Diagnosing and Resolving Sporadic OS Client Response Delays in PCS 7

Sporadic 10-second response delays on SIMATIC PCS 7 OS Clients while redundant OS Servers update smoothly typically result from running WinCC Graphics Runtime directly on the OS Servers, Terminal Bus DNS name resolution delays, or saturated client script queues. Operating visualization screens on the OS Servers consumes local DCOM threads and SQL resources, starving client requests. To eliminate the delay, deactivate graphics runtime on dedicated OS Servers, populate static IP-to-hostname mappings in the Windows 'hosts' and 'lmhosts' files across all clients and servers, and use the 'apdiag' tool to detect and optimize scripts with excessive cyclic triggers.

Architectural Root Causes of OS Client Latency and 10-Second Freezes

In a multi-client SIMATIC PCS 7 architecture (e.g. redundant OS Servers serving 8 OS Clients), operators occasionally report intermittent lag where faceplates take 5 to 10 seconds to open, alarm acknowledgments freeze, or graphic values pause. Meanwhile, inspecting the OS Server reveals that local screens and process values are updating instantaneously.

A characteristic 10-second delay in Windows networking is the standard TCP/IP socket connection timeout and NetBIOS/DNS lookup retry period. When an OS Client attempts to acquire tag values or query an alarm segment from the server, if the primary communication path encounters a name resolution delay or if server worker threads are saturated, the client blocks until the internal DCOM timeout expires.

The Hazard of Running Graphics Runtime on Dedicated OS Servers

A frequent design error in PCS 7 deployments is treating dedicated OS Servers as local operator consoles by keeping WinCC Graphics Runtime active on the server console:

Resource Contention: The WinCC OS Server must manage cyclic AS communication over the Plant Bus, execute archive segment logging into Microsoft SQL Server, maintain redundancy synchronization with its partner server, and stream data to all 8 remote clients.
Thread Starvation: Executing complex PDL graphic screens with dynamic C-scripts, alarm banners, and trend curves on the server forces the graphic engine (pdlrt.exe) and script subsystem to compete directly with client DCOM request handlers. Under peak alarm or process bursts, client requests are deprioritized, creating severe lag on the plant floor.
Siemens Architectural Rule: Dedicated OS Servers must run headless or strictly in server mode without active operator screen navigation.

Terminal Bus Network Optimization and Name Resolution Tuning

To eliminate network-induced latency across the Terminal Bus:

Static Hosts and LMHosts Configuration: Windows DNS queries frequently stall when resolving local industrial machine names. On every OS Server and OS Client, configure static entries in:

C:\Windows\System32\drivers\etc\hosts and lmhosts

Map the exact computer name and Terminal Bus IP address for all servers, redundant partners, and engineering stations.
2. Network Card Power Management: Open Windows Device Manager on each OS Client, navigate to the Network Adapter properties, open the Power Management tab, and uncheck Allow the computer to turn off this device to save power.
3. Terminal Bus Adapter Priority: In Windows Advanced Network Settings, verify that the Terminal Bus network adapter is positioned at the top of the network binding order.

WinCC Script Queue Diagnostics Using APDiag

To diagnose whether client-side script congestion is causing the freeze:

On the lagging OS Client, open the command prompt or Run dialogue and execute: apdiag.
In the APDiag window, click Output Window to capture live diagnostic messages.
Click Info -> Count of functions in queue:
- A healthy system maintains 0 to 5 queued functions.
- If the queue count exceeds 50 or steadily climbs, scripts are blocking execution.
Check for cyclic triggers: Ensure graphic objects use Upon Change triggers linked to specific tags rather than high-frequency cyclic timers (e.g. 250ms or 500ms cyclic triggers).

OS Client Latency Diagnostic and Performance Matrix

The following table correlates client response delay symptoms with root causes and corrective measures:

Observed SymptomMeasured LatencyRoot Cause MechanismResponsible SubsystemCorrective ActionFaceplate opening delayExactly 10 to 15sDNS / NetBIOS hostname resolution timeoutWindows TCP/IP & Name ResolutionPopulate hosts and lmhosts filesSporadic value update freeze5 to 10s intermittentlyActive graphic runtime on OS ServerServer CPU & DCOM Thread PoolDeactivate local runtime on serverClient UI unresponsivenessContinuous lag (>10s)APDiag script queue overflow (>100 functions)WinCC Scripting (script.exe)Replace cyclic script triggers with Tag ChangeFirst click delay after idle5 to 10s after inactivityNetwork card Energy-Efficient Ethernet (EEE)NIC Power ManagementDisable NIC power savings in Device Manager

[!NOTE]
Commercial Accuracy Notice: Operating system client-server performance tuning is an architectural and network engineering > task. Deploying up to the maximum certified client count (e.g. up to 32 OS Clients on a redundant OS Server pair) is fully > supported under standard PCS 7 OS Client licensing (6ES7658-2FX58-0YH5) without supplementary server licenses.

Verification

Confirm that OS Client latency is permanently resolved:

Faceplate Response Benchmark: Open multiple complex motor and valve faceplates in rapid succession on an OS Client. Confirm that faceplates render within 0.5 to 1.0 second without hesitation.
APDiag Queue Stability: Check the APDiag function queue on the client during screen changes. Confirm that the queue count immediately clears to 0.
Server Resource Utilization: Monitor OS Server CPU and memory utilization in Windows Performance Monitor. Confirm that CPU usage remains stable below 30% under steady-state operation.

Limitations and Operational Precautions

Adhere to these operational rules for multi-client PCS 7 systems:

Never Navigate Screens on OS Servers: Train maintenance technicians never to use the physical OS Server console for daily process monitoring; provide dedicated OS Clients or Web Clients.
Terminal Bus Bandwidth: Maintain a dedicated Gigabit Ethernet network for the Terminal Bus; never mix plant bus traffic (AS communication) on the same physical switch without VLAN segmentation.
Synchronize System Time: Ensure all OS Servers and OS Clients synchronize their system clocks via a central NTP time master (e.g. SICLOCK TC400 or domain controller).

Engineer FAQ

Why do my OS Clients freeze for exactly 10 seconds?

A 10-second freeze is the classic Windows network timeout when attempting to resolve a hostname or connect to a DCOM socket. Adding entries to the Windows hosts file eliminates the DNS lookup delay.

Is it forbidden to view process screens on a PCS 7 OS Server?

While WinCC technically permits it, Siemens strongly recommends against running operational graphics on dedicated OS Servers in multi-client systems. Local graphics consume server resources needed to service client requests.

How do I access the APDiag diagnostic tool?

Open the Windows Run dialog (Win+R), type apdiag, and press Enter. The tool is installed automatically with WinCC under the Siemens bin directory.

Can network card power savings really cause 10-second delays?

Yes. Modern Gigabit network adapters enter low-power sleep states when traffic drops. Waking the physical interface and re-negotiating TCP flow control can induce 5 to 10 seconds of latency.