PropelZ Multiplexer: Moving a Million Mainframe Data Sets from z/OS to the Cloud in Parallel

How do you move a million mainframe data sets from IBM z/OS to the cloud without a multi-year project? VirtualZ Computing Executive Chairman Marc Sokol and CTO Vince Re walk through the PropelZ Multiplexer — the component that turns a list of 100,000 or 1,000,000 files into a single, self-managing job.

Moving mainframe data isn't the hard part. PropelZ does that at 56,000 records per second, mainframe to cloud. The timeline balloons because someone has to account for every data set, submit every transfer, and verify every result. The Multiplexer takes that manual effort off the plate: hand it the inventory and a parallelism ceiling, and it schedules every PropelZ transfer, monitors resource utilization minute by minute, and floats the number of parallel tasks up and down to match what your system, network, and target can actually absorb.

In this episode of Skyward Data, the podcast from VirtualZ Computing:

Why a data movement project measured in data sets, not terabytes, quietly becomes a multi-month or multi-year effort

How the Multiplexer runs 100 PropelZ transfers in parallel and adapts the count in real time to zIIP utilization, available bandwidth, and target response — opening the throttle overnight and pulling it back during business hours

The customer result: 9,000 mixed disk, tape, and GDG data sets that ran from Friday night to Monday morning on a competitor's product finished in a couple of hours with the PropelZ Multiplexer

Why the 5% rule of thumb — a typical PropelZ operation uses about 5% of one zIIP engine — sets the ceiling and floor for how much parallelism your environment can sustain

How single-pass, direct z/OS-to-target movement on zIIP processors avoids the multi-tier gateway architectures where failures are hard to trace

How per-file status flows to a simple log and the operator console for integration with OPS/MVS, NetView, and other z/OS automation

Using the same technique to split one multi-terabyte table into a hundred parallel operations

Where this shows up: application modernization, virtual tape library relocation, daily partner data feeds, log files from dozens of LPARs, and feeding Databricks and Snowflake for AI and analytics

PropelZ is VirtualZ Computing's no-code tool for extracting and transforming IBM Z mainframe data to any target. It's part of a four-product portfolio alongside Lozen (real-time read-write access to live mainframe data without moving it), FlowZ (mainframe files to cloud object storage), and Zaac (cloud and SAN presented as native z/OS disk and tape).