Home / Products / PDI Data Quality for Mainframe
New Product · Adabas · Datacom · Db2 for i · Db2 for z/OS · IDMS · IMS

PDI Data Quality for Mainframe

Profile 100 million mainframe rows in a 30-minute window — from one Linux server, one config file, zero host installs. The mainframe does the counting; you get an audit-ready DAMA scorecard.

6
mainframe datastores, one identical scorecard
Db2 z/OS · Db2 i · Datacom · IDMS · Adabas · IMS
100M
row design budget per 30-minute profiling window
PDI internal benchmark · 8-core, 256 GB Azure instance · 2 TB SSD
800K
rows/sec rule-scoring per CPU core
Reproducible via the included make smoke harness · 8-core, 256 GB Azure instance · 2 TB SSD
0
code changes to add rules, columns, or datastores
INI-declared configuration
The problem

The most critical data is the data quality tools can't reach.

The oldest, most business-critical data in the enterprise lives on the mainframe. Moving 100 million rows off-host just to count nulls is slow, expensive, and risky. The solution inverts that: the counting runs where the data lives.

Pushdown-first

Aggregates run on the host

Row counts, nulls, distincts, and length bounds are computed by the engine that owns the data, via one aggregate statement. One result row crosses the wire — not 100 million.

Minimal movement

Streams only what must move

Only the profiled columns travel, in 10,000-row array fetches, under read-only/UR isolation — no locks on production workloads.

Pattern scoring

High-performance linear-time validation

Emails (RFC 5322 subset), phones (E.164 + NANP tiering), URLs against the live IANA TLD list, IDs, and junk or placeholder values — scored on the Linux side across all cores.

✓ Output

DAMA scorecard, audit trail included

Completeness, Validity, Uniqueness, and Consistency per column — plus a rule-level audit trail with registry IDs and the exact SQL shipped to the host. Text and CSV, identical across all six datastores.

Datastore coverage

Six stores, one motion.

The same preset that profiles a Db2 table profiles an IMS segment or an Adabas file — using the vendor connectivity your shop already licenses.

DatastoreAccess from LinuxPushdown
Db2 for z/OSDb2 Connect / DRDAFull — DRDA work is zIIP-eligible under IBM's processing rules
Db2 for iIBM i Access ODBCFull
CA DatacomDatacom Server SQLFull
CA IDMSIDMS Server SQL OptionAggregates
AdabasAdabas SQL Gateway (CONNX)Aggregates
IMSODBM / IMS ConnectStreamed
Proof

Proof in five minutes, on a laptop.

Demo in the box

make smoke

The full pipeline runs on generated data with no mainframe access required — the scorecard, the tiered phone logic, and junk detection, live in front of your team.

Reproducible benchmark

Verify the throughput yourself

The same command is the throughput proof point. The 800K rows/sec-per-core figure was measured on an 8-core, 256 GB Azure instance with 2 TB SSD — and your team can re-run it on your own hardware.

Available per data-quality assessment

The solution is licensed as a one-time purchase per profiling assessment: one complete profiling run across your declared datastores, tables, and rule chains, producing the full DAMA scorecard and audit trail. Pricing is scoped to the assessment during a demo.

Availability. Built and demoable today — the make smoke demonstration requires nothing from your environment.

Let your mainframe grade its own data.

A demo runs the full pipeline in front of your stewards in five minutes — no host access needed.

Request a demo