Microsoft's Datacenter TCP (DCTCP): TCP Congestion Control for Datacenters

Large datacenters necessarily need a large number of network switches to interconnect the servers in the datacenter. Therefore, a datacenter can greatly reduce its capital expenditure by leveraging low cost switches. However, low cost switches tend to have limited queue capacities and thus are more susceptible to packet loss due to congestion. Network traffic in the datacenter is often a mix of short and long flows, where the short flows require low latency and the long flows require high throughput. Datacenters also experience incast bursts, where many endpoints send traffic to a single server at the same time. For example, this is a natural consequence of MapReduce algorithms. The worker nodes complete at approximately the same time, and all reply to the master node concurrently. These factors place some conflicting demands on the queue occupancy of a switch: The queue must be short enough that it does not impose excessive latency on short flows. The queue must be long enough to buffer sufficient data for the long flows to saturate the path bandwidth. The queue must be short enough to absorb incast bursts without excessive packet loss. Standard TCP congestion control relies on segment loss to detect congestion. This does not meet the demands described above. First, the short flows will start to experience unacceptable latencies before packet loss occurs. Second, by the time TCP congestion control kicks in on the sender, most of the incast burst has already been dropped. describes a mechanism for using Explicit Congestion Notification (ECN) from the switch for early detection of congestion, rather than waiting for segment loss to occur. However, this method only detects the presence of congestion, not the extent. In the presence of mild congestion, it reduces the TCP congestion window too aggressively and unnecessarily affects the throughput of long flows. Datacenter TCP (DCTCP) enhances ECN processing to estimate the fraction of bytes that encounter congestion, rather than simply detecting that some congestion has occurred. DCTCP then scales the TCP congestion window based on this estimate. This method achieves high burst tolerance, low latency, and high throughput with shallow-buffered switches. This document describes DCTCP as implemented in Microsoft Windows Server 2012. Since publication of the first versions of this document, the Linux and FreeBSD operating systems have also implemented support for DCTCP in a way that is believed to follow this document.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in .

There are three components involved in the DCTCP algorithm: The switch (or other intermediate device on the network) detects congestion and sets the Congestion Encountered (CE) codepoint in the IP header. The receiver echoes the congestion information back to the sender using the ECN-Echo (ECE) flag in the TCP header. The sender reacts to the congestion indication by reducing the TCP congestion window (cwnd).

The switch indicates congestion to the end nodes by setting the CE codepoint in the IP header as specified in Section 5 of . For example, the switch may be configured with a congestion threshold. When a packet arrives at the switch and its queue length is greater than the congestion threshold, the switch sets the CE codepoint in the packet. For example, Section 3.4 of suggests threshold marking with a threshold K > (RTT * C)/7, where C is the sending rate in packets per second. However, the actual algorithm for marking congestion is an implementation detail of the switch and will generally not be known to the sender and receiver. Therefore, sender and receiver MUST NOT assume that a particular marking algorithm is implemented by the switching fabric.

According to Section 6.1.3 of , the receiver sets the ECE flag if any of the packets being acknowledged had the CE code point set. The receiver then continues to set the ECE flag until it receives a packet with the Congestion Window Reduced (CWR) flag set. However, the DCTCP algorithm requires more detailed congestion information. In particular, the sender must be able to determine the number of sent bytes that encountered congestion. Thus, the scheme described in does not suffice. One possible solution is to ACK every packet and set the ECE flag in the ACK if and only if the CE code point was set in the packet being acknowledged. However, this prevents the use of delayed ACKs, which are an important performance optimization in datacenters. Instead, DCTCP introduces a new Boolean TCP state variable, DCTCP Congestion Encountered (DCTCP.CE), which is initialized to false and stored in the Transmission Control Block (TCB). When sending an ACK, the ECE flag MUST be set if and only if DCTCP.CE is true. When receiving packets, the CE codepoint MUST be processed as follows: If the CE codepoint is set and DCTCP.CE is false, send an ACK for any previously unacknowledged packets and set DCTCP.CE to true. If the CE codepoint is not set and DCTCP.CE is true, send an ACK for any previously unacknowledged packets and set DCTCP.CE to false. Otherwise, ignore the CE codepoint.

The sender estimates the fraction of sent bytes that encountered congestion. The current estimate is stored in a new TCP state variable, DCTCP.Alpha, which is initialized to 1 and MUST be updated as follows: DCTCP.Alpha = DCTCP.Alpha * (1 - g) + g * M where g is the estimation gain, a real number between 0 and 1. The selection of g is left to the implementation. See for further considerations. M is the fraction of sent bytes that encountered congestion during the previous observation window, where the observation window is chosen to be approximately the Round Trip Time (RTT). In particular, an observation window ends when all the sent bytes in flight at the beginning of the window have been acknowledged. In order to update DCTCP.Alpha, the TCP state variables defined in are used, and three additional TCP state variables are introduced: DCTCP.WindowEnd: The TCP sequence number threshold for beginning a new observation window; initialized to SND.UNA. DCTCP.BytesSent: The number of bytes sent during the current observation window; initialized to zero. DCTCP.BytesMarked: The number of bytes sent during the current observation window that encountered congestion; initialized to zero. The congestion estimator on the sender MUST process acceptable ACKs as follows: Compute the bytes acknowledged (TCP SACK options are ignored): BytesAcked = SEG.ACK - SND.UNA Update the bytes sent: DCTCP.BytesSent += BytesAcked If the ECE flag is set, update the bytes marked: DCTCP.BytesMarked += BytesAcked If the sequence number is less than or equal to DCTCP.WindowEnd, then stop processing. Otherwise, the end of the observation window was reached, so proceed to update the congestion estimate as follows: Compute the congestion level for the current observation window: M = DCTCP.BytesMarked / DCTCP.BytesSent Update the congestion estimate: DCTCP.Alpha = DCTCP.Alpha * (1 - g) + g * M Determine the end of the next observation window: DCTCP.WindowEnd = SND.NXT Reset the byte counters: DCTCP.BytesSent = DCTCP.BytesMarked = 0 Rather than always halving the congestion window as described in , when the sender receives an indication of congestion, the sender MUST update cwnd as follows: cwnd = cwnd * (1 - DCTCP.Alpha / 2) Thus, when no sent byte experienced congestion, DCTCP.Alpha equals zero, and cwnd is left unchanged. When all sent bytes experienced congestion, DCTCP.Alpha equals one, and cwnd is reduced by half. Lower levels of congestion will result in correspondingly smaller reductions to cwnd. Just as specified in , TCP should not react to congestion indications more than once every window of data. The setting of the "Congestion Window Reduced" (CWR) bit is also exactly as per .

As noted in , the implementation must choose a suitable estimation gain. provides a theoretical basis for selecting the gain. However, it may be more practical to use experimentation to select a suitable gain for a particular network and workload. The Microsoft implementation of DCTCP in Windows Server 2012 uses a fixed estimation gain of 1/16. The implementation must also decide when to use DCTCP. Datacenter servers may need to communicate with endpoints outside the datacenter, where DCTCP is unsuitable or unsupported. Thus, a global configuration setting to enable DCTCP will generally not suffice. DCTCP may be configured based on the IP address of the remote endpoint. Microsoft Windows Server 2012 also supports automatic selection of DCTCP if the estimated RTT is less than 10 msec, under the assumption that if the RTT is low, then the two endpoints are likely on the same datacenter network.

Since DCTCP relies on congestion marking by the switch, DCTCP can only be deployed in datacenters where the network infrastructure supports ECN. The switches may also support configuration of the congestion threshold used for marking. provides a theoretical basis for selecting the congestion threshold, but as with estimation gain, it may be more practical to rely on experimentation or simply to use the default configuration of the device. DCTCP requires changes on both the sender and the receiver, so both endpoints must support DCTCP. Furthermore, DCTCP provides no mechanism for negotiating its use, so both endpoints must be configured through some out-of-band mechanism to use DCTCP. A variant of DCTCP that can be deployed unilaterally and only requires standard ECN behavior has been described in , but requires additional experimental evaluation.

DCTCP relies on the sender's ability to reconstruct the stream of CE codepoints received by the remote endpoint. To accomplish this, DCTCP avoids using a single ACK packet to acknowledge segments received both with and without the CE codepoint set. However, if an ACK packet is dropped, it's possible that a subsequent ACK will indeed acknowledge a mix of CE and non-CE segments. This will, of course, result in a less accurate congestion estimate. There are some potential mitigations: Even with a degraded congestion estimate, DCTCP may still perform better than . If the estimation gain is small relative to the packet loss rate, the estimate may not be degraded much. If packet losses mostly occur under heavy congestion, most drops will occur during an unbroken string of CE packets, and the estimate will be unaffected. However, the affect of packet drops on DCTCP under real world conditions has not been analyzed. DCTCP provides no mechanism for negotiating its use. Thus, there is additional management and configuration overhead required to ensure that DCTCP is not used with non-DCTCP endpoints. The affect of using DCTCP with a standard ECN endpoint has been analyzed in . Furthermore, it's possible that other implementations may also modify behavior without negotiation, causing further interoperability issues. Much like standard TCP, DCTCP is biased against flows with longer RTTs. A method for improving the fairness of DCTCP has been proposed in , but requires additional experimental evaluation.

DCTCP enhances ECN and thus inherits the security considerations discussed in . The processing changes introduced by DCTCP do not exacerbate these considerations or introduce new ones. In particular, with either algorithm, the network infrastructure or the remote endpoint can falsely report congestion and thus cause the sender to reduce cwnd. However, this is no worse than what can be achieved by simply dropping packets.

This document has no actions for IANA.

The DCTCP algorithm was originally proposed and analyzed in by Mohammad Alizadeh, Albert Greenberg, Dave Maltz, Jitu Padhye, Parveen Patel, Balaji Prabhakar, Sudipta Sengupta, and Murari Sridharan. Lars Eggert has received funding from the European Union's Horizon 2020 research and innovation program 2014-2018 under grant agreement No. 644866. This document reflects only the authors' views and the European Commission is not responsible for any use that may be made of the information it contains.