Network Resiliency – Resilient Ethernet Protocol (REP)

Part 2 – Resilient Ethernet Protocol (REP)

Resilient plant-wide network architectures play a pivotal role in maintaining overall plant uptime and productivity. Industrial Automation and Control System (IACS) application requirements such as availability and performance drive the choice of resiliency technology.

When selecting resiliency technology, various plant application factors should be evaluated, including physical layout of IACS devices (geographic dispersion), resiliency performance, uplink media type, tolerance to data latency and jitter and future-ready requirements.

In this network resiliency series, we will highlight the various network resiliency protocols, such as DLR, REP, FlexLinks, EtherChannel, and see how they may be applied in various IACS applications.

This week we will feature Resilient Ethernet Protocol, or REP.

What is REP?

Resilient Ethernet Protocol (REP) is a technology implemented on Cisco distribution switches and Cisco IE and Rockwell Automation Stratix® IES. REP is designed to provide fast network and application convergence in case of a media or network failure, without a negative impact on most network applications.

REP is a segment protocol that integrates easily into existing CPwE Cell/Area Zone LANs. Although REP disables STP on interfaces where REP is enabled, it can coexist with STP as part of the same Cell/Area Zone LAN. Since REP can also notify STP about potential topology changes, it allows for interoperability between the two.

REP is a distributed and secure control plane protocol that does not rely on a master switch controlling the status of the ring. Therefore, failures can be detected locally, either through loss of signal (LOS) or loss of connectivity to a neighboring switch. By default, REP automatically elects an alternate port (the switch port being blocked). Any REP port within the REP topology can initiate a switchover to unblock the alternate port.

REP Operation

A REP segment is a chain of switch ports connected to each other and configured with the same segment ID. Each end of a segment terminates on what is called the “edge port” of an edge switch. Note that each switch in a segment has exactly two REP-enabled ports.

With REP, in order to prevent a loop in the network, one switch port (the alternate port) is always blocked in any given segment. The blocked port helps achieve loop-free traffic within the segment by requiring traffic flow to exit only one of the edge ports. Therefore, when a failure occurs in the segment, REP opens the alternate port so traffic can reach the edge of the segment. The figures below show the basic operation of REP to converge the network when a disruption occurs.

What Stratix® switches support REP?

  • Stratix 5400 switches
  • Stratix 5410 switches
  • Stratix 5700 switches (LITE and FULL)
  • Stratix 5800
  • Stratix 8000 switches
  • Stratix 8300 switches

Single Ring (Single Media)

In a single access ring design consisting of any IES model with up to 50 switches, REP should generally be used for resiliency, since it provides significantly better reaction time following a disruption than other protocols. The REP segment should be configured with the edges co-located on the primary distribution switch, as shown in the figure below. All other ports in the ring should be configured as members of the segment.

In this example the Stratix 5410 or Stratix 5400 is used as the distribution platform configured in HSRP mode with REP.

Single Ring (Dual Media)

If the access ring is constructed entirely from Stratix 5400 switches, which has four SFP gigabit uplinks available, then an alternative to the previous design is a dual media ring. In this design, the access switches have two links between each switch that are grouped into an EtherChannel, as shown in the figure below. REP is still recommended for resiliency and is configured as in the single media case.

In this example the Stratix 5410 is used as the distribution platform configured in HSRP mode with EtherChannel and REP.

Multiple Ring Segments

The following use case represents the requirement for configuring multiple REP segments using one pair of Stratix 5410 switches in an HSRP configuration.

Each REP segment will have a segment edge on each of the HSRP nodes (that is, the primary REP segment edge on the active HSRP node and the secondary REP segment edge on the standby HSRP node).

This design requires an additional REP segment to be configured as the trunk between the two HSRP nodes. Preferably, this REP segment will be configured on two 10 Gigabit ports to accommodate a higher level of traffic.

Multiple VLANs can exist in each REP segment. In addition, any or all VLANs can span across multiple REP segments if necessary. The preferred approach, however, is to configure a VLAN in only one REP segment to avoid Layer 2 traffic crossing between segments. This use case has been validated for Layer 2 traffic within a VLAN and for Layer 3 traffic between VLANs in the same ring and between VLANs in different rings.

Differences between REP and DLR Protocols

  • REP is a Cisco proprietary protocol. DLR is managed by ODVA for EtherNet/IP.
  • REP can support multiple VLANs in a single ring. DLR only allows for one VLAN in a single ring.
  • REP can only have Cisco and Stratix industrial Ethernet switches (IES) as ring nodes. DLR can have both Stratix switches (5700, 5400, 5800) and individual devices intermixed as ring node members.
  • REP can support dual-media rings. DLR can only support single-media ring topologies.
  • REP can support up to 50 IES in the ring. DLR can support up to 24 IES in the ring.
  • REP does not require a ring supervisor to manage the ring. DLR requires that one ring node member be identified as the ring supervisor.
  • REP has a longer convergence time to detect and correct a fault than DLR. DLR offers the fastest convergence time of all of the resiliency protocols.

REP Resources:

CPwE Resources:

Stratix Resources:

Looking for more information?
Contact your local automation specialist or account manager at The Reynolds Company to discuss your automation applications. Find your specialist.