| DeNet WG | L. Chen |
| Internet-Draft | S. Qian |
| Intended status: Standards Track | SEU |
| Expires: November 7, 2019 | T. Ao |
| ZTE | |
| May 6, 2019 |
DetNet Packet Jitter, Bandwidth and Misordering Passive Performance Measurement
draft-detnet-jitter-bandwidth-misordering-00
Deterministic Networking (DetNet) can provide data transmission with end-to-end bounded latency and extremely low packet loss for user’s services. In order to better control and manage deterministic network services. It is necessary to measure and monitor DetNet QoS information. As introduced in [I-D.chen-detnet-loss-delay], packet loss rates and end-to-end delay can be measured by using passive Performance Measurement (PM) in MPLS-based DetNet encapsulation.
This document implement three new QoS related attribute to support passive Performance Measurement for 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 November 7, 2019.
Copyright (c) 2019 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.
As defined in [I-D.ietf-detnet-architecture], DetNet can provide data transmission with end-to-end bounded latency and extremely low packet loss for user’s services. In order to ensure the efficiency and accuracy of information collection and management, it is necessary to measure and monitor DetNet QoS information.
DetNet QoS information includes minimum and maximum delay, bounded jitter, packet loss ratio and an upper bound on misordering packet in general. Apart from this, link bandwidth information also play an important role in resource allocation.
As introduced in [I-D.chen-detnet-loss-delay], packet loss rates and end-to-end delay can be measured by using passive Performance Measurement (PM) in MPLS-based DetNet encapsulation. Which defines two new flags in the d-CW(control word) and three new TLVs to LM and DM messages.
Inspird by that, we implement protocol mechanisms to support passive Performance Measurement for bounded jitter, misordering packet and used bandwidth.
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 [RFC2119].
The terminology is defined as [I-D.ietf-detnet-architecture], [RFC3270], [RFC6374], [RFC2475] and [RFC2474].
MPLS-based DetNet encapsulation introduces an S-Lable and a d-CW. Meanwhile [I-D.chen-detnet-loss-delay] document defines two new flags in the d-CW. Here this document defines a new flag in the d-CW(as shown in Figure 1). The B bit is defined to indicate whether the bandwidth measurement is enabled.
+-----------------+
~ IP/MPLS Tunnel ~
+-----------------+ <--\
| Service Label | |
+-----------------+ +-- Service Layer Header
+----| Control Word | |
| +-----------------+ <--/
| | Payload |
| +-----------------+
|
| +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+--->|0 0 0 0|L|D|B| Sequence Number |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 1: DetNet Control Word
where:
Assume a DetNet service path between node A and node B, where node A is the ingress node, and node B is the egress node. The packets with same interval number belong to the same measurement interval.
[Editor notes: The detail of measurement interval and flags in DetNet Control Word can be found in [I-D.chen-detnet-loss-delay]]
In brief, jitter can be calculated by delay.To measure the delay of a packet, the D bit of the d-CW MUST be set.
If the D bit of the d-CW is set, recording both the flow-id of DetNet flow and the node-id of node.
At the ingress node, record the time when sending the packet, with the timestamp indexed by the sequence number. At the egress node, when receiving a packet with D bit set, record the time when the packet was received, with the timestamp indexed by the sequence number.
When the measurement interval comes, all the information of node-id, flow-id, sequence number and timestamps need to be sent to the centralized controller.
The mechanism for sending information to a centralized controller is out side the scope of this document.
After that, the centralized controller can find the sequence number and timestamps between adjacent nodes such as AB by flow-id and node-id.
Then, with the timestamps from the ingress and egress nodes, and the sequence number, the packet delay can be calculated as below.
Delay[n] = B_RxT[n] - A_TxT[n], where:
After getting multiple sets of Delay[n] informatin, the packet jitter in the No. "t" measurement interval can be calculated as below.
Jitter[t] = Max(Delay[i] - Delay[j]), where:
To measure the used bandwidth of DetNet flows between A and B, both of the D bit and B bit of the d-CW MUST be set.
If the B and D bit of the d-CW are set, recording both the flow-id of DetNet flow and the node-id of node.
At the ingress node, record the time and accumulate the packet bytes when sending the packet, with the timestamp indexed by the sequence number. At the egress node, when receiving a packet with B and D bit set, record the time and accumulate the packet bytes when the packet was received, with the timestamp indexed by the sequence number.
When the measurement interval comes, all the information of node-id, flow-id, sequence number, timestamps and packet bytes need to be sent to the centralized controller. Then the packet bytes need to be reset to zero for next measurement interval.
After that, the centralized controller can find the sequence number, timestamps and packet bytes between adjacent nodes such as AB by flow-id and node-id.
Then, with the timestamps and packet bytes from the ingress and egress nodes, and the sequence number, the used bandwidth in No. "t" measurement interval can be calculated as below.
Bandwidth[t] = Max(A_TxB[t], B_RxB[t])*8/Delay [t], where:
Since the measurement of the maximum misordering packets only requires the sequence number of the DetNet flows, there is no need to additionally add a flag bit to the MPLS control word. The D bit of the d-CW can re reused.
To measure the maximum misordering packets, the D bit of the d-CW MUST be set.
At the ingress node, record the time when sending the packet, with the timestamp indexed by the sequence number. At the egress node, when receiving a packet with D bit set, record the time when the packet was received, with the timestamp indexed by the sequence number.
When the measurement interval comes, all the information of node-id, flow-id, sequence number and timestamps need to be sent to the centralized controller.
After that, the centralized controller can find the sequence number and timestamps between adjacent nodes such as AB by flow-id and node-id.
A packet can be classified as a misordering packet if it has a sequence number smaller than its predecessors.
Specifically, let M DetNet flows, denoted as (S1, . . . ,Sm), be the total number of flows sent from node A to B.
In each flow Si consisting of K packets, we assign to each packet j a sequence number aj which is a successive integer from 1 to K in the order of the packet emission and so we create the source sequence as (a1, . . . ,aK). Assume an output sequence (b1, . . . , bP) of Si observed at the receiving node B, where P ≤ K be the total number of packets received out of the K packets sent. Due to loss, the amount K may less than P.
The sequence is said to be in order if for any index k (1 ≤ k ≤ P) holds bk ≤ bq (0 ≤ q ≤ k), else the flow is said to reached at the destination midordering, and the packet k is a reordered packet in the reordered flow. The total number of reordered packets in flow Si is written as Li.
For example, for the sequence of an arrived reordered flow (1,2,3,5,4,7,6,8), there are 2 reordered packets (packet 4 and packet 6), which leads to L = 2. Note that in this document reordering does not correlate with loss (same as [2][8][9]). For example, a received flow (1,2,3,4,5,6,8) is considered as in order.
TBD.
TBD.
TBD.
| [RFC2475] | Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z. and W. Weiss, "An Architecture for Differentiated Services", RFC 2475, DOI 10.17487/RFC2475, December 1998. |