IMD SolutionComputing LibraryShop main store

Comparison

Move DeltaV Continuous Historian data into a PI system

Edge computing and IIoT Published 9 min read

Short answer

When PI points exist, collect DeltaV live values with OPC DA and recover history with OPC HDA; when points and AF structure are missing, let the DeltaV Smart Connector create them from EntHist.xml. Add PI-to-PI only for a second office-LAN server. Prove buffering, service identity and archive coverage by test before the bulk load.

On this page
  1. Which connection moves DeltaV Continuous Historian data into PI?
  2. When should you use the DeltaV Smart Connector instead of the OPC interfaces?
  3. Why does a PI point update but its trend stay empty?
  4. What do the 7000 IO and 5500 IO capacities actually decide?
  5. Do you need PI-to-PI between the DMZ server and the office LAN?
  6. Where do the interfaces run and what must buffering survive?
  7. How do you commission DA collection and HDA backfill before the bulk load?
  8. Do you keep the DeltaV Continuous Historian after migration?
  9. Questions asked

Run OPC DA for live values and OPC HDA for historical recovery when the PI points already exist and DeltaV's OPC servers are configured; use the DeltaV Smart Connector when the points and the AF structure still have to be created; add PI-to-PI only when a second PI server on the office LAN must hold its own copy. Those three decisions settle the architecture for a DeltaV Continuous Historian installation feeding a PI system. Where the interfaces run, what buffering must survive, what the 7000 IO and 5500 IO capacities decide, and which tests gate the bulk history load all follow from them.

Which connection moves DeltaV Continuous Historian data into PI? #

If the PI points exist on the target server and DeltaV's OPC servers are configured, connect with the OPC DA interface for live values plus OPC HDA for historical recovery; choose the DeltaV Smart Connector when the points and their AF structure must be created.

The four options an architecture drawing usually shows are not four ways to do the same job. Each fills a different gap, and the first three combine rather than compete.

Connection Best fit What it covers Verify before the bulk load
OPC DA interface PI points exist and the DeltaV OPC DA server is configured Current values and quality into the existing points Item mappings, service identity, buffering
OPC HDA interface Continuous Historian archives must reach PI Historical records on request Archive coverage, item mapping, time ranges
DeltaV Smart Connector PI points and AF structure still missing Automatic point and attribute provisioning from EntHist.xml Installed build, AF client version, generated attributes
PI-to-PI A second PI server on the office LAN holds its own copy Selected data copied between two PI servers Destination mappings, permissions, second-hop recovery
Direct DeltaV-to-office PI The approved network design permits the path Reaches the office PI without a DMZ server Control-network boundaries, firewall policy, outage behavior

Rows one and two are the default for this installation, which already runs a Continuous Historian and a DeltaV OPC DA server: DA writes current values and quality into points that exist, HDA reads archive records on request. Row three earns its license when provisioning, not collection, is the bottleneck. Rows four and five are architecture decisions — a copy between servers, or a path that skips the DMZ server and moves the boundary problem onto the firewall policy.

When should you use the DeltaV Smart Connector instead of the OPC interfaces? #

Use the Smart Connector when the PI server has no points yet: it reads the enterprise historian configuration file EntHist.xml and creates and configures the PI points and attributes for the DeltaV parameters automatically, which manual interface mapping would otherwise do tag by tag.

EntHist.xml is the enterprise historian configuration file. The connector reads that file to create the PI tags for the DeltaV parameters and to configure OPC DA collection against them, so provisioning scales with the DeltaV configuration rather than with hand-built tag lists. It provisions every point needed for time-series collection, not only the ones somebody happens to trend. What it does not do is prove history: after provisioning, archive coverage, HDA retrieval and buffering still need the same tests as the interface path.

Two checks come before the installer. First, identify the installed product: vendor documentation uses Smart Connector and Asset Connector names for closely related provisioning roles, so match the manual to the installed build instead of assuming. Second, check the AF client version: the Smart Connector 3.0.0.169 installer stops when it cannot detect PI Analysis Framework Client version 2.1.0.3522 or later, so an out-of-date AF client blocks installation before a single point is created.

Why does a PI point update but its trend stay empty? #

Yes — OPC DA carries only current values and quality, so a point can look healthy from the day collection starts while its trend stays empty for everything before that day; only OPC HDA reads the archive records the Continuous Historian already holds.

The split is deliberate: the OPC Foundation's Data Access specification carries present-time values, and its Historical Data Access specification carries reads from an archive. A point commissioned on Tuesday therefore trends from Tuesday onward, and nothing in a healthy DA connection backfills the days before it. Historical recovery needs four things at once: an accessible archive, a valid historical item mapping, an explicit time range, and a destination point that accepts the returned records — and a green DA status proves none of the four.

Diagnose an empty trend in three steps. If the point holds recent records but a display shows none, fix the display's point reference or time range. If the archive holds nothing recent, trace the collection interface: its source item, the quality it reports, and the server access the service identity actually has. If recent records exist and earlier ones do not, the gap is the HDA retrieval path or the archive coverage — the same discipline as backfilling PI from an existing plant historian: prove one bounded window, compare the returned records against the DeltaV source, then widen the range.

What do the 7000 IO and 5500 IO capacities actually decide? #

They are two separate stated capacities, so they decide scope only after one required signal list is reconciled against both: a 6,200-signal list fits the 7000 IO historian with 800 IO spare but leaves 700 signals without a live path on a 5500 IO OPC DA server.

An IO count is a capacity statement, and two of them neither add nor imply each other. The Continuous Historian's 7000 IO sizes the DeltaV-side archive; the OPC DA server's 5500 IO sizes its configured live exposure. Neither number says whether a given signal has a record, a DA item, or both — only a signal-by-signal comparison of identifiers, archive inclusion and live exposure does. The historical server's own capacity is a third number that this installation's summary does not state, and it needs the same reconciliation.

Worked example, assuming one IO per signal and a migration scope of 6,200 required signals:

  • Continuous Historian: 6,200 ≤ 7000 IO, so the archive side holds with 800 IO of headroom.
  • OPC DA server: 6,200 − 5500 IO = 700 signals beyond the server's configured exposure.
  • Decision, per signal among those 700: widen the DA server's exposure where a live value is required, keep it history-only where the archive record is enough, or drop it from scope. Only the first outcome needs a DeltaV configuration change before migration; the other two need none.

Do you need PI-to-PI between the DMZ server and the office LAN? #

No — one PI server in the DMZ that office clients read directly is a single-server design, so PI-to-PI earns its place only when a second PI server on the office LAN must hold its own copy of the data.

Count PI servers before designing transfers. With one server in the DMZ, collection lands once and office clients read across the approved path; adding PI-to-PI brings a second hop with its own destination mappings, transfer permissions and recovery behavior, and buys nothing the single server does not already have to provide. The second server is the requirement that justifies those costs: reporting load isolation on the office side, a different retention policy, or an office application that must own its server.

When that second server exists, treat the copy as a transfer between historians, much like moving PI historian data to another PI server: select the data, map destination points, prove recovery at the second hop after a deliberate break. PI-to-PI is a copy between PI servers, never a substitute for getting collection right at the first hop.

Where do the interfaces run and what must buffering survive? #

Run the DA and HDA interfaces on the DeltaV application stations, enable buffering on each instance, and validate both under the service identity that will actually run collection, because OPC DA rides COM/DCOM and an interactive login succeeding proves nothing about that identity.

Placement follows the data path: the interfaces read the DeltaV OPC servers, so they belong on the application stations next to them, subject to the supported software arrangement for the installed DeltaV release and the approved network design. From there they write to the PI server in the DMZ across the firewall — the link that drops during maintenance windows, certificate replacements and policy changes.

Buffering decides whether those drops cost data. The decision: buffering enabled on every interface instance, the buffer store on the interface station's own disk, with the store's location and recovery diagnostics taken from the interface's own buffering configuration rather than a checked box in a dialog. Then prove it: block the path to the DMZ server for a fixed commissioning window — one hour is a practical test — and confirm afterwards that every event arrived with its original timestamp and no gap. A buffer that has never survived a deliberate outage is a hope, not a feature.

Access follows identity, not the LAN: validate server access under the interface's service identity, since a process's run identity decides OPC access even when desktop clients work — the same rule that governs OPC browsing, where OPC browsing identity is the path, not the LAN. One safety line: never open unrestricted DCOM or firewall access between the control network and the DMZ; configure the documented, least-privilege access for the interface and keep historian traffic on the approved path.

How do you commission DA collection and HDA backfill before the bulk load? #

Commission in this order: confirm the destination, build a representative signal list, provision points, prove live DA collection, prove HDA on one bounded interval, test buffering with a deliberate outage, and settle overlapping writes before any bulk history load.

  1. Confirm the destination: collection targets the DMZ PI server alone, or an office server also holds a copy — this decision fixes whether PI-to-PI has a role at all.
  2. Build a representative signal list with known live changes and known historical records, recording each DeltaV source identifier beside its intended PI destination.
  3. Provision the points: map existing PI points to their source items, or validate connector-generated attributes and AF structure on a sample before provisioning scales.
  4. Validate access as the service identity that will run collection, not under an interactive login.
  5. Commission live DA first: values, quality, timestamps and destination writes on the representative list, before any historical transfer starts.
  6. Commission HDA on one bounded interval with known source records, compare the returned records against the DeltaV history, and widen the range only after the comparison passes.
  7. Test buffering with the deliberate outage window described above and confirm full recovery.
  8. Settle overlapping writes: define how live collection and historical recovery handle overlapping timestamps on the same point, and validate the behavior on test points before bulk migration.

Do you keep the DeltaV Continuous Historian after migration? #

Keep it — DeltaV Insight and DeltaV Reporter draw on the Continuous Historian, so a plant that plans to use either keeps it; filling PI is a parallel integration task, and nothing in this migration calls for retiring the source historian.

Collection reads what the DeltaV side already produces, so the design leaves DeltaV configuration untouched apart from any DA exposure widened during the capacity reconciliation. Insight's configuration, Reporter's reports and their data-source bindings do not migrate with the data — check each against its own documentation before anyone repoints it. Running old and new historians side by side through the transition is the normal pattern, and running two Plant SCADA Historian connectors during migration shows the same parallel-run discipline worked out for another historian pair.

Questions asked

Does the DeltaV Smart Connector license depend on historian size?

No. The vendor's FAQ for the embedded Enterprise PI historian states the Smart Connector license carries no size component, so the same license applies regardless of the embedded Enterprise PI historian's size. Licensing and sizing stay separate checks: the PI server's own sizing and the interface point counts follow their own documentation.

Does opening a trend in DeltaV create the matching PI point?

No. Point provisioning comes from the enterprise historian configuration file EntHist.xml when the connector runs, or from manual item mapping when interfaces are configured by hand. Browsing to a parameter or opening its trend in DeltaV changes nothing on the PI server, so a missing point is never fixed by looking at it.

What happens to live values when an outage outlasts the buffer?

Events the buffer cannot hold are lost on the DA path, but they are not necessarily lost overall: the Continuous Historian archive retains what it recorded while the path was down, and OPC HDA recovery can read that interval into PI afterwards. This is why archive coverage is checked per signal before the migration, not after the first incident.

Do office PI clients need collection software of their own?

No. With a single PI server in the DMZ, office PI clients read that server across the approved network path and run no collection components. The engineering work on the office side is the firewall policy, the approved path and client connectivity; the collection interfaces, their buffering and their service identities all stay on the DeltaV application stations.

Keep reading

  • Explainer

    Do you need a historian or just an OPC tag server?

    No. An OPC tag server (OPC DA) supplies the current value, quality and timestamp of each item and stores nothing, so monthly figures need a historian or a SQL logger collecting from that same server. Keep a reporting client like ActiveFactory only if you also run the historian it expects.

    Edge computing and IIoT

  • Troubleshooting

    Why Power BI cannot connect to FactoryTalk Historian

    Power BI has no native FactoryTalk Historian connector because the PI-based archive is a proprietary time-series store, not SQL tables. Reach it through the licensed 64-bit OLE DB or ODBC client provider installed on the Power BI machine, aggregate on the server before import, or stage the data through an Excel PI DataLink workbook when no provider is licensed.

    Edge computing and IIoT

  • How-to

    Backfilling PI from an existing plant historian

    Split the historian-to-PI backfill into four independent stages — extract, intermediate store, external compression, destination load — and qualify throughput with a measured pilot. Rows divided by elapsed seconds, not tag count, sets the schedule: 12,000 tags from 2020 at 500 rows/s is about 203 days. Check the native Historian 2023 replication route before writing pipeline code.

    Edge computing and IIoT