Honeywell Uniformance PHD  ·  Migration & archive recovery

Migrate your Honeywell PHD history. Even after the server is gone.

Archivian is an independent, done-for-you service. We read your PHD archive files straight off disk and load the history into PI, InfluxDB, Snowflake or Parquet, named and timestamped, by time range. It is the only approach that still works with no live server, no PHD license and no OPC gateway.

Runs in place on your infrastructure, or on a copy you send. Independent service, not affiliated with Honeywell.

The problem

Getting history out of PHD usually means keeping PHD alive.

Every sanctioned way to get history out of Honeywell PHD assumes a live, licensed, running server. So a migration means keeping the old system alive long after you wanted it gone, and even then the native tools can miss data you can't pull back by time range. If the server is already retired, the history is simply stranded.

You have to keep the old server alive

Every bulk export route runs through the PHD API or an OPC gateway, so you keep paying to license and run a system you are trying to switch off, just to read data you already own.

Cutovers go live with gaps

The standard PI Interface for PHD is built for live collection, not arbitrary-range historical backfill. In the migrations we see, that leaves holes in the record that only surface weeks after go-live.

No server means no data at all

Once a system is decommissioned or divested, you are left with disks full of .dat/.idx files and nothing that can read them. The native "RECOVER" path still needs PHD installed.

How it works

Files in. Clean, named, time-ranged data out.

It is a done-for-you service, and your data does not have to leave the plant. We can run the extraction in place on your infrastructure, on-site or over a supervised remote session, or work from a copy you send. Either way you do not install anything or touch your production system.

We come to the data, or you send it

We read your PHD archive files (SCAN*, MANV*, CHAR* .dat/.idx) and, if you have it, the PIMS catalog dump. We work in place on your infrastructure so nothing leaves your network, air-gapped if needed, or from a copy you send under NDA for decommissioned and divested systems.

We decode and validate

We read the archives directly at the byte level, reconstruct timestamps and quality, reattach tag names and units from the catalog, and run a reconciliation sample against your reference data.

You get your data

Delivered as CSV or Parquet, or loaded straight into AVEVA PI, InfluxDB, Snowflake or your warehouse. Complete, by time range, ready to use.

Speed

How fast is reading PHD archive files vs. an OPC gateway?

Reading Honeywell PHD archive files directly off disk is far faster than an OPC HDA gateway. On real production data we measured about 436,000 valid samples per second on a single thread, against the ~1 to 2 MB/s typical of a throttled OPC route. That puts a 10 TB archive at days on one thread, against the ~58 days an OPC gateway would take, and our parallel reader brings it down to hours. OPC HDA is a per-tag request and response protocol that has to throttle so it does not slow the live production server. Reading offline copies of the raw files has no such contention, and the work splits cleanly across cores because each archive segment is independent.

Archive sizeOPC gatewayMeasured, single threadParallel reader
10 TB~58 days~5.3 days~6 hours (projected)
50 TB~290 days~26 days~1-2 days (projected)
100 TB~1.6 years~53 days~2-3 days (projected)

Time to extract a full archive. Scroll the table sideways to see every column. The OPC column uses the faster 2 MB/s end of the 1 to 2 MB/s range reported in independent historian-integration analyses; at 1 MB/s it is roughly twice as long. The "Measured, single thread" column is real, timed on production PHD data on one core. The parallel-reader column is projected from sustained SSD read throughput (~500 MB/s) and not yet timed end to end. We benchmark on your files before any quote.

~436,000
valid samples per second, measured on a single thread on real production data, after discarding empty and uninitialized records. At the average record size in those archives that is roughly 20 MB/s sustained, well past a throttled OPC gateway, and it scales across cores and files.

Because we read offline copies, extraction has zero impact on your live system, so it can run flat out instead of working around production load.

Why we’re different

How does reading PHD archive files compare to OPC, the PI Interface and native tools?

Every other route requires a live, licensed PHD instance. Here is how the options stack up.

Capability Archivian OPC HDA gateway PI Interface for PHD Native RECOVER / PHDMAN 3rd-party connectors1
Works with no running PHD serverYesNoNoNoNo
Works with no PHD licenseYesNoNoNoNo
Recovers decommissioned / archived files onlyYesNoNoPartialNo
History recovery by arbitrary time rangeYesYesLimitedPartialYes
Zero impact on production systemYesNoNoNoNo
Extraction speed (10 TB)Hours~58 daysSlowVariesSlow
Parallel / linear scalingYesNoNoNoLimited
Reattaches tag names, units & rangesYesSeparate exportPartialPartialYes
Handles SCAN, MANV & CHAR familiesYesYesPartialYesPartial
Output: CSV / Parquet / PI / InfluxDB / SnowflakeAnyVia loaderPI onlyRelationalPlatform-specific
Infrastructure to set upMinimal; in place or copyOPC + PHD clonePI interface nodePHD installConnector + live link
Fully done-for-youYesDIYDIYDIYVaries

Scroll the table sideways to see every column. Only Archivian works with no running PHD server or license; the alternatives require a live, licensed PHD instance. 1 Live connectors such as Seeq, Cognite or PARCview, which read PHD through its API or OPC. Comparison reflects our understanding of each option's standard configuration from vendor documentation and field experience; capabilities vary by version and setup.

The only option that still works once PHD is gone.

Point us at one archive segment, in place or as a copy, and we return a free, reconciled sample from your own data, so you can check accuracy before you commit to anything.

Get a free migration assessment

Where it fits

Four common situations where this fits.

Historian migration

Moving to PI, InfluxDB, Snowflake or Parquet. Backfill history by time range, as complete as the surviving files allow, without keeping PHD alive through the cutover.

Decommissioning

Site closing or already closed. Meet your data-retention obligations by recovering the full record from the archive files alone, with no PHD to stand back up.

M&A and handover

You received the data files in a divestiture but not the system. We make them readable, named, and verifiable.

Forensic and legal

A documented, quality-coded extraction from the archive files, reconciled against your reference data, to support investigations, audits, or regulatory requests.

Why trust the output

Reverse-engineered to the byte, and checked against reality.

Empirically validated

The format specification is verified against real production PHD data. Every structural check passes across SCAN, MANV/RECV and CHAR archives (21 of 21 on a SCAN archive), covering header layout, timestamp decoding, quality-code mapping and index alignment. Samples are filtered by PHD quality codes so stale, uninitialized memory never reaches your output.

All three archive families

Continuous samples (SCAN), single-value events (MANV/RECV), and operator-mode strings (CHAR) are auto-detected and decoded, including the C-ISAM index for direct time-range lookup.

Real tag identities

We recover the PIMS tag catalog, including tag names, descriptions, engineering units, ranges and source-DCS mapping, so you get meaningful series instead of anonymous tag numbers.

Reconciliation first

Before any full extraction we deliver a free sample and reconcile it against your reference data, so accuracy is proven on your archives before you commit.

Delivered into

CSV Apache Parquet AVEVA PI System InfluxDB Snowflake Amazon S3 / object storage Your data warehouse

Reference

What are Honeywell PHD SCAN, MANV/RECV and CHAR archive files?

A Honeywell Uniformance PHD historian stores its history on disk as three archive families, each a pair of .dat (data) and .idx (index) files, with the tag catalog held separately in the PIMS Oracle database.

SCAN

Continuous, periodically sampled tags. Each .dat record packs hundreds of fixed-width sample slots with values and PHD quality codes; the matching .idx is a C-ISAM B-tree index used for time-range lookup.

MANV / RECV

Single-value events: manually entered or recovered points, written one record at a time rather than on a fixed scan interval.

CHAR

Operator-entry string data, such as mode or status text, that does not fit the numeric SCAN model.

Archivian reads all three families and the PIMS catalog directly off disk, so the history can be recovered from the files alone, with no running PHD server, PHD license or OPC gateway.

These same SCAN, MANV/RECV and CHAR archives appear across the Uniformance PHD release line, from the R2xx and R3xx series through R400 and R430, including PHD R210, R215, R310, R320, R321 and R340. We confirm the exact format on your specific archives with a free reconciliation sample before any full run, whether the data sits on a live server or a decommissioned one.

FAQ

Common questions.

Can you migrate or recover PHD data with no running server or license?
Yes. We read the raw archive files (.dat / .idx) directly off disk. We don’t call the PHD API, OPC, or any Honeywell software, so a live, licensed PHD instance isn’t required. This is the case we’re built for: decommissioned, divested or archived systems where only the files remain.
Is it legal to recover our own PHD archive data this way?
You own your historical process data and the archive files. We read files you provide and never touch Honeywell software or licenses. Archivian is an independent service and is not affiliated with or endorsed by Honeywell.
How accurate is the recovered data?
We have validated the file format against real production PHD data. Every structural check passes across SCAN, MANV/RECV and CHAR archives (21 of 21 on a SCAN archive), and every sample is quality-filtered. Before any full run we provide a free reconciliation sample so you can compare our output directly against your reference data.
How much faster is this than an OPC HDA gateway?
An OPC gateway is a throttled per-tag protocol that has to protect the live server, and independent integration analyses commonly report figures around 1 to 2 MB/s. Reading offline file copies has no such contention and runs in parallel. We measure roughly 436,000 valid samples per second on a single thread, and we benchmark on your own files before quoting any timeline.
We already work with an SI or connector vendor. Why you?
We are not replacing your integrator. We do the one thing their toolchain usually cannot: read the PHD archive with no live server, then hand over clean, named, time-ranged data they can load. Many engagements are exactly this split, where we recover and reconcile the history and your SI takes it into the target system.
Which output formats can you deliver?
CSV, Parquet, or a direct load into AVEVA PI System, InfluxDB, Snowflake, or object storage such as Amazon S3. Tag names, units, descriptions and engineering ranges are reattached from the PIMS catalog.
Do our PHD files have to leave our network?
No. By default we run the extraction in place on your infrastructure, on-site or over a supervised remote session into a machine you control, so your archive files and history never leave your network. Air-gapped sites are supported. If you would rather hand off a copy, common for decommissioned or divested systems where the plant network is gone, we work under NDA.
How do you handle data security?
Our default is to process in place on your infrastructure so nothing leaves your network. We work under NDA, support on-site or air-gapped processing for sensitive data, and when you do send a copy we return your data and destroy working copies on completion.

Get started

Get a free migration assessment.

Tell us what you’re working with and we’ll tell you what’s recoverable, how fast, and in what format, with no obligation. The more detail you give, the sharper the answer. Engagements are fixed-fee per archive, quoted up front after a free reconciliation sample, so there are no surprises.

We reply within one business day. Covered by NDA on request. We never share your details.