This introductory to VM/370 video was shown to IBM Field Engineers (FE) in 1972. Field Engineers were responsible for maintaining and repairing equipment installed at customer sites. This video describes the how virtual machines work, its features and advantages of the virtual storage technique in improved computing.
In college in 1983, we installed VM/370 onto a secondary mainframe (IBM 370) with 4Mb of ram. We then installed VM/370 on top of VM/370 to see if it could be done. After a few test tries, we continued stacking pancakes on top of pancakes and at around 20 deep, the 370 suddenly froze up, all of the early LED's glowing red. It was cool, and we knew this would be the wave of the future!
As an undergrad student at Brown University (1973 to 1977) I had access to the university’s mainframe. That mainframe was an IBM 360/67 running CP/CMS. This was the prototype IBM for the 370 line. CP for Control Program was renamed VM for virtual machine. The only real difference was renaming the technology. Before the 370 line was announced and delivered, my classmates and I were giving the CP/CMS system a hard test. Some of the more advanced student worked hard to crash the entire mainframe. I met one who successfully crashed the entire mainframe. FYI crashing your own virtual machine was easy, but that only impacted yourself. Crashing the entire mainframe was the true challenge.
Nice Descriptive Video from 51 years ago👌
Before virtual memory our programs had to "overlay" the next phase of the program over the previously executed phase thus reusing the same memory. It violated the sin of self-modifying code but that was the only way to crowbar it all into memory. LOL!
It’s pretty cool they were able to get Garfield to voice this.
The feal stuff was related to the fact they implemented some funtions in the hardware so its well shared between hard and soft, the advantage of designing both in a single company. Ibm hardware are able to share part of a core , tenth in fact, not full one like amd64 architecture, which is pretty limiting.
16MB?? Unheard of! Seriously for 72 that was huge considering in the 80's we were still dealing in KB. In 93 a BBS admin told me he had a 5G drive and I couldn't figure who would ever need that much storage.
So where are we today? Now (with current Intel-VT/AMD-V) we have EPT and hardware VMCS (and ARM64 has NEVE) to support nested virtualization. Because, yo dawg, we heard you like virtual machines, so now you can nest one Virtual Machine Monitor inside another, so you can have VM’s running inside your VM’s. But wait, there’s more: sometimes you don’t want a whole VMM + guest kernel and userland, so we have containers where kernel control groups just make it look like you have your own host. “But wait!”, you exclaim, what if I want to put docker containers in my LXD container, running on my KVM VM, nested in my Hyper-V VM?!!”. Good news, because with nested container support, you can nest your container, inside a container, running on a VM, nested in a VM. Now, you might be thinking: “that’s nonsense, nobody actually does that!”. Right now, I have a build and test system that spins up Hyper-V hosted Linux VM’s, running nested on an AzureVM (also running on Hyper-V), to launch a series of LXD system containers, which then deploy a sequence of docker app containers that run tests. And yes, it all works great. And no I’m not insane. At least I don’t think I am.
In 1972, this must have been distributed on U-matic tape (en.wikipedia.org/wiki/U-matic)
IBM invented just about everything we think of as modern computing -- but they did it 50 years ago!
Hello, VM/370. Nice to meet you. Not really. I operated a number of IBM systems for a few years and I never could get used to the paradigms they employed. Unix/GCOS were more intuitive to me.
Maybe the beginning of VR concept also..
Its the first commercially available virtualization solution I'm aware of. But the terminology for everything prior to the 70s is so different it makes it hard to evaluate the documentation without a really close read
@@christopherjackson2157A big difference between mainframes and PCs is that hardware functions on mainframes are handled by essentially separate computers running their own firmware (microcode or licensed internal code in IBM-speak). 3270 user terminals would go through a communication controller like the 3174 Establishment Controller. A 3480 tape drive has its own controller box with microprocessor and floppy-loaded firmware. This is the key to the scalability of mainframes. What's relevant is that the operating system on the actual mainframe processor handles very little of the hardware tasks that OS drivers do today. Therefore, the OS is much simpler and virtualization is much simpler. Instead of faking hardware to isolate guests, like PC virtualization has to do, it becomes more of a virtual memory and data routing problem. The guest OS, CMS, is a very lightweight single-user OS. It's more complex to run multiple instances of MS-DOS prompts under Windows 3.1, which used virtual machine techniques to get real-mode MS-DOS programs to multitask in a 386 environment. To that extent, the 286/386 virtual memory system was referred to as virtual machines through the early 90's. Like they talk about here, each process has its own address space that's completely independent from other processes, and the OS manages hardware. The motivation of modern PC virtualization is that the OS itself (what they call the control program here) got so complex and services provided got so intertwined between processes, we have to add another layer and run multiple OSes.
@@christopherjackson2157had Burroughs not a comparable solution? They were the first with the virtual memory, so maybe they were earlier, but maybe not as transparent and thought out as IBM.