How to Design Fault-Tolerant Plant and Terminal Bus Architectures in SIMATIC PCS 7 - PRP, S7-REDCONNECT, and Teaming

In SIMATIC PCS 7, software-based failover between two independent Named Connections (such as ISO Transport and TCP/IP) on a single communications processor is unsupported because WinCC process tags are strictly bound to a single connection ID during compilation. To achieve high-availability industrial communication, redundancy must be implemented at the physical and protocol layers. On the Plant Bus, fault-tolerant S7 connections powered by SIMATIC NET S7-REDCONNECT provide 4-way redundancy across redundant S7-400H controllers and SCALANCE X optical rings. On the Terminal Bus, high availability is achieved using Parallel Redundancy Protocol (PRP) compliant switches or dual network interface cards configured with Intel Advanced Network Services (ANS) adapter teaming.

Industrial Ethernet Redundancy Architecture in SIMATIC PCS 7

In process control systems, network communications are divided into two segregated physical or logical domains:
* Plant Bus (Process Communication): Interconnects Automation Stations (AS S7-400 / CPU 410-5H) and Operator Station (OS) Servers. It carries real-time S7 communication, cyclic process data, alarms, and engineering downloads.
* Terminal Bus (Supervisory Communication): Interconnects OS Servers, OS Clients, Engineering Stations (ES), and Process Historians. It carries WinCC server-client synchronization, SQL database queries, and terminal desktop traffic.

Attempting to achieve redundancy by creating two named connections in WinCC (one ISO and one TCP) fails because WinCC tags compile against a single static connection. Redundancy must be engineered into the underlying network layers.

Plant Bus High Availability: S7-REDCONNECT and 4-Way Redundancy

For maximum availability on the Plant Bus, Siemens provides the S7-REDCONNECT communication package:

Hardware Topology:
Fault-tolerant controller: S7-400H (e.g., dual CPU 410-5H or CPU 417-4H).
Communications Processors: Two CP 443-1 modules (6GK7443-1EX30-0XE0), one in Rack 0 and one in Rack 1.
Operator Station: Industrial PC equipped with dual network adapters (CP 1623 / CP 1613 or standard industrial NICs running Softnet-S7).
Network Switches: Redundant SCALANCE X-300 / X-400 managed industrial switches arranged in dual optical rings.
4-Way Redundancy Mechanism:
S7-REDCONNECT establishes four concurrent communication paths behind a single logical Named Connection:
1. OS NIC 1 -> Switch Ring A -> CP 443-1 Rack 0
2. OS NIC 1 -> Switch Ring B -> CP 443-1 Rack 1
3. OS NIC 2 -> Switch Ring A -> CP 443-1 Rack 0
4. OS NIC 2 -> Switch Ring B -> CP 443-1 Rack 1
If any fiber link, switch, or CP fails, failover occurs in under 30 ms without tag loss.

Terminal Bus High Availability: Parallel Redundancy Protocol (PRP)

Starting with PCS 7 V8.1 and V8.2, Parallel Redundancy Protocol (PRP) according to IEC 62439-3 has become the preferred standard for Terminal Bus redundancy:

Operating Principle:
PRP operates by transmitting duplicate Ethernet frames simultaneously across two completely independent, unlinked networks: LAN A and LAN B.
The receiving node accepts the first arriving frame and discards the duplicate.
In the event of a cable break or switch failure on LAN A, the packet on LAN B is received with zero millisecond failover time (bumpless redundancy).
PRP Implementation Modes:
Hardware-Based (RedBox): Non-PRP nodes connect to SCALANCE X-204RNA Redundancy Boxes, which handle frame duplication and deduplication.
Software-Based (SIMATIC NET SOFTNET-IE RNA): Installed directly on IPCs, allowing standard dual-port network cards to participate natively in PRP networks without external RedBoxes.

Network Adapter Teaming (Intel ANS) on Operator Stations

Before PRP, terminal bus redundancy relied primarily on Network Adapter Teaming:

Intel Advanced Network Services (ANS):
Uses two dedicated Intel server Gigabit network adapters.
Configured in Windows Device Manager under Adapter Fault Tolerance (AFT) or Switch Fault Tolerance (SFT).
The teaming driver presents a single virtual IP address and MAC address to Windows and WinCC.
If the primary physical cable disconnects, the secondary adapter assumes traffic handling within 500 ms.
Operational Constraint: Network adapter teaming requires compatible hardware drivers; starting with Windows 10 and Windows Server 2016, Intel deprecated legacy ANS teaming in favor of native Windows Server NIC Teaming (LBFO) or PRP.

Comparison of Industrial Network Redundancy Mechanisms

The table below compares the high-availability networking solutions deployed across PCS 7 systems.

Redundancy MechanismTarget Bus DomainFailover TimeUnderlying ProtocolHardware / Software RequiredS7-REDCONNECTPlant Bus (AS - OS)< 30 msS7 Fault-TolerantS7-400H, dual CP 443-1, S7-REDCONNECT licensePRP (IEC 62439-3)Terminal Bus & Plant Bus0 ms (Bumpless)Dual-frame EthernetSCALANCE X-204RNA or SOFTNET-IE RNAMRP (IEC 62439-2)Plant Bus / PROFINET< 200 msMedia Redundancy RingSCALANCE X managed switches in ringIntel ANS TeamingTerminal Bus200 to 500 msEthernet AFT / SFTDual Intel server NICs, ANS driverDual Named ConnectionsOS Tag ManagementUnsupportedManual switchingNot possible; compiler binds to single connection

Verification

To confirm that network redundancy operates correctly:
1. On the OS Server, open WinCC Channel Diagnosis: verify that the S7-REDCONNECT connection state reports State: OK and active path displays Rack 0 or Rack 1.
2. Disconnect the primary Plant Bus fiber optic cable from SCALANCE Switch A: verify that WinCC faceplates remain active with zero # bad quality displays and failover logs are recorded in SIMATIC NET diagnostics.
3. If using PRP on the Terminal Bus, launch SOFTNET-IE RNA Diagnosis: confirm both LAN A and LAN B links report Link Active with 100% duplicate packet reception.
4. Unplug the LAN A Ethernet cable from the OS Server: verify that OS Client displays, alarm banners, and SQL database queries continue without delay or interruption.
5. Reconnect all physical cables: confirm that both redundant paths resume healthy standby states within 5 seconds.

Limitations and Operational Precautions

Never bridge LAN A and LAN B together with an Ethernet patch cable in a PRP network; interconnecting the two networks will create a broadcast storm that disables both segments.
S7-REDCONNECT requires dedicated licenses on all OS Servers communicating with redundant S7-400H controllers; running without licenses triggers recurring ALM violation pop-ups.
Intel ANS network teaming is strictly unsupported on Windows 10 client operating systems in modern PCS 7 releases; deploy PRP or standard managed switch rings instead.
Always validate final network switches, communication processors, and software licenses against the applicable Siemens catalog and authorized distributor quotation.

Engineer FAQ

Can I use PRP on the Plant Bus as well as the Terminal Bus?

Yes. SIMATIC PCS 7 V8.2 and V9.0 fully support PRP on both the Terminal Bus and the Plant Bus when using SCALANCE X-204RNA RedBoxes or modern CPU 410-5H PROFINET configurations.

Why did WinCC lose connection for 2 seconds during switch failover?

If the industrial switch ring uses Spanning Tree (RSTP) rather than Media Redundancy Protocol (MRP) or PRP, network reconfiguration takes 1 to 2 seconds, which exceeds the S7 communication watchdog threshold.

Does S7-REDCONNECT require CP 1623 hardware cards?

No. While CP 1623 provides high-performance hardware microprocessor offloading, S7-REDCONNECT can also run on standard commercial industrial network cards using the SOFTNET-S7 RedConnect VM license.