achieves a fast utilization convergence by
decoupling it from fairness convergence, which is inspired by XCP.
o Third, HPCC++ enables the switch-guided multiplicative increase
(MI) by defining the ``inflight byte'' to quantify the link load.
The inflight byte tells both the underload and overload of the
link precisely and thus it allows the flow to increase/decrease
the rate multiplicatively and safely. By contrast, traditional
approaches of using the queue length or RTT as the feedback cannot
guide the rate increase and instead have to rely on additive
increase (AI) with heuristics. As the link speed contines to
grow, this becomes increasingly slow in reclaiming the unused
bandwidth. Besides, queue-based feedback mechanisms subject to
latency inflation.
o Last, HPCC++ uses TX rate instead of RX rate used by XCP and RCP.
As detailed in [SIGCOMM-HPCC], we view the TX rate is more precise
because RX rate and queue length are overlapped and thus it causes
oscillation.
9.3. Work with QoS queuing
Under the use of QoS (Quality of service) priority queuing in
switches, the length of flow's own queue cannot tell the actual
queuing time and the exact extent of congestion. Although general
approaches for running congestion control with QoS queuing are out of
Miao, et al. Expires September 8, 2022 [Page 16]
Internet-Draft HPCC++ March 2021
the scope of this document, we provide a few hints for HPCC++ running
friendly with QoS queuing. In this case, HPCC++ can leverage the
packet sojourn time (the egress timestamp minus the ingress
timestamp) instead of the queue length to quantify the packet's
actual queuing delay. In addition, the operators typically use the
Deficit Weighted Round Robin (DWRR) instead of the strict priority
(SP) as their QoS scheduling to prevent traffic starvation. DWRR
provides a minimum bandwdith guarantee for each queue so that HPCC++
can leverage it for precise rate update to avoid congestion.
9.4. Path migration
HPCC++ allows switches and end-hosts to share precise information of
network utilization, which suggests a framework for path selection
and rate control at end-hosts. The framework HPCC++ enabled is to
leverage each switch to report its link load information via inband
telemetry. The end-host fetches inband telemetry along the traffic
routes and makes a timely and accurate decision on path selection and
traffic admission.
10. Acknowledgments
The authors would like to thank RTGWG members for their valuable
review comments and helpful input to this specification.
11. Contributors
The following individuals have contributed to the implementation and
evaluation of the proposed scheme, and therefore have helped to
validate and substantially improve this specification: Pedro Y.
Segura, Roberto P. Cebrian, Robert Southworth and Malek Musleh.
12. References
12.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119,
DOI 10.17487/RFC2119, March 1997,
<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,
May 2017, <https://www.rfc-editor.org/info/rfc8174>.
Miao, et al. Expires September 8, 2022 [Page 17]
Internet-Draft HPCC++ March 2021
12.2. Informative References
[Dukkipati-RCP]
Dukkipati, N., "Rate Control Protocol (RCP): Congestion
control to make flows complete quickly.", Stanford
University , 2008.
[I-D.ietf-avtcore-cc-feedback-message]
Sarker, Z., Perkins, C., Singh, V., and M. A. Ramalho,
"RTP Control Protocol (RTCP) Feedback for Congestion
Control", draft-ietf-avtcore-cc-feedback-message-09 (work
in progress), November 2020.
[I-D.ietf-ippm-ioam-data]
"Data Fields for In-situ OAM", March 2020,
<https://tools.ietf.org/html/draft-ietf-ippm-ioam-data-
09>.
[I-D.ietf-kumar-ippm-ifa]
"Inband Flow Analyzer", February 2019,
<https://tools.ietf.org/html/draft-kumar-ippm-ifa-01>.
[Katabi-SIGCOMM2002]
Katabi, D., Handley, M., and C. Rohrs, "Congestion Control
for High Bandwidth-Delay Product Networks", ACM
SIGCOMM Pittsburgh, Pennsylvania, USA, October 2002.
[P4-INT] "In-band Network Telemetry (INT) Dataplane Specification,
v2.0", February 2020, <https://github.com/p4lang/p4-
applications/blob/master/docs/INT_v2_0.pdf>.
[SIGCOMM-HPCC]
Li, Y., Miao, R., Liu, H., Zhuang, Y., Fei Feng, F., Tang,
L., Cao, Z., Zhang, M., Kelly, F., Alizadeh, M., and M.
Yu, "HPCC: High Precision Congestion Control", ACM
SIGCOMM Beijing, China, August 2019.
[Zhu-SIGCOMM2015]
Zhu, Y., Eran, H., Firestone, D., Guo, C., Lipshteyn, M.,
Liron, Y., Padhye, J., Raindel, S., Yahia, M., and M.
Zhang, "Congestion Control for Large-Scale RDMA
Deployments", ACM SIGCOMM London, United Kingdom, August
2015.
Miao, et al. Expires September 8, 2022 [Page 18]
Internet-Draft HPCC++ March 2021
Authors' Addresses
Rui Miao
Alibaba Group
525 Almanor Ave, 4th Floor
Sunnyvale, CA 94085
USA
Email: miao.rui@alibaba-inc.com
Hongqiang H. Liu
Alibaba Group
108th Ave NE, Suite 800
Bellevue, WA 98004
USA
Email: hongqiang.liu@alibaba-inc.com
Rong Pan
Intel, Corp.
2200 Mission College Blvd.
Santa Clara, CA 95054
USA
Email: rong.pan@intel.com
Jeongkeun Lee
Intel, Corp.
4750 Patrick Henry Dr.
Santa Clara, CA 95054
USA
Email: jk.lee@intel.com
Changhoon Kim
Intel Corporation
4750 Patrick Henry Dr.
Santa Clara, CA 95054
USA
Email: chang.kim@intel.com
Miao, et al. Expires September 8, 2022 [Page 19]
Internet-Draft HPCC++ March 2021
Barak Gafni
Mellanox Technologies, Inc.
350 Oakmead Parkway, Suite 100
Sunnyvale, CA 94085
USA
Email: gbarak@mellanox.com
Yuval Shpigelman
Mellanox Technologies, Inc.
Haim Hazaz 3A
Netanya 4247417
Israel
Email: yuvals@nvidia.com
Jeff Tantsura
Microsoft Corporation
One Microsoft Way
Redmond, Washington 98052-6399
USA
Email: jefftantsura@microsoft.com
Miao, et al. Expires September 8, 2022 [Page 20]
