CODE SALVAGE
JobCS-014
SubjectLegacy and failing hardware
MoveDisk image onto an emulator, a small on-site PC, or the cloud that runs that architecture
SpareGolden image, checksummed copies, original stays up until the copy boots
Still
in
production

The hardware can fail. The job stays up.

This is the lift. The chassis, the disk, or both are what can no longer be the only copy.

Where this sits

The practice is still keep, rehost, or rebuild. Use this page when the risk is the machine itself: a VAX or PDP-11, OpenVMS on Alpha or Itanium, an HP 3000, PA-RISC HP-UX, SPARC Solaris, SCO or Xenix, OS/2 Warp, an old Windows box, NetWare bindery, IBM i, AIX, or IBM Z.

Paths

VAX/VMS and PDP-11 Take a disk image. Run it under SIMH or Charon, on a Linux VM or on a small on-site x86 PC when a device has to stay on the LAN. Copy the image with TCP/IP FTP or DECnet COPY, whichever that machine already has. Telnet stays on the private side. Remote hands are a serial console and an outbound tunnel.
Alpha and Itanium OpenVMS An emulator, or a move forward onto OpenVMS on x86.
HP 3000 Charon-HPA.
PA-RISC HP-UX Charon-PAR.
SPARC Solaris Charon-SSP.
SCO, Xenix, OS/2 Warp, old Windows Already x86. QEMU on a Linux VM. Warp 4 can use a normal VPN. ArcaOS is the maintained OS/2 path.
NetWare 3.x and 4.x bindery QEMU on a VM. IPX clients need an Ethernet bridge across the tunnel. A normal IP VPN does not carry them.
IBM i and AIX A Power cloud: IBM Power Virtual Server or Skytap. Not an x86 emulator.
IBM Z An IBM Z cloud. Not an x86 emulator.

Which cloud

One cloud does not boot every architecture.

GCP or AWS The Linux VM for SIMH, Charon, and QEMU, and for OpenVMS on x86. This is the emulator case and the x86 case.
On the LAN A small x86 PC, a serial adapter, and a BNC or AUI converter when the remaining devices are still on thinnet or AUI. Same emulators. The devices never have to cross the internet.
IBM Power IBM i and AIX, on IBM Power Virtual Server or Skytap.
IBM Z IBM Z only. Separate from the Power cloud, and separate from GCP and AWS.

Disaster recovery

Keep a golden image of the disk. Keep checksummed copies of that image. Keep a spare, and boot the spare while the original is still up. The original stays in production until that copy has booted. The spare is not theoretical after that. It is a machine you have watched come up.

The same spare is the refresh when the hardware is already on a VM. Checksum the image, boot the copy, leave the live one alone until the copy is the one you trust.

Software, if it is needed

The lift freezes the system as it stood. When that is not enough, the same job can keep the old software running on the new host, apply the updates that still exist, or move the application forward: OpenVMS on x86, ArcaOS, or a current COBOL or MultiValue port.

Bring
The hardware, the operating system, and what still has to stay on the LAN.
Next
Request an interview Keep, rehost, or rebuild