Guide · Healthcare imaging

PACS and VNA migration at scale.

Why imaging-archive migrations take months, how to separate moving the bytes from reconciling the DICOM data, how long the bulk copy really takes, and a checklist for a faster, verifiable cutover.

Updated November 18, 2026 · by the Zettar team
The short answer. A PACS or VNA migration is two jobs that are usually run as one: moving petabytes of image data, and reconciling the DICOM records that describe it. Run them as parallel layers. A DICOM-aware migration tool handles reconciliation, tag correction, and indexing; a high-speed data-movement layer copies the underlying files or objects storage to storage at line rate, with every file checksummed. At line rate, a 100 Gbps link moves about a petabyte a day — so the bulk copy stops being the long pole, and every day of running two archives side by side costs less.

Why do imaging migrations take so long?

  • Volume and file count. Archives hold years of studies — hundreds of terabytes to petabytes — spread across very large numbers of individual image files. Tools that move one file or one stream at a time crawl through them.
  • Export through the old system. Many migrations pull every study back out through the legacy archive’s DICOM interface, which was built to serve reading workstations, not to empty itself in bulk.
  • Data reconciliation. Patient identifiers, accession numbers, and DICOM tags must be matched, corrected, and validated against the new system — work that runs study by study.
  • Clinical uptime. Radiology cannot stop reading, so migration traffic has to coexist with production — often only in off-hours windows.
  • Security and integrity. Protected health information has to stay encrypted and provably intact the whole way.

Separate moving the bytes from reconciling the data

The two jobs have different bottlenecks, so give each the right tool:

LayerWhat it doesWhat sets its speed
DICOM-aware migrationReconciles patient and study records, corrects tags, indexes studies into the new PACS or VNAData quality and the rules your team applies
High-speed data movementCopies the underlying image files or objects between storage systems, sites, or the cloud, with integrity checksThroughput of the whole path — storage, hosts, software, network

Where the source archive’s storage can be read directly as files or objects, the bulk copy can run at line rate while reconciliation proceeds alongside it. Where a legacy archive only exposes its data through DICOM export, the copy speed is bounded by that interface — worth knowing before you set a cutover date.

How long does the bulk copy take?

Archive size10 Gbps25 Gbps100 Gbps
250 TB2.6 days25 h6.2 h
1 PB10 days4.1 days25 h
3 PB31 days12 days3.1 days

At line rate — about 90% of the link. Formula: hours = data (TB) ÷ (link Gbps × 0.9 × 0.45 TB per hour per Gbps). Tools that realize a fraction of the link take proportionally longer — at 20% utilization, a petabyte on 100 Gbps takes about 4.6 days instead of one.

What does a long cutover cost?

Until the last study lands and you cut over, you run both archives: storage, licenses, support, and staff on two systems at once. Every week the migration window stretches is another week of that overlap. Put a number on it with the migration-window calculator.

A checklist for a faster, verifiable migration

  • Inventory first. Size the archive, count the files, and confirm whether the source storage can be read directly or only through DICOM export.
  • Split the work. Run reconciliation and the bulk copy as parallel tracks, not one serial pipeline.
  • Move in parallel, at line rate. Keep many files in flight across many streams and nodes, so file count does not cap throughput.
  • Verify every file. End-to-end checksums on every file or object, so “we think it copied” becomes “we can prove it did.”
  • Encrypt in transit and stay in control. Keep data on your own infrastructure, encrypted, inside the controls your security team already enforces.
  • Keep native formats. Land studies exactly as they were, with no proprietary wrapper to unwind later.
  • Protect clinical traffic. Cap migration bandwidth during reading hours and open it up off-hours.
  • Sync to cutover. Keep copying new and changed studies until the switch, so nothing is left behind.

Security and compliance

Zettar does not retain your data — it stages it only briefly in the appliance’s NVMe burst buffer and moves it, encrypted with TLS and verified with unconditional end-to-end checksums, on your own infrastructure. It works through your identity provider and supports per-transfer bandwidth caps, so migration traffic stays within the policies you set. It fits within the HIPAA and security controls your organization already enforces; it is not itself a compliance certification. See security & trust.

Proof

In one large-scale file migration, a customer cut migration time by 80% with hands-off automation (case study). And with SLAC National Accelerator Laboratory and ESnet (U.S. DOE), Zettar moved 1 PB in 29 hours at 96% link utilization, encrypted and checksummed. The Zettar zx Appliance brings that bulk-copy layer turnkey — including a compact Mini for smaller sites.

Estimate your overlap cost → Healthcare & imaging →

FAQ

Common questions

How long does a PACS or VNA migration take?

The bulk copy depends on throughput: at line rate, a 100 Gbps link moves about a petabyte a day, and a 10 Gbps link about 97 TB a day. The overall project also depends on DICOM reconciliation and on how the legacy archive exposes its data, which is why running the two as parallel tracks shortens the timeline.

Why are imaging archive migrations so slow?

Archives hold hundreds of terabytes to petabytes across very large numbers of files; many migrations export every study back through the legacy archive's DICOM interface; patient and study records must be reconciled; and migration traffic has to coexist with clinical reading.

Does a high-speed data mover replace a DICOM migration tool?

No. A DICOM-aware migration tool reconciles records, corrects tags, and indexes studies into the new PACS or VNA. A high-speed data-movement layer copies the underlying files or objects at line rate with integrity checks. The two run side by side.

Can we migrate imaging archives to the cloud?

Yes — Zettar moves file and object data between sites and to the cloud at line rate, bypassing most of the performance penalty of standard cloud transfer paths, with every file encrypted in transit and checksummed.

Does it fit HIPAA requirements?

Zettar does not retain your data — it stages it only briefly in the appliance's NVMe burst buffer and moves it encrypted with TLS and verified with unconditional end-to-end checksums, on your own infrastructure, through your identity provider. It fits within the HIPAA and security controls your organization already enforces; it is not itself a compliance certification.

See it for yourself

Shorten your imaging migration.

See the appliance copy your archive at line rate — or talk through your migration plan with an engineer.