Internet-Draft | DetNet Bounded Latency | April 2022 |
Finn, et al. | Expires 10 October 2022 | [Page] |
This document presents a timing model for sources, destinations, and DetNet transit nodes. Using the model, it provides a methodology to compute end-to-end latency and backlog bounds for various queuing methods. The methodology can be used by the management and control planes and by resource reservation algorithms to provide bounded latency and zero congestion loss for the DetNet service.¶
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/.¶
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 10 October 2022.¶
Copyright (c) 2022 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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
The ability for IETF Deterministic Networking (DetNet) or IEEE 802.1 Time-Sensitive Networking [IEEE8021TSN] to provide the DetNet services of bounded latency and zero congestion loss depends upon¶
As explained in [RFC8655], DetNet flows are notably characterized by¶
That latency guarantee, in turn, provides the opportunity for the network to supply enough buffer space to guarantee zero congestion loss. It is assumed in this document that the paths of DetNet flows are fixed. Before the transmission of a DetNet flow, it is possible to calculate end-to-end latency bounds and the amount of buffer space required at each hop to ensure zero congestion loss; this can be used by the applications identified in [RFC8578].¶
This document presents a timing model for sources, destinations, and the DetNet transit nodes; using this model, it provides a methodology to compute end-to-end latency and backlog bounds for various queuing mechanisms that can be used by the management and control planes to provide DetNet qualities of service. The methodology used in this document account for the possibility of packet reordering within a DetNet node. The bounds on the amount of packet reordering is out of the scope of this document and can be found in [PacketReorderingBounds]. Moreover, this document references specific queuing mechanisms, mentioned in [RFC8655], as proofs of concept that can be used to control packet transmission at each output port and achieve the DetNet quality of service.¶
Using the model presented in this document, it is possible for an implementer, user, or standards development organization to select a set of queuing mechanisms for each device in a DetNet network, and to select a resource reservation algorithm for that network, so that those elements can work together to provide the DetNet service. Section 7 provides an example application of the timing model introduced in this document on a DetNet IP network with a combination of different queuing mechanisms.¶
This document does not specify any resource reservation protocol or control plane function. It does not describe all of the requirements for that protocol or control plane function. It does describe requirements for such resource reservation methods, and for queuing mechanisms that, if met, will enable them to work together.¶
This document uses the terms defined in [RFC8655]. Moreover, the following terms are used in this document:¶
This document assumes that the following paradigm is used to admit DetNet flows:¶
This paradigm can be implemented using peer-to-peer protocols or using a central controller. In some situations, a lack of resources can require backtracking and recursing through the above list.¶
Issues such as service preemption of a DetNet flow in favor of another, when resources are scarce, are not considered here. Also not addressed is the question of how to choose the path to be taken by a DetNet flow.¶
In this calculation, all of the DetNet flows are known before the calculation commences. This problem is of interest to relatively static networks, or static parts of larger networks. It provides bounds on latency and buffer size. The calculations can be extended to provide global optimizations, such as altering the path of one DetNet flow in order to make resources available to another DetNet flow with tighter constraints.¶
This calculation may be more difficult to perform than the dynamic calculation (Section 3.1.2), because the DetNet flows passing through one port on a DetNet transit node affect each other's latency. The effects can even be circular, from a node A to B to C and back to A. On the other hand, the static calculation can often accommodate queuing methods, such as transmission selection by strict priority, that are unsuitable for the dynamic calculation.¶
This calculation is dynamic, in the sense that DetNet flows can be added or deleted at any time, with a minimum of computation effort, and without affecting the guarantees already given to other DetNet flows.¶
Dynamic latency-calculation can be done based on the static one described in Section 3.1.1; when a new DetNet flow is created or deleted, the entire calculation for all DetNet flows is repeated. If an already-established DetNet flow would be pushed beyond its latency requirements by the new DetNet flow request, then the new DetNet flow request can be refused, or some other suitable action taken.¶
The choice of queuing methods is critical to the applicability of the dynamic calculation. Some queuing methods (e.g., CQF, Section 6.6) make it easy to configure bounds on the network's capacity, and to make independent calculations for each DetNet flow. Some other queuing methods (e.g., strict priority with the credit-based shaper defined in [IEEE8021Q] section 8.6.8.2) can be used for dynamic DetNet flow creation, but yield poorer latency and buffer space guarantees than when that same queuing method is used for static DetNet flow creation (Section 3.1.1).¶
A model for the operation of a DetNet transit node is required, in order to define the latency and buffer calculations. In Figure 1 we see a breakdown of the per-hop latency experienced by a packet passing through a DetNet transit node, in terms that are suitable for computing both hop-by-hop latency and per-hop buffer requirements.¶
In Figure 1, we see two DetNet transit nodes that are connected via a link. In this model, the only queues, that we deal with explicitly, are attached to the output port; other queues are modeled as variations in the other delay times (e.g., an input queue could be modeled as either a variation in the link delay (2) or the processing delay (4).) There are six delays that a packet can experience from hop to hop.¶
Not shown in Figure 1 are the other output queues that we presume are also attached to that same output port as the queue shown, and against which this shown queue competes for transmission opportunities.¶
In this analysis, the measurement is from the point at which a packet is selected for output in a node to the point at which it is selected for output in the next downstream node (that is the definition of a "hop"). In general, any queue selection method that is suitable for use in a DetNet network includes a detailed specification as to exactly when packets are selected for transmission. Any variations in any of the delay times 1-4 result in a need for additional buffers in the queue. If all delays 1-4 are constant, then any variation in the time at which packets are inserted into a queue depends entirely on the timing of packet selection in the previous node. If the delays 1-4 are not constant, then additional buffers are required in the queue to absorb these variations. Thus:¶
End-to-end latency bounds can be computed using the delay model in Section 3.2. Here, it is important to be aware that for several queuing mechanisms, the end-to-end latency bound is less than the sum of the per-hop latency bounds. An end-to-end latency bound for one DetNet flow can be computed as¶
The two terms in the above formula are computed as follows.¶
First, at the h-th hop along the path of this DetNet flow, obtain an upper-bound per-hop_non_queuing_delay_bound[h] on the sum of the bounds over the delays 1,2,3,4 of Figure 1. These upper bounds are expected to depend on the specific technology of the DetNet transit node at the h-th hop but not on the T-SPEC of this DetNet flow [RFC9016]. Then set non_queuing_delay_bound = the sum of per-hop_non_queuing_delay_bound[h] over all hops h.¶
Second, compute queuing_delay_bound as an upper bound to the sum of the queuing delays along the path. The value of queuing_delay_bound depends on the information on the arrival curve of this DetNet flow and possibly of other flows in the network, as well as the specifics of the queuing mechanisms deployed along the path of this DetNet flow. Note that arrival curve of DetNet flow at source is immediately specified by the T-SPEC of this flow. The computation of queuing_delay_bound is described in Section 4.2 as a separate section.¶
For several queuing mechanisms, queuing_delay_bound is less than the sum of upper bounds on the queuing delays (5,6) at every hop. This occurs with (1) per-flow queuing, and (2) aggregate queuing with regulators, as explained in Section 4.2.1, Section 4.2.2, and Section 6. For other queuing mechanisms the only available value of queuing_delay_bound is the sum of the per-hop queuing delay bounds.¶
The computation of per-hop queuing delay bounds must account for the fact that the arrival curve of a DetNet flow is no longer satisfied at the ingress of a hop, since burstiness increases as one flow traverses one DetNet transit node. If a regulator is placed at a hop, an arrival curve of a DetNet flow at the entrance of the queuing subsystem of this hop is the one configured at the regulator (also called shaping curve in [NetCalBook]); otherwise, an arrival curve of the flow can be derived using the delay-jitter of the flow from the last regulation point (the last regulator in the path of the flow if there is any, otherwise the source of the flow) to the ingress of the hop; more formally, assume a DetNet flow has arrival curve at the last regulation point equal to 'alpha(t)', and the delay-jitter from the last regulation point to the ingress of the hop is 'V'. Then, the arrival curve at the ingress of the hop is 'alpha(t+V)'.¶
For example, consider a DetNet flow with T-SPEC "Interval: tau, MaxPacketsPerInterval: K, MaxPayloadSize: L" at source. Then, a leaky-bucket arrival curve for such flow at source is alpha(t)=r * t+ b, t>0; alpha(0)=0, where r is the rate and b is the bucket size, computed as¶
where L' is the size of any added networking technology-specific encapsulation (e.g., MPLS label(s), UDP, and IP headers). Now, if the flow has delay-jitter of 'V' from the last regulation point to the ingress of a hop, an arrival curve at this point is r * t + b + r * V, implying that the burstiness is increased by r*V. A more detailed information on arrival curves is available in [NetCalBook].¶
With such mechanisms, each flow uses a separate queue inside every node. The service for each queue is abstracted with a guaranteed rate and a latency. For every DetNet flow, a per-node latency bound as well as an end-to-end latency bound can be computed from the traffic specification of this DetNet flow at its source and from the values of rates and latencies at all nodes along its path. An instance of per-flow queuing is IntServ's Guaranteed-Service, for which the details of latency bound calculation are presented in Section 6.5.¶
With such mechanisms, multiple flows are aggregated into macro-flows and there is one FIFO queue per macro-flow. A practical example is the credit-based shaper defined in section 8.6.8.2 of [IEEE8021Q] where a macro-flow is called a "class". One key issue in this context is how to deal with the burstiness cascade: individual flows that share a resource dedicated to a macro-flow may see their burstiness increase, which may in turn cause increased burstiness to other flows downstream of this resource. Computing delay upper bounds for such cases is difficult, and in some conditions impossible [CharnyDelay][BennettDelay]. Also, when bounds are obtained, they depend on the complete configuration, and must be recomputed when one flow is added. (The dynamic calculation, Section 3.1.2.)¶
A solution to deal with this issue for the DetNet flows is to reshape them at every hop. This can be done with per-flow regulators (e.g., leaky bucket shapers), but this requires per-flow queuing and defeats the purpose of aggregate queuing. An alternative is the interleaved regulator, which reshapes individual DetNet flows without per-flow queuing ([SpechtUBS], [IEEE8021Qcr]). With an interleaved regulator, the packet at the head of the queue is regulated based on its (flow) regulation constraints; it is released at the earliest time at which this is possible without violating the constraint. One key feature of per-flow or interleaved regulator is that, it does not increase worst-case latency bounds [LeBoudecTheory]. Specifically, when an interleaved regulator is appended to a FIFO subsystem, it does not increase the worst-case delay of the latter; in Figure 1, when the order of packets from output of queuing subsystem at node A to the entrance of regulator at node B is preserved, then the regulator does not increase the worst-case latency bounds; this is made possible if all the systems are FIFO or a DetNet packet-ordering function (POF) is implemented just before the regulator. This property does not hold if packet reordering occurs from the output of a queuing subsystem to the entrance of next downstream interleaved regulator, e.g., at a non-FIFO switching fabric.¶
Figure 2 shows an example of a network with 5 nodes, aggregate queuing mechanism and interleaved regulators as in Figure 1. An end-to-end delay bound for DetNet flow f, traversing nodes 1 to 5, is calculated as follows:¶
In the above formula, Cij is a bound on the delay of the queuing subsystem in node i and interleaved regulator of node j, and S4 is a bound on the delay of the queuing subsystem in node 4 for DetNet flow f. In fact, using the delay definitions in Section 3.2, Cij is a bound on sum of the delays 1,2,3,6 of node i and 4,5 of node j. Similarly, S4 is a bound on sum of the delays 1,2,3,6 of node 4. A practical example of queuing model and delay calculation is presented Section 6.4.¶
REMARK: If packet reordering does not occur, the end-to-end latency bound calculation provided here gives a tighter latency upper-bound than would be obtained by adding the latency bounds of each node in the path of a DetNet flow [TSNwithATS].¶
A sender can be a DetNet node which uses exactly the same queuing methods as its adjacent DetNet transit node, so that the latency and buffer bounds calculations at the first hop are indistinguishable from those at a later hop within the DetNet domain. On the other hand, the sender may be DetNet-unaware, in which case some conditioning of the DetNet flow may be necessary at the ingress DetNet transit node.¶
This ingress conditioning typically consists of a FIFO with an output regulator that is compatible with the queuing employed by the DetNet transit node on its output port(s). For some queuing methods, this simply requires added buffer space in the queuing subsystem. Ingress conditioning requirements for different queuing methods are mentioned in the sections, below, describing those queuing methods.¶
It is sometimes desirable to build a network that has both DetNet-aware transit nodes and DetNet-unaware transit nodes, and for a DetNet flow to traverse an island of DetNet-unaware transit nodes, while still allowing the network to offer delay and congestion loss guarantees. This is possible under certain conditions.¶
In general, when passing through a DetNet-unaware island, the island may cause delay variation in excess of what would be caused by DetNet nodes. That is, the DetNet flow might be "lumpier" after traversing the DetNet-unaware island. DetNet guarantees for delay and buffer requirements can still be calculated and met if and only if the following are true:¶
The ingress conditioning is exactly the same problem as that of a sender at the edge of the DetNet domain. The requirement for bounds on the latency variation across the DetNet-unaware island is typically the most difficult to achieve. Without such a bound, it is obvious that DetNet cannot deliver its guarantees, so a DetNet-unaware island that cannot offer bounded latency variation cannot be used to carry a DetNet flow.¶
When the input rate to an output queue exceeds the output rate for a sufficient length of time, the queue must overflow. This is congestion loss, and this is what deterministic networking seeks to avoid.¶
To avoid congestion losses, an upper bound on the backlog present in the regulator and queuing subsystem of Figure 1 must be computed during resource reservation. This bound depends on the set of flows that use these queues, the details of the specific queuing mechanism and an upper bound on the processing delay (4). The queue must contain the packet in transmission plus all other packets that are waiting to be selected for output. A conservative backlog bound, that applies to all systems, can be derived as follows.¶
The backlog bound is counted in data units (bytes, or words of multiple bytes) that are relevant for buffer allocation. For every flow or an aggregate of flows, we need one buffer space for the packet in transmission, plus space for the packets that are waiting to be selected for output.¶
Let¶
Then a bound on the backlog of traffic in the queue at this output port is¶
The above bound is over the backlog caused by the traffic entering the queue from the input ports of a DetNet node. If the DetNet node also generates packets (e.g., creation of new packets, replication of arriving packets), the bound must accordingly incorporate the introduced backlog.¶
In this section, we present a general queuing data model as well as some examples of queuing mechanisms. For simplicity of latency bound computation, we assume leaky-bucket arrival curve for each DetNet flow at source. Also, at each DetNet transit node, the service for each queue is abstracted with a minimum guaranteed rate and a latency [NetCalBook].¶
Sophisticated queuing mechanisms are available in Layer 3 (L3, see, e.g., [RFC7806] for an overview). In general, we assume that "Layer 3" queues, shapers, meters, etc., are precisely the "regulators" shown in Figure 1. The "queuing subsystems" in this figure are FIFO. They are not the province solely of bridges; they are an essential part of any DetNet transit node. As illustrated by numerous implementation examples, some of the "Layer 3" mechanisms described in documents such as [RFC7806] are often integrated, in an implementation, with the "Layer 2" mechanisms also implemented in the same node. An integrated model is needed in order to successfully predict the interactions among the different queuing mechanisms needed in a network carrying both DetNet flows and non-DetNet flows.¶
Figure 3 shows the general model for the flow of packets through the queues of a DetNet transit node. The DetNet packets are mapped to a number of regulators. Here, we assume that the PREOF (Packet Replication, Elimination and Ordering Functions) are performed before the DetNet packets enter the regulators. All Packets are assigned to a set of queues. Packets compete for the selection to be passed to queues in the queuing subsystem. Packets again are selected for output from the queuing subsystem.¶
Some relevant mechanisms are hidden in this figure, and are performed in the queue boxes:¶
Ideally, neither of these actions are performed on DetNet packets. Full queues for DetNet packets occurs only when a DetNet flow is misbehaving, and the DetNet QoS does not include "yellow" service for packets in excess of committed rate.¶
The queue assignment function can be quite complex, even in a bridge [IEEE8021Q], since the introduction of per-stream filtering and policing ([IEEE8021Q] clause 8.6.5.1). In addition to the Layer 2 priority expressed in the 802.1Q VLAN tag, a DetNet transit node can utilize the information from the non-exhaustive list below to assign a packet to a particular queue:¶
The "Transmission selection" function decides which queue is to transfer its oldest packet to the output port when a transmission opportunity arises.¶
In [IEEE8021Q] and [IEEE8023], the transmission of a frame can be interrupted by one or more "express" frames, and then the interrupted frame can continue transmission. The frame preemption is modeled as consisting of two MAC/PHY stacks, one for packets that can be interrupted, and one for packets that can interrupt the interruptible packets. Only one layer of frame preemption is supported -- a transmitter cannot have more than one interrupted frame in progress. DetNet flows typically pass through the interrupting MAC. For those DetNet flows with T-SPEC, latency bounds can be calculated by the methods provided in the following sections that account for the effect of frame preemption, according to the specific queuing mechanism that is used in DetNet nodes. Best-effort queues pass through the interruptible MAC, and can thus be preempted.¶
In [IEEE8021Q], the notion of time-scheduling queue gates is described in section 8.6.8.4. On each node, the transmission selection for packets is controlled by time-synchronized gates; each output queue is associated with a gate. The gates can be either open or closed. The states of the gates are determined by the gate control list (GCL). The GCL specifies the opening and closing times of the gates. The design of GCL must satisfy the requirement of latency upper bounds of all DetNet flows; therefore, those DetNet flows that traverse a network that uses this kind of shaper must have bounded latency, if the traffic and nodes are conformant.¶
Note that scheduled traffic service relies on a synchronized network and coordinated GCL configuration. Synthesis of GCL on multiple nodes in network is a scheduling problem considering all DetNet flows traversing the network, which is a non-deterministic polynomial-time hard (NP-hard) problem [Sch8021Qbv]. Also, at this writing, scheduled traffic service supports no more than eight traffic queues, typically using up to seven priority queues and at least one best effort.¶
In this queuing model, it is assumed that the DetNet nodes are FIFO. We consider the four traffic classes (Definition 3.268 of [IEEE8021Q]): control-data traffic (CDT), class A, class B, and best effort (BE) in decreasing order of priority. Flows of classes A and B are DetNet flows that are less critical than CDT (such as studio audio and video traffic, as in IEEE 802.1BA Audio-Video-Bridging). This model is a subset of Time-Sensitive Networking as described next.¶
Based on the timing model described in Figure 1, contention occurs only at the output port of a DetNet transit node; therefore, the focus of the rest of this subsection is on the regulator and queuing subsystem in the output port of a DetNet transit node. The input flows are identified using the information in (Section 5.1 of [RFC8939]). Then they are aggregated into eight macro flows based on their service requirements; we refer to each macro flow as a class. The output port performs aggregate scheduling with eight queues (queuing subsystems): one for CDT, one for class A flows, one for class B flows, and five for BE traffic denoted as BE0-BE4. The queuing policy for each queuing subsystem is FIFO. In addition, each node output port also performs per-flow regulation for class A and B flows using an interleaved regulator (IR), called Asynchronous Traffic Shaper [IEEE8021Qcr]. Thus, at each output port of a node, there is one interleaved regulator per-input port and per-class; the interleaved regulator is mapped to the regulator depicted in Figure 1. The detailed picture of scheduling and regulation architecture at a node output port is given by Figure 4. The packets received at a node input port for a given class are enqueued in the respective interleaved regulator at the output port. Then, the packets from all the flows, including CDT and BE flows, are enqueued in queuing subsystem; there is no regulator for CDT and BE flows.¶
Each of the queuing subsystems for classes A and B, contains a Credit-Based Shaper (CBS). The CBS serves a packet from a class according to the available credit for that class. As described in Section 8.6.8.2 and Annex L.1 of [IEEE8021Q], the credit for each class A or B increases based on the idle slope (as guaranteed rate), and decreases based on the sendslope (typically equal to the difference between the guaranteed and the output link rates), both of which are parameters of the CBS. The CDT and BE0-BE4 flows are served by separate queuing subsystems. Then, packets from all flows are served by a transmission selection subsystem that serves packets from each class based on its priority. All subsystems are non-preemptive. Guarantees for classes A and B traffic can be provided only if CDT traffic is bounded; it is assumed that the CDT traffic has a leaky bucket arrival curve with two parameters r_h as rate and b_h as bucket size, i.e., the amount of bits entering a node within a time interval t is bounded by r_h * t + b_h.¶
Additionally, it is assumed that the classes A and B flows are also regulated at their source according to a leaky bucket arrival curve. At the source, the traffic satisfies its regulation constraint, i.e., the delay due to interleaved regulator at the source is ignored.¶
At each DetNet transit node implementing an interleaved regulator, packets of multiple flows are processed in one FIFO queue; the packet at the head of the queue is regulated based on its leaky bucket parameters; it is released at the earliest time at which this is possible without violating the constraint.¶
The regulation parameters for a flow (leaky bucket rate and bucket size) are the same at its source and at all DetNet transit nodes along its path in the case where all clocks are perfect. However, in reality there is clock non-ideality throughout the DetNet domain even with clock synchronization. This phenomenon causes inaccuracy in the rates configured at the regulators that may lead to network instability. To avoid that, when configuring the regulators, the rates are set as the source rates with some positive margin. [ThomasTime] describes and provides solutions to this issue.¶
A delay bound of the queuing subsystem ((4) in Figure 1) of a given DetNet node for a flow of classes A or B can be computed if the following condition holds:¶
If the condition holds, the delay bounds for a flow of class X (A or B) is d_X and calculated as:¶
where L_min_X is the minimum packet lengths of class X (A or B); c is the output link transmission rate; b_t_X is the sum of the b term (bucket size) for all the flows of the class X. Parameters R_X and T_X are calculated as follows for class A and class B, separately:¶
If the flow is of class A:¶
where I_A is the idle slope for class A; L_nA is the maximum packet length of class B and BE packets; L_n is the maximum packet length of classes A,B, and BE; r_h is the rate and b_h is the bucket size of CDT traffic leaky bucket arrival curve.¶
If the flow is of class B:¶
where I_B is the idle slope for class B; L_A is the maximum packet length of class A; L_BE is the maximum packet length of class BE.¶
Then, as discussed in Section 4.2.2; an interleaved regulator does not increase the delay bound of the upstream queuing subsystem; therefore an end-to-end delay bound for a DetNet flow of class X (A or B) is the sum of d_X_i for all node i in the path the flow, where d_X_i is the delay bound of queuing subsystem in node i which is computed as above. According to the notation in Section 4.2.2, the delay bound of queuing subsystem in a node i and interleaved regulator in node j, i.e., Cij, is:¶
More information of delay analysis in such a DetNet transit node is described in [TSNwithATS].¶
The delay bound calculation requires some information about each node. For each node, it is required to know the idle slope of CBS for each class A and B (I_A and I_B), as well as the transmission rate of the output link (c). Besides, it is necessary to have the information on each class, i.e., maximum packet length of classes A, B, and BE. Moreover, the leaky bucket parameters of CDT (r_h,b_h) must be known. To admit a flow/flows of classes A and B, their delay requirements must be guaranteed not to be violated. As described in Section 3.1, the two problems, static and dynamic, are addressed separately. In either of the problems, the rate and delay must be guaranteed. Thus,¶
The choice of the static values of R and b_t at all nodes and classes must be done in a prior configuration phase; R controls the bandwidth allocated to this class at this node, b_t affects the delay bound and the buffer requirement. The value of R must be set such that¶
where I_X is the idleslope of credit-based shaper for class X={A,B}, c is the transmission rate of the output link and r_h is the leaky-bucket rate of the CDT class.¶
Guaranteed-Service Integrated service (IntServ) is an architecture that specifies the elements to guarantee quality of service (QoS) on networks [RFC2212].¶
The flow, at the source, has a leaky bucket arrival curve with two parameters r as rate and b as bucket size, i.e., the amount of bits entering a node within a time interval t is bounded by r * t + b.¶
If a resource reservation on a path is applied, a node provides a guaranteed rate R and maximum service latency of T. This can be interpreted in a way that the bits might have to wait up to T before being served with a rate greater or equal to R. The delay bound of the flow traversing the node is T + b / R.¶
Consider a Guaranteed-Service IntServ path including a sequence of nodes, where the i-th node provides a guaranteed rate R_i and maximum service latency of T_i. Then, the end-to-end delay bound for a flow on this can be calculated as sum(T_i) + b / min(R_i).¶
The provided delay bound is based on a simple case of Guaranteed-Service IntServ where only a guaranteed rate and maximum service latency and a leaky bucket arrival curve are available. If more information about the flow is known, e.g., the peak rate, the delay bound is more complicated; the details are available in [RFC2212] and Section 1.4.1 of [NetCalBook].¶
Annex T of [IEEE8021Q] describes Cyclic Queuing and Forwarding (CQF), which provides bounded latency and zero congestion loss using the time-scheduled gates of [IEEE8021Q] section 8.6.8.4. For a given class of DetNet flows, a set of two or more buffers is provided at the output queue layer of Figure 3. A cycle time T_c is configured for each class of DetNet flows c, and all of the buffer sets in a class of DetNet flows swap buffers simultaneously throughout the DetNet domain at that cycle rate, all in phase. In such a mechanism, the regulator, mentioned in Figure 1, is not required.¶
In the case of two-buffer CQF, each class of DetNet flows c has two buffers, namely buffer1 and buffer2. In a cycle (i) when buffer1 accumulates received packets from the node's reception ports, buffer2 transmits the already stored packets from the previous cycle (i-1). In the next cycle (i+1), buffer2 stores the received packets and buffer1 transmits the packets received in cycle (i). The duration of each cycle is T_c.¶
The cycle time T_c must be carefully chosen; it needs to be large enough to accommodate all the DetNet traffic, plus at least one maximum packet (or fragment) size from lower priority queues, which might be received within a cycle. Also, the value of T_c includes a time interval, called dead time (DT), which is the sum of the delays 1,2,3,4 defined in Figure 1. The value of DT guarantees that the last packet of one cycle in a node is fully delivered to a buffer of the next node in the same cycle. A two-buffer CQF is recommended if DT is small compared to T_c. For a large DT, CQF with more buffers can be used, and a cycle identification label can be added to the packets.¶
The per-hop latency is determined by the cycle time T_c: a packet transmitted from a node at a cycle (i), is transmitted from the next node at cycle (i+1). Then, if the packet traverses h hops, the maximum latency experienced by the packet is from the beginning of cycle (i) to the end of cycle (i+h); also, the minimum latency is from the end of cycle (i) before the DT, to the beginning of cycle (i+h). Then, the maximum latency is:¶
and the minimum latency is:¶
Ingress conditioning (Section 4.3) may be required if the source of a DetNet flow does not, itself, employ CQF. Since there are no per-flow parameters in the CQF technique, per-hop configuration is not required in the CQF forwarding nodes.¶
This section provides an example application of the timing model presented in this document to control the admission of a DetNet flow on a DetNet-enabled IP network. Consider Figure 5, taken from Section 3 of [RFC8939], that shows a simple IP network:¶
Consider a fully centralized control plane for the network of Figure 5 as described in Section 3.2 of [I-D.ietf-detnet-controller-plane-framework]. Suppose end-system 1 wants to create a DetNet flow with traffic specification destined to end-system 2 with end-to-end delay bound requirement D. Therefore, the control plane receives a flow establishment request and calculates a number of valid paths through the network (Section 3.2 of [I-D.ietf-detnet-controller-plane-framework]). To select a proper path, the control plane needs to compute an end-to-end delay bound at every node of each selected path p.¶
The end-to-end delay bound is d1 + d2_p + d3_p, where d1 is the delay bound from end-system 1 to the entrance of relay node 1, d2_p is the delay bound for path p from relay node 1 to entrance of the first node in sub-network 2, and d3_p the delay bound of path p from the first node in sub-network 2 to end-system 2. The computation of d1 is explained in Section 6.5. Since the relay node 1, sub-network 1 and relay node 2 implement aggregate queuing, we use the results in Section 4.2.2 and Section 6.4 to compute d2_p for the path p. Finally, d3_p is computed using the delay bound computation of Section 6.6. Any path p such that d1 + d2_p + d3_p <= D satisfies the delay bound requirement of the flow. If there is no such path, the control plane may compute new set of valid paths and redo the delay bound computation or reject the DetNet flow.¶
As soon as the control plane selects a path that satisfies the delay bound constraint, it allocates and reserves the resources in the path for the DetNet flow (Section 4.2 [I-D.ietf-detnet-controller-plane-framework]).¶
Detailed security considerations for DetNet are cataloged in [RFC9055], and more general security considerations are described in [RFC8655].¶
Security aspects that are unique to DetNet are those whose aim is to provide the specific QoS aspects of DetNet, specifically bounded end-to-end delivery latency and zero congestion loss. Achieving such loss rates and bounded latency may not be possible in the face of a highly capable adversary, such as the one envisioned by the Internet Threat Model of BCP 72 [RFC3552] that can arbitrarily drop or delay any or all traffic. In order to present meaningful security considerations, we consider a somewhat weaker attacker who does not control the physical links of the DetNet domain but may have the ability to control or change the behavior of some resources within the boundary of the DetNet domain.¶
Latency bound calculations use parameters that reflect physical quantities. If an attacker finds a way to change the physical quantities, unknown to the control and management planes, the latency calculations fail and may result in latency violation and/or congestion losses. An example of such attacks is to make some traffic sources under the control of the attacker send more traffic than their assumed T-SPECs. This type of attack is typically avoided by ingress conditioning at the edge of a DetNet domain. However, it must be insured that such ingress conditioning is done per-flow and that the buffers are segregated such that if one flow exceeds its T-SPEC, it does not cause buffer overflow for other flows.¶
Some queuing mechanisms require time synchronization and operate correctly only if the time synchronization works correctly. In the case of CQF, the correct alignments of cycles can fail if an attack against time synchronization fools a node into having an incorrect offset. Some of these attacks can be prevented by cryptographic authentication as in Annex K of [IEEE1588] for the Precision Time Protocol (PTP). However, the attacks that change the physical latency of the links used by the time synchronization protocol are still possible even if the time synchronization protocol is protected by authentication and cryptography [DelayAttack]. Such attacks can be detected only by their effects on latency bound violations and congestion losses, which do not occur in normal DetNet operation.¶
This document has no IANA actions.¶
We would like to thank Lou Berger, Tony Przygienda, John Scudder, Watson Ladd, Yoshifumi Nishida, Ralf Weber, Robert Sparks, Gyan Mishra, Martin Duke, Eric Vyncke, Lars Eggert, Roman Danyliw, and Paul Wouters for their useful feedback on this document.¶
RFC 7322 limits the number of authors listed on the front page to a maximum of 5. The editor wishes to thank and acknowledge the following author for contributing text to this document¶
Janos Farkas Ericsson Email: janos.farkas@ericsson.com¶