C135 PCS 7 (PCS7)
6C
6 Cielo Ciencia y Tecnología
·Publicada en C135 PCS 7 (PCS7)

Resolving SFCache Flooding and Missing Archive Tags in SIMATIC Process Historian

Resolving SFCache Flooding and Missing Archive Tags in SIMATIC Process Historian

Massive Store and Forward Cache (SFCache) accumulation (up to 90 GB) and subsequent missing archive tags in SIMATIC Process Historian (PH) occur when OS project downloads alter local project GUIDs, creating duplicate 'Data Sources' in the PH Management Console. When redundant servers failover under duplicate registration, both servers buffer unacknowledged data into local disk cache. While merging data sources in PH reconciles alarm logging, Tag Logging tag IDs become desynchronized. Executing a full OS compilation with memory reset, re-registering the OS station in PH, and resetting the runtime cache restores continuous tag archiving.

Table of Contents

Process Historian Store and Forward (SFCache) Architecture

In SIMATIC PCS 7 enterprise architectures, the SIMATIC Process Historian (PH) serves as the central long-term archive repository for plant-wide WinCC OS Servers.

To ensure zero data loss during network interruptions or database maintenance, each OS Server employs the Store and Forward Cache (SFCache) mechanism:
* The local WinCC service CCProcessHistorian.exe continuously monitors TCP connectivity to the central PH server.
* Under normal conditions, process values and alarms stream directly from WinCC Tag Logging into the PH Microsoft SQL Server database.
* If communication is interrupted, or if the PH server throttles ingestion due to database locks, the OS Server redirects incoming archive transactions into local binary cache files stored in the directory C:\Program Files (x86)\Siemens\WinCC\Diagnostics\SFCache.
* Once connection and registration are re-established, the OS Server flushes buffered records chronologically to the central historian.

Root Causes of Duplicate Data Sources and Cache Accumulation

When redundant OS Servers generate dozens of gigabytes in SFCache and drop process tags following a failover, the fault originates from project identification mismatches:

  1. Duplicate Data Source Registration: In Process Historian, each OS Server is identified by a unique combination of Computer Name, Project Name, and Project GUID. If an engineer performs a complete OS download from the ES without preserving the project container, WinCC generates a new internal project GUID. PH interprets this as a brand-new, unrelated OS server, creating a duplicate "Data Source" entry in the PH Management Console.
  2. Failover Cache Avalanche: When Server 1 switches from master to standby, the newly active Server 2 attempts to write archive records under the new GUID, which PH rejects or queues. Both servers begin buffering every single historical sample into their local SFCache directories, rapidly consuming 50 to 100 GB of local hard drive space.
  3. Tag ID Desynchronization After Merge: Using the PH Console to "Merge" duplicate data sources reconciles the station entity, but Tag Logging tags rely on internal SQL integer IDs (ArchiveTagID). If the tag dictionary mapping between WinCC's CCAlgRtServer and PH's Runtime.ProcessValueArchive is broken, alarm messages transfer successfully while analog tag streams are silently dropped.

Step-by-Step Restoration and Database Reconciliation Workflow

To clear the massive cache safely and restore live tag streaming to Process Historian:

  1. Reconcile Data Sources in Process Historian:
    * On the Process Historian server, open Process Historian Management Console.
    * Navigate to Management -> Data Sources.
    * Identify duplicate entries for the OS servers. Select the duplicate and click Merge.
    * Wait approximately 15 minutes for the SQL Server stored procedures to re-index historical partitions. Restart the Process Historian Server Service via the console.
  2. Flush and Normalize SFCache on OS Servers:
    * Monitor the SFCache folder on both OS servers. Allow the background transfer service to drain the buffered files into PH.
    * Ensure that hard disk capacity on the OS servers does not reach 100%; expand disk volumes dynamically if necessary to prevent SQL Server crash.
  3. Synchronize OS Tag Logging Dictionary:
    * Once SFCache drops to zero bytes, open SIMATIC Manager on the ES.
    * Right-click the OS project, select Compile, choose Complete compilation, and check With memory reset.
    * Execute an OS download to both redundant servers (Standby first, then Master).
  4. Re-register OS Stations in PH:
    * On each OS Server, verify that the configuration file C:\Program Files (x86)\Siemens\WinCC\bin\CCProcessHistorian.ini contains the exact, correct PH server computer name and IP.
    * Restart WinCC Runtime to force a fresh handshaking session with Process Historian.

Verification

  • In Process Historian Management Console, verify that System Status displays a green check mark with State: Active.
  • Check the Data Sources view. Confirm that both OS Server A and Server B show Status: Connected and Archive State: Streaming.
  • Open Process Historian Information Server (IS) or WinCC OnlineTrendControl querying the long-term archive.
  • Confirm that trend curves for analog process tags populate with current, live timestamps without flatlining or data gaps.

Limitations and Operational Precautions

  • Never manually delete binary files inside the SFCache folder unless instructed by Siemens Technical Support. Deleting cache files permanently destroys historical process data captured during the outage.
  • Disabling SFCache is strictly prohibited in regulated plants (e.g., FDA 21 CFR Part 11 compliant pharmaceutical facilities).
  • Sizing OS Server system drives must account for potential SFCache accumulation. Always reserve at least 150 GB of dedicated free disk space on OS Server data partitions.

Engineer FAQ

  • What does the 'Merge' function in Process Historian actually do?
    Merge associates historical archive records previously assigned to an orphaned or outdated OS Project GUID with the newly active OS Project GUID, uniting historical trends under a single station name.
  • Why did alarm messages work while process tags failed?
    Alarm Logging and Tag Logging utilize separate transport pipelines inside CCProcessHistorian.exe. Alarms use message numbers as universal keys, whereas Tag Logging relies on internal numeric ArchiveTagIDs that desynchronize more easily during uncoordinated project downloads.
  • How fast does SFCache drain over the network?
    Depending on network bandwidth and PH database write load, SFCache typically drains at rates between 1 GB and 5 GB per hour over a dedicated gigabit Terminal Bus.

Technical References

  • SIMATIC Process Historian and Information Server System Manual (A5E41356290-AA).
  • Siemens Industry Online Support Entry ID 109801081: Process Historian Store and Forward Architecture and Sizing.
  • WinCC V7/V8 Long-Term Archiving with Microsoft SQL Server Enterprise.

Related Articles

Related Tags

PCS 7, WinCC, Process Historian, SFCache, Tag Logging, Alarm Logging, SQL Server, Redundancy

20 respuestas