Beckhoff EtherCAT Terminals
Industrial Ethernet Book Issue 120 / 15
Request Further Info   Print this Page   Send to a Friend  

FRER vs. PRP protocol in Time Sensitive Networks

The availability of a network is the probability that the network is in service and available for use at any instant in time. A comparison between PRP/HSR and FRER solutions discussed in this article, used to achieve static redundancy reveals critical technical differences in both system costs and performance.

VARIOUS PROTOCOLS CAN BE USED TO PROVIDE high availability in EtherNet/IP systems. The protocols focused on in this article are the Time Sensitive Networks (TSN) feature Frame Replication and Elimination for Reliability (FRER) defined in the IEEE 802.1CB-2017 standard; and the Parallel Redundancy Protocol (PRP) and High-availability Seamless Redundancy (HSR) protocol, both defined in the IEC 62439 standard.

Critical system applications are often required to maintain high availability of communication network components. For critical infrastructures and time sensitive processes, downtime is never allowed. The protocols focused on here (FRER, PRP, HSR) provide zero recovery time. We will compare and contrast these protocols in reference to network topology, frame structures, network convergence and cost to deploy.


Static Redundancy.

High availability

High availability is based on the concept of availability. The availability of a network is the probability (in percent) that the network is in service and available for use at any instant in time.

High availability is represented as a percentage, usually referred to as the 9s. If the availability metric is specified as five nines, it is understood to mean that the network should be functional for 99.999% of the desired duty cycle (24-hours/day).

Availability is expressed using the following measures of reliability.

Availability = MTTF / (MTTF+MTTR) (1) where MTTF is the mean time to failure; a measure of the reliability of a network, otherwise known is its failure rate. The MTTF is the interval in which the network or element can provide service without failure.

MTTR is the mean time to repair; a measure of reliability that represents the time it takes to resume service after a failure has been experienced.

As equation above shows, the availability of a network can be increased by designing network elements that are highly reliable (high MTTF), and/or by reducing the time required to repair the network and return it to service (low MTTR).

Since it is impossible to create networks that never fail, the key to high availability is to make recovery time as brief as possible. Availability is increased in networks by introducing redundancy.

Redundancy

High availability can be achieved economically by using techniques that detect points of failure and avoid service interruptions through redundancy in the system. There are two forms of redundancy, dynamic and static.

In dynamic (standby) redundancy the replicated components activate after a failure has been detected. Dynamic redundancy does not actively participate in the control. Switchover logic determines when to insert and activate redundancy.

In static (parallel) redundancy the replicated components are active concurrently. Static redundancy usually participates in the control. No special processing is needed on errors. This provides bumpless (0 ms) switchover, with continuously exercised redundancy and increased point-of-failure detection with fail-safe behavior. Static redundancy is provided at the cost of duplication.

The two set of protocols discussed here provide Static Redundancy. They are:

  • Parallel Redundancy Protocol (PRP) and Highly-available Seamless Redundancy (HSR) protocol, both defined in the IEC 62439-2 standard;
  • Frame Replication and Elimination for Reliability (FRER) defined in the IEEE 802.1CB standard.

Sample Target Solution

A Sample Target Network shown above will be used as illustration for the comparison of PRP/HSR and FRER.

The Sample Target Network is a complex network consisting of the following: a basic Star network centered around the E01 Ethernet Switch; A Ring network with the Ethernet Switch E02 as the Head of the Ring; and a Line network with Ethernet switch E03 at the Head of the Line.

The Sample network includes control applications consisting of Single Port Sensor and Drive devices (Sx), Multiple Port Sensor and IO devices (Mx), Ethernet Switches (Ex), and cables. A video monitoring network (Camera, Monitor, and Recorder) has be converged with the control network.

IEC 62439-3 PRP/HSR Solution

The IEC 62439-3 standard specifies two redundancy protocols designed to provide seamless recovery in case of single failure of an inter-bridge link or bridge in the network, which are based on the same scheme: duplication of the LAN, and/or duplication of the transmitted information. Further improvements in recovery time require managing of redundancy in the end nodes, by equipping the end nodes with several, redundant communication links. In general, doubly attached end nodes provide enough redundancy.

Conversion of the Ring

The conversion of the Sample Target Network the Ring portion of the network will be converted to an HSR Network. The Multiport Devices will be converted to Doubly Attached Nodes with HSR Protocol (DANH) and will stay arranged as a ring. These Dual Attached Nodes within the ring are restricted to be HSR-capable bridging nodes, thus avoiding the use of dedicated switches.

Single-port Devices, known as Singly Attached Nodes (SANs) cannot be attached directly to the ring, but need attachment through a Redundancy Box (RedBox). Ethernet Switch (E04) will need to be replaced with an HSR Redundancy Box (R06) to connect the Single-port Camera (S06).

Each DANH has two identical interfaces, port A and port B. For each frame, the source node sends one copy over each of its two ports. The source node removes the frames it injected into the ring. Each node (between source and destination) relays a frame it receives from port A to port B and vice-versa, except if already forwarded. The destination node consumes the first frame of a pair and discards the duplicate. If the ring is broken, frames still arrive over the intact path, with no impact on the application. Loss of a path is easily detected since duplicates cease to arrive. HSR is not defined in this CIP specification but defined in the sub-clauses 5 and 7 of the IEC 62439-3 standard (IEC 62439-3:2012-7).


Star and Line Networks

The Star and Line Networks will be converted to two duplicate line networks and implement the PRP redundancy protocol. The PRP redundancy protocol implements redundancy in the devices such as Double attached nodes implementing PRP (DANPs) and Redundancy Boxes (Red Box).

A DANP is attached to two independent Local Area Networks (LANs) of similar topology (LAN_A [Blue] and LAN_B [Red]) which operate in parallel. One DANP (a source) sends the same frame over both LANs to another DANP (Destination) who receives it from both LANs, consumes the first frame and discards the duplicate. The same mechanism of duplicate generation and rejection can be implemented by a Red-Box. A Red-Box does the transition between a Singly Attached Node (SAN) and the doubled LANs (LAN_A and LAN_B). The Red-Box mimics the SANs connected behind it (called VDAN or virtual DANs) and multicasts supervision frames on their behalf. The Red-Box is itself a DANP and has its own IP address for management purposes, but it may also perform application functions.

The two LANs are identical in protocol at the MAC-LLC level, but they can differ in performance and topology. Transmission delays may also be different, especially if one of the networks reconfigures itself, for example using RSTP, to overcome an internal failure. The two LANs follow configuration rules that allow the network management protocols such as Address Resolution Protocol (ARP) to operate correctly.

The two LANs shall have no connection between them and are assumed to be fail-independent. Redundancy can be defeated by single points of failure, such as a common power supply or a direct connection whose failure brings both networks down. Refer to the installation guidelines in the IEC 62439-3 standard (IEC 62439-3:2012-7) to provide guidance to the installer to achieve fail-independence. PRP is not defined in the CIP specification but defined in the sub-clauses 4 and 7 of the IEC 62439-3 standard (IEC 62439-3:2012-7)

IEEE 802.1CB FRER Solution

FRER is a static (parallel) redundancy high availability capability as defined in the IEEE 802.1CB-2017 standard. This solution is illustrated by transforming the Sample Target Network into an FRER supported network.

The additional network interconnections needed within the network are representing by the dashed lines. These interconnections change an existing network into a Mesh network.

FRER provides increased reliability (reduced packet loss rates) for a Stream by using a sequence numbering scheme, and by replicating every stream packet in the source. FRER also eliminates those replicated stream packets in the destination. FRER provides:

  • Packet replication: sending replicated frames on separate paths, and then using inserted identification information to eliminate replicates, reducing the probability of frame loss.
  • Multicast or unicast: A path on which a Stream is sent can be a point-to-point path or a point-to-multipoint tree.
  • Latent error detection: some means of detecting a failure to deliver copies of each packet is provided at the point that the replicated packets are discarded.
  • Interoperability: a small number of controls are provided that make interoperation with other standards possible.
  • Backward compatibility: To provide the ability to be connected to a network that is not aware of FRER, and for a network of conformant relay systems to offer these benefits to unaware end systems.
  • Zero congestion loss: provide a Stream with zero (or very low) packet loss due to congestion.

The FRER protocol provides increased reliability (reduced packet loss rates) for a Stream by using a sequence numbering scheme, and by replicating every stream packet in the source end system and/or in relay systems in the network. FRER also eliminates those replicates in the destination end system and/or in other relay systems. The devices types described in the standard are:

End Systems (ES): End Systems may contain a Talker component, a Listener component, or both. End Systems are represented by: The Single-Port Devices (Sx): Controller; Drives, Sensor; IP Camera; Camera Recorder; and Monitor.

Relay Systems (RS): Relay Systems will either transfer packets belonging to redundant streams, or act as a proxy Talker or Listener for End Systems not capable of handling redundant streams. Relay Systems are represented by the Ethernet Switch(es).

Relay-End Systems (RES): Relay-End Systems are not defined in the IEEE Standard but are elements within the EtherNet/IP Network. The Rely-End System is created by combining the FRER End system and Relay System capabilities. Relay-End Systems are represented by the Multi-Port IO Device(s) and one Multi-Port Sensors.


Cost Comparison: PRP/HSR vs. TSN-FRER

PRP/HSR Topology Performance

This section will discuss to operations within the HSR Ring in the sample PRP/HSR network as comparted to a similar operation within the FRER Mesh. One key to the technology is the message flow between the MP IO Device (D08) and the SP Controller (D01).

The message from IO Device (D08) is duplicated along two paths. The messages transition to the HSR Redundancy Boxes whose job it is to transfer the messages to the respective PRP networks. The message traverses the two separate PRP networks until reaching the PRP Redundancy Box supporting the SP Controller (D01).

The messages will also traverse back through the HSR Network to their source to be removed from the network. This is demonstrating the need to account to double the bandwidth for messages within an HSR Network.

One message traverses 8 hops, while the second message only traverses 7 hops. The one message will arrive at R01 first and be selected to generate the message to the D01 controller. The other will be dropped once it reached R01, as it has arrived later.

An error may occur within the HSR Ring. The error in this example is a break in the cable between R02B and D06 MP IO Device.

The message from IO Device (D08) is duplicated along two paths. One message has stopped transitioning due to the cable break, but the second message continues to the HSR Redundancy.

One message continues to traverse the 8 hops, while the other has been stopped. The first message will be used this time by R01 since it is the only message to arrive, though it will arrive 1 hop later.

FRER Mesh topology performance

This section describes the use of FRER in an EtherNet/IP network example. The mesh network is supported by any of the Network Control Protocols defined in the IEEE 802.1Q-2018 Standard (i.e. RSTP, MSTP, SBB, etc.).

In this scenario the D08 MP IO Device is an End System (RES) acting as a Talker that transmits a redundant stream to the D01 SP Controller (ES). The Talker proxy will generate redundant streams:

  • Sequencing information in frames;
  • Replicates each frame passed to it, assigning each replicate a different stream handle, at most one of which can be the same as the original passes unchanged;
  • Triggers the sending of one stream (Blue) before the other (Red) to keep the propagation delay within the network the same at the destination (7 hops).
  • Encodes the sequencing information into the frame in a manner such that it can be decoded by its peer.

The S01 Ethernet Switch (RS) are connected to the D01 SP Controller and acting as Listener for the Redundant stream from D08 MP IO Device. When the Listener proxy receives redundant streams:

  • Extracts and decodes the sequencing information from a received frame.
  • Examines this sequencing information in received frames packets and discards frames indicated to be a duplicate of a frame previously received and forwarded; Also monitors the variables to detect latent errors of streams.

An error may occur within the Mesh Network. The error can be illustrated as a break in the cable between So2 Ethernet Switch and D06 MP IO Device. The network recovery time of these control protocols is irrelevant due to the seamless redundancy nature of the FRER protocol.

In this scenario the D08 MP IO Device is an End System (RES) acting as a Talker that transmits a redundant stream to the D01 SP Controller (ES). The Talker proxy will generate redundant streams:

  • Sequencing information in frames;
  • Replicates each frame passed to it, assigning each replicate a different stream handle, at most one of which can be the same as the original passes unchanged;
  • Triggers the sending of one stream before the other to keep the propagation delay within the network the same at the destination (7 hops).
  • Encodes the sequencing information into the frame in a manner such that it can be decoded by its peer.

George Ditzel, Ethernet Architect, Schneider Electric.


Source: Industrial Ethernet Book Issue 120 / 15
Request Further Info    Print this Page    Send to a Friend  

Back

Sponsors:
sps connect: Automation goes digital
Accelerate your HART data at the speed of Ethernet
Industry of Things World

Get Social with us:



© 2010-2020 Published by IEB Media GbR · Last Update: 28.10.2020 · 16 User online · Privacy Policy · Contact Us