| detnet | N. Finn |
| Internet-Draft | Huawei Technologies Co. Ltd |
| Intended status: Informational | P. Thubert |
| Expires: September 20, 2018 | Cisco |
| March 19, 2018 |
Deterministic Networking Problem Statement
draft-ietf-detnet-problem-statement-03
This paper documents the needs in various industries to establish multi-hop paths for characterized flows with deterministic properties .
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 September 20, 2018.
Copyright (c) 2018 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.
The Deterministic Networking Use Cases document illustrates that beyond the classical case of industrial automation and control systems (IACS), there are in fact multiple industries with strong and yet relatively similar needs for deterministic network services with latency guarantees and ultra-low packet loss.
The generalization of the needs for more deterministic networks have led to the IEEE 802.1 AVB Task Group becoming the Time-Sensitive Networking (TSN) Task Group (TG), with a much-expanded constituency from the industrial and vehicular markets.
Along with this expansion, the networks in consideration are becoming larger and structured, requiring deterministic forwarding beyond the LAN boundaries. For instance, IACS segregates the network along the broad lines of the Purdue Enterprise Reference Architecture (PERA) [ISA95], typically using deterministic local area networks for level 2 control systems, whereas public infrastructures such as Electricity Automation require deterministic properties over the Wide Area. The realization is now coming that the convergence of IT and Operational Technology (OT) networks requires Layer-3, as well as Layer-2, capabilities.
While the initial user base has focused almost entirely on Ethernet physical media and Ethernet-based bridging protocol (from several Standards Development Organizations), the need for Layer-3 expressed above, must not be confined to Ethernet and Ethernet-like media, and while such media must be encompassed by any useful DetNet architecture, cooperation between IETF and other SDOs must not be limited to IEEE or IEEE 802. Furthermore, while the work completed and ongoing in other SDOs, and in IEEE 802 in particular, provide an obvious starting point for a DetNet architecture, we must not assume that these other SDOs' work confines the space in which the DetNet architecture progresses.
The properties of deterministic networks will have specific requirements for the use of routed networks to support these applications and a new model must be proposed to integrate determinism in IT technology. The proposed model should enable a fully scheduled operation orchestrated by a central controller, and may support a more distributed operation with probably lesser capabilities. In any fashion, the model should not compromise the ability of a network to keep carrying the sorts of traffic that is already carried today in conjunction with new, more deterministic flows.
Once the abstract model is agreed upon, the IETF will need to specify the signaling elements to be used to establish a path and the tagging elements to be used identify the flows that are to be forwarded along that path. The IETF will also need to specify the necessary protocols, or protocol additions, based on relevant IETF technologies, to implement the selected model.
As a result of this work, it will be possible to establish a multi-hop path over the IP network, for a particular flow with given timing and precise throughput requirements, and carry this particular flow along the multi-hop path with such characteristics as low latency and ultra-low jitter, duplication and elimination of packets over non-congruent paths for a higher delivery ratio, and/or zero congestion loss, regardless of the amount of other flows in the network.
Depending on the network capabilities and on the current state, requests to establish a path by an end-node or a network management entity may be granted or rejected, an existing path may be moved or removed, and DetNet flows exceeding their contract may face packet declassification and drop.
The Internet is not the only digital network that has grown dramatically over the last 30-40 years. Video and audio entertainment, and control systems for machinery, manufacturing processes, and vehicles are also ubiquitous, and are now based almost entirely on digital technologies. Over the past 10 years, engineers in these fields have come to realize that significant advantages in both cost and in the ability to accelerate growth can be obtained by basing all of these disparate digital technologies on packet networks.
The goals of Deterministic Networking are to enable the migration of applications that use special-purpose fieldbus technologies (HDMI, CANbus, ProfiBus, etc... even RS-232!) to packet technologies in general, and the Internet Protocol in particular, and to support both these new applications, and existing packet network applications, over the same physical network.
Considerable experience ([ODVA]/[EIP],[AVnu], [Profinet],[HART],[IEC62439], [ISA100.11a] and [WirelessHART], etc...) has shown that these applications need a some or all of a suite of features that includes:
Time synchronization techniques need not be addressed by an IETF Working Group; there are a number of standards available for this purpose, including IEEE 1588, IEEE 802.1AS, and more.
The multicast, latency, loss ratio, and non-throttling needs are made necessary by the algorithms employed by the applications. They are not simply the transliteration of fieldbus needs to a packet-based fieldbus simulation, but reflect fundamental mathematics of the control of a physical system.
With classical forwarding latency- and loss-sensitive packets across a network, interactions among different critical flows introduce fundamental uncertainties in delivery schedules. The details of the queuing, shaping, and scheduling algorithms employed by each bridge or router to control the output sequence on a given port affect the detailed makeup of the output stream, e.g. how finely a given flow's packets are mixed among those of other flows.
This, in turn, has a strong effect on the buffer requirements, and hence the latency guarantees deliverable, by the next bridge or router along the path. For this reason, the IEEE 802.1 Time-Sensitive Networking Task Group has defined a new set of queuing, shaping, and scheduling algorithms that enable each bridge or router to compute the exact number of buffers to be allocated for each flow or class of flows.
Robustness is a common need for networking protocols, but plays a more important part in real-time control networks, where expensive equipment, and even lives, can be lost due to misbehaving equipment.
Reserving resources before packet transmission is the one fundamental shift in the behavior of network applications that is impossible to avoid. In the first place, a network cannot deliver finite latency and practically zero packet loss to an arbitrarily high offered load. Secondly, achieving practically zero packet loss for un-throttled (though bandwidth limited) flows means that bridges and routers have to dedicate buffer resources to specific flows or to classes of flows. The requirements of each reservation have to be translated into the parameters that control each host's, bridge's, and router's queuing, shaping, and scheduling functions and delivered to the hosts, bridges, and routers.
In some use cases, the end point which run the application is involved in the deterministic networking operation, for instance by controlling certain aspects of its throughput such as rate or precise time of emission. In that case, the deterministic path is end-to-end from application host to application host.
On the other end, the deterministic portion of a path may be a tunnel between and ingress and an egress router. In any case, routers and switches in between should not need to be aware whether the path is end-to-end of a tunnel.
While it is clear that DetNet does not aim at setting up deterministic paths over the global Internet, there is still a lack of clarity on the limits of a domain where a deterministic path can be set up. These limits may depend in the technology that is used to seu th epath up, whether it is centralized or distributed.
Deterministic forwarding can only apply on flows with well-defined characteristics such as periodicity and burstiness. Before a path can be established to serve them, the expression of those characteristics, and how the network can serve them, for instance in shaping and forwarding operations, must be specified.
A centralized routing model, such as provided with a PCE, enables global and per-flow optimizations. The model is attractive but a number of issues are left to be solved. In particular:
To enable a centralized model, DetNet should produce the complete SDN architecture with describes at a high level the interaction and data models to:
Whether a distributed alternative without a PCE can be valuable could be studied as well. Such an alternative could for instance inherit from the Resource ReSerVation Protocol (RSVP-TE) flows. But the focus of the work should be to deliver the centralized approach first.
To enable a RSVP-TE like functionality, the following steps would take place:
In some cases the duplication and elimination of packets over non-congruent paths is required to achieve a sufficiently high delivery ratio to meet application needs. In these cases, a small number of packet formats and supporting protocols are required (preferably, just one) to serialize the packets of a DetNet stream at one point in the network, replicate them at one or more points in the network, and discard duplicates at one or more other points in the network, including perhaps the destination host. Using an existing solution would be preferable to inventing a new one.
Security in the context of Deterministic Networking has an added dimension; the time of delivery of a packet can be just as important as the contents of the packet, itself. A man-in-the-middle attack, for example, can impose, and then systematically adjust, additional delays into a link, and thus disrupt or subvert a real-time application without having to crack any encryption methods employed. See [RFC7384] for an exploration of this issue in a related context.
Typical control networks today rely on complete physical isolation to prevent rogue access to network resources. DetNet enables the virtualization of those networks over a converged IT/OT infrastructure. Doing so, DetNet introduces an additional risk that flows interact and interfere with one another as they share physical resources such as Ethernet trunks and radio spectrum. The requirement is that there is no possible data leak from and into a deterministic flow, and in a more general fashion there is no possible influence whatsoever from the outside on a deterministic flow. The expectation is that physical resources are effectively associated with a given flow at a given point of time. In that model, Time Sharing of physical resources becomes transparent to the individual flows which have no clue whether the resources are used by other flows at other times.
Security must cover:
This document does not require an action from IANA.
The authors wish to thank Lou Berger, Jouni Korhonen, Erik Nordmark, George Swallow, Rudy Klecka, Anca Zamfir, David Black, Thomas Watteyne, Shitanshu Shah, Craig Gunther, Rodney Cummings, Wilfried Steiner, Marcel Kiessling, Karl Weber, Ethan Grossman, Patrick Wetterwald, Subha Dhesikan, Rudy Klecka and Pat Thaler for their various contribution to this work.