CP/CMS, VM/370, and the Reinvention of One Computer as Many Virtual Machines
IBM's CP/CMS and VM lineage showed that one physical mainframe could present many isolated virtual computers, turning virtualization into a practical operating-system architecture decades before modern cloud computing.
Virtualization began as a way to share expensive mainframes
In the 1960s, computers were costly institutional resources. IBM’s Cambridge Scientific Center explored whether one System/360 machine could be divided into multiple independent computing environments so users could experiment without destabilizing one another.
IBM’s VM timeline traces that work from CP-40 through CP-67 and eventually VM/370.[1][2]
CP and CMS deliberately separated two responsibilities
The architecture split resource virtualization from the interactive user environment. Control Program, or CP, managed the physical machine and created virtual machines; the Cambridge Monitor System, or CMS, provided a compact interactive operating environment inside a virtual machine.[3]
This separation made the hypervisor responsible for illusion and isolation while the guest system focused on user services.
Each user could see a complete machine
Rather than merely scheduling processes inside one operating system, CP attempted to reproduce the hardware interface closely enough that guest operating systems could behave as though they owned a machine.
Isolation made experimentation safer
A malfunction inside one virtual machine was less likely to crash other users’ environments. IBM histories identify this isolation as a practical advantage of the VM design.[4]
CP-40 proved the concept on modified hardware
CP-40 was developed on a specially modified System/360 Model 40 with address-translation support. IBM’s historical material places the work in 1964–1966 and describes it as the foundation of later VM systems.[1][5]
The project showed that hardware assistance and a carefully designed control program could make virtualization efficient enough for serious use.
CP-67 moved virtualization onto the System/360 Model 67
The Model 67 included Dynamic Address Translation hardware. CP-67 used those facilities and spread to research and customer sites, turning the virtual-machine idea into a broader community rather than a single laboratory experiment.[5]
Source sharing accelerated improvement
Early CP/CMS software circulated among users, who modified and exchanged it. This community history anticipated later open-source collaboration even though the licensing and institutional context were very different.
Virtual disks made storage part of the illusion
Virtualization had to extend beyond processors and memory. Guest systems needed virtualized storage and I/O so that each environment could behave like a coherent independent computer.
VM/370 turned research into an official IBM product
IBM announced VM/370 on August 2, 1972 for System/370 machines with virtual-memory hardware.[1][2] That release established a commercial lineage that would continue through VM/SP, VM/ESA and today’s z/VM.
Virtual machines had moved from research technique to supported production infrastructure.
Virtualization changed testing and system development
Developers could run experimental operating systems, new releases and potentially disruptive workloads inside separate virtual machines on one physical host. That reduced the need for dedicated hardware for every test environment.
The same idea later became essential to software testing, hosting, disaster recovery and data-center consolidation.
The hypervisor became a new systems layer
Traditional operating systems multiplex processes; a hypervisor multiplexes whole machine environments. This added a layer beneath guest operating systems that could control scheduling, memory, I/O and isolation without requiring every guest to share one kernel.
Modern virtualization technologies differ greatly in implementation, but the architectural distinction remains recognizable.
Why IBM’s VM lineage matters to cloud history
Cloud computing did not invent the idea of dividing physical hardware into isolated logical computers. IBM’s VM family had been doing that on mainframes for decades.[2]
What changed later was scale, commodity hardware, automation, networking and on-demand provisioning. The underlying conceptual move—one machine presented as many—was already a mature operating-system idea.
Works Cited
- 01IBM — z/VM History Timeline vm.ibm.com
- 02IBM — VM 50th Anniversary ibm.com
- 03
- 04
- 05
CodeHistory is a living archive. Citations document the evidence used for this edition; later evidence may refine the account.
Submit a research lead