An 18 Year Old Buys a Mainframe [video]
youtube.com
youtube.com
I had a sorta-kinda similar experience in college years ago. My college (a small community college in the middle of nowhere) had received an IBM AS/400 machine as a donation. I signed up for some "Survey of Operating Systems" class, and when I showed up the first day, the instructor (who already knew me) goes "Phil, your job is to take the AS/400, install an OS and get it on the Internet". Needless to say, at that point in time, I'd never seen or touched an AS/400 before.
Anyway, the instructor pointed me to a room with an AS/400, a huge stack of tapes (for the OS), a huge stack of manuals, and basically said "just do it." It was an interesting experience, but in the end, I got the thing on the 'net so you could telnet to it and work on it. And that indirectly led to the beginnings of my career in IT, as my first real IT job was as an AS/400 operator / Netware admin, and it was a job this instructor connected me with after this class.
So here's to taking vintage IBM equipment and dorking around with it and putting it on the Internet!
When everything goes perfect, they are still a complete nightmare. I can't imagine how you pulled it off, frankly.
I'm honestly interested in all kinds of IT horror stories, I need a reminder every now and then to keep the IT part of free from things that break and require tinkering (starting with unnecessary on-premise servers in small biz settings). Mainframes are obviously nine orders of magnitude beyond what I ever to, but still: the lessons of expensive and proprietary can be somewhat universal.
Btw, the original design was an awesome capability-secure system called System/38:
http://people.cs.clemson.edu/~mark/fs.html
http://www.garlic.com/~lynn/submain.html#futuresys
It never got off the ground. System/38 and AS/400 are the only visible remnants of it, survivors from an alien world.
"though running a Linux VM/partition/etc is more for sheer fun of it than work."
I thought that was to get benefits of modern stacks running side-by-side with legacy OS/400 apps. You don't use it that way? And while we're at it, did they ever provide an easy, machine-usable interface between those instead of a web app simulating a terminal typing individual characters in like some mainframe places do? That seemed kind of ridiculous.
http://www.jfsowa.com/computer/memo125.htm
...that illustrates the difference between knowledge and wisdom very well. Soya has a number of wise observations and recommendations that apply in enterprise software projects today. I've seen the same arguments. Just usually not that many at once given the narrow scope of any given enterprise tech.
He's definitely made me consider the Single-Level Store a weakness until proven otherwise. His argument that the customers saw no benefit except the performance tradeoff is disturbing and probably still applies a lot. I mean, the ability to avoid crashes, hacks, much technical debt, and many failed integrations with high-level machine are huge enablers. All people cared about was price for raw performance ratio, though.
Interestingly, I've been thinking similar to him on a number of angles as I look at these new projects. It's why I pushed CHERI processor over SAFE despite liking SAFE more. It's because CHERI is compatible with FreeBSD and MIPS while probably having less overhead. It's also why I pushed Java processors like JOP or Vega3 over another Secure Ada Target (or even Rust target) for embedded or enterprise servers. Python runtime or machine over LISP. And so on.
Time shows the path of least resistance with greatest performance and legacy integration must be taken. So, anyone wanting high-reliability or high-security has to work in a box. That's been rained on continuously for decades. ;)
For some things having silly amounts of RAM and a few hundred disk drives and multi gigabit interconnects just works really well.
IMO, it's less about OS than it is specialized hardware, you don't see multiple terabytes of RAM on normal motherboards. As such plenty of mainframes ran on a flavor of UNIX just like plenty of supercomputers used UNIX.
But, I can see people using other definitions.
https://scs.senecac.on.ca/~ibc233/MS400.html
I still poke at them over this. Another was them running Hotmail on FreeBSD instead of Windows. Lots of problems when they switched. Least both experiences were good for debugging purposes. ;)
Note: The only thing better would be finding out Apple ran on a mainframe or AS/400 given Jobs' hatred of IBM. I doubt I'll get the satisfaction.
The most interesting thing was while researching about them almost a decade later, I came to discover their programming model with a common runtime and a kernel level JIT for software.
.NET on Windows/WP, Android and iOS/bitcode might be the closest thing we have on current systems about that programming model.
Their reputation for reliability is well deserved. The only time ours went down was for upgrades, and the few times we ran out of UPS power during storms.
I'm surprised that you didn't have UPS for stopgap until the generators kick in. If you have a mainframe, it's probably for something mission critical to where a generator (or two or three for redundancy) is a good idea.
I'm curious what kind of business you're in where you can justify a mainframe and UPS but not generator, if you can talk about it.
Me and another guy were the only Windows programmers there. We put a friendly UI in front of the AS/400 via network calls (SNA APPC, later TCP/IP sockets) to an endpoint that the RPG guys wrote for us. Later, we wrote an Excel add-in to allow the users to access their financial data from the AS/400, and this became so successful that the entire company is now based around the same technology. Sadly, I don't get a percentage.
"You're gonna learn to swim the Stennis way."
RPG, at least back then, didn't suit me. It was very much column oriented, with this oddball tool (SEU - Source Entry Utility) that you had to program in. I understand the ILE version(s) of RPG have gone to more of a free-form format, but back then it was a royal pain.
Luckily, yes. It had an Ethernet port and TCP/IP, so once it was up and running, getting it on the 'net wasn't too bad. The biggest challenge was actually getting the IT dept. to assign us an IP address, etc. Luckily I knew the IT guy pretty well from doing work-study with him earlier, so it wasn't too bad.
On the other hand, while you could telnet to the thing, you couldn't really use plain old command line telnet, which usually assumes something like a VT100 or VT200 terminal. AS/400's usually want 5250 terminal emulation, not VT100/220. So I had to find a 5250 terminal emulator to use to connect to the system and actually work with it.
A bigger issue can be with trying to send each of the 24 function keys when a VT102 didn't have any.
Perhaps a different person writing about a different system each month.
Thanks. ;-)
Interview with Linus:
What's the most amused you've ever been by the collaborative development process (flame war, silly code submission, amazing accomplishment)?
I think my favorite part is when somebody does something utterly crazy using Linux. Things that just make no sense at all, but are impressive from a technical angle (and even more impressive from a "they spent many months doing that?" angle).
Like when Alan Cox was working on porting Linux to the 8086. Or the guy who built his own computer using an 8-bit microcontroller that he wired up to some RAM and an SD card, then wrote an ARM emulator for it, and booted Linux (really really slowly) on his board.
https://www.linux.com/news/special-feature/linux-developers/...
Every year, IBM is at the user conference for the software we currently use. They're always showing off their newest Linux machine with 200 cores and I don't know why, because half the people there aren't even using bar coding, and everyone there is a small fry.
Here's where a niche developer comes in. You can use Java which is supported by IBM, Zend has a nice PHP setup, and IBM's bastardized version of Apache. Want to build an API to interact with this stone age fucker? You can do it. Build a Slack integration and look like a hero. Heck build some crappy web app because it's easier and faster to do development with your favorite framework than it is to do it on the IBM with RPG and whatever the fuck that thing runs. SQL your brains out!
These customers aren't leaving the IBM without investing a half a million dollars into a new platform. Every year, there are less and less people there, but they'll gladly spend $50k to integrate some simple workflow.
Rent a booth for $1,000, give away an iPad or some FitBits and pass out some business cards. Go to 3-4 of these a year and you should be able to keep busy for a while. And you already know SQL, PHP, Apache, you just get to discover the 'intricacies' of doing it on the IBM.
https://www.reddit.com/r/IBM/comments/3relk4/i_just_bought_a...
I hate it when I assembled computers and I ended up needing extra screws... or worse, I ended up with "spare" screws which left me wondering what exactly I missed. Can't imagine doing it to something this big.
Or like when you buy something that comes packed neatly..and as soon as you unpack it it becomes impossible to put it back and make it fit as before.
Maybe I'm the problem... uhmmm..
Nah, that's just Rap's Law of Inanimate Reproduction in action!
http://www.anvari.org/fortune/Laws/1348_raps-law-of-inanimat...
That's one very wise law. Never heard of that before.
I wish 100 Dollar bills had screws...
Then you can go back and see what you missed.
At most I laid down a white sheet of paper and put the screws in sort of the same shape as I picked them up from, so I knew these two long ones were at the top, the short ones at the bottom, etc.
But still... one accidental hit to the table and boom..everything is messed up.
This is especially helpful if what you're disassembling has a ton of slightly-different-length screws. Also helps to keep a record of the cable routing which can be tricky on laptops and cameras.
That's just amazing. I can't believe this is a real effect.
I wonder how much has this been researched? Do you have any more links related to this effect?
I even got to learn about HIPPI while at it. I wish I had more adventures like that.
There's some considerable engineering in those machines. Though you could replace almost all the boards with one or two FPGAs these days
What are you using it with? BONES?
I have two signals coming out. I have a HD 1080p link coming from its HD serializers (actually four outputs via two serializers) which I laid out to several inputs (Tricaster, custom built PC with 10-bit input, monitors, etc..). For when I need full 2K/10-bit, machine works at 6fps, not realtime. Currently, machine has a HIPPI output and with that I have limited options. Well, one option to be specific. An SGI machine with HIPPI card and Phantom Transfer Engine on it which then transfers DPX files to a SAN. I would like BONES, or anything else for that matter, but that would require me to get a GSN instead of HIPPI and BONES, of course. I'm still looking for options out there to see if there are any. Preferably if there's an option for PC based solution and HIPPI. If not, GSN/BONES is what I'm considering. Price of that is a bit steep though (around $30k I think), but it's still an option that is considered.
Then most folk in London got rid of their Spirits.
Now this year half the big tent-pole movies are shooting 35mm film again!
It's a strange time for telecines, with so many companies having disappeared but newer, much more compact designs appearing, for archival scanning presumably. If you haven't discovered it already the archives of the TiG mailing list are a great resource for TK stuff - e.g. this thread about the new Blackmagic/Digital Vision/MWA/Kinetta scanners... http://tig.colorist.org/pipermail/tig/2016-February/thread.h...
Pin registration is not what telecine is about. That's where you get scanners and their area sensors vs telecine's linear array. We operate a few telecines (Cintel, MWA, Spirit) and a scanner (Arri), and tbh I've seen that "wobble" only on older telecines like FDL series, also from Philips/Thomson/Grasva.. they've changed quite a few companies in the past :) This particular telecine has a sprocket for film and is rock solid.
Telecines are extremely useful, even to these days. There's a difference between scanners and telecine, of course. Telecine's strength is that it's fast. It's realtime and, in case of this machine, it can work all day in shifts and not break a sweat. This is extremely useful if you have a lot of material. Not all of material on film is feature film. There is a ton of material out there on 16mm and 35mm, most from TV stations. This machine can do 35mm, 16mm, optical audio, magnetic audio. All in realtime and gives you 1080p/10bit with sound. Sensor isn't all there in 1080p as it could be, but for most material it's more than fine. 2K is important if you're doing 4:3 material from film since you get 1556 pixels of vertical resolution which is more than 1080. So, in a sense, you get a machine, for relatively cheap, that can be operated all day for all kinds of archival work.
There are better machines now, of course. Newer Spirits, Scanity, and you can even retrofit some of the older machines with Xena: http://digitalcinemasystems.net/?page_id=348
There was a lot of talk about Blackmagic's buying of Cintel and their mini telecine. Personally, considering it's Blackmagic and experience I have with their products, I would avoid their stuff in a wide arc. Especially if I would to run it all day long or, god forbid, run a business that relied on their stuff. They are just not reliable, both on product basis and as a company. MWA (https://mwa-nova.com/) has interesting small scanners for 16mm, something which we also have. It's a nice machine, but Datacine/Spirit is on another level hardware-wise/reliability.
Kinetta is an interesting approach, and something I've had my eyes on. I would consider that for fragile/sensitive material considering that's what it's all about anyways.
Cool thing about modern workflow with film is that hardware is more or less the same. Telecine, 16mm/35mm optical blocks, light filters, scanners, (re)winders, wet gate system, audio heads, Lipsner smith film cleaners.. it's all the same. New things are new light sources (LED vs Xenon - though there's a debate which one is better), camera/sensors, and software. Software replaced a lot of "boxes" around the hardware. If you have a reliable machinery that outputs a good image base then it's all about software. Especially in restoration area. It got a bit competitive in that area over the recent years. Everybody seems to have some sort of restoration software as an offering now. Heck, even I have some of my own software that I wrote a few years back for automating most of the tasks cleaning up images that come through from film.
There was an office space in the middle of our data center at my first big IT job. Only Unisys people were allowed to enter, and crossing the threshold into that room, or opening a Unisys cabinet was a serious offense.
I'd be secretive too! That "mainframe" was a $25,000 Xeon server with a mainframe emulator running on it. They had these sweet contracts where they would lease the server based on performance -- MIPS/mo.
The IBM mainframes are big power boxes with custom asics for i/o and lots of mumbo jumbo to make it sound magical.
It's a little more than my two monthly salaries (after tax) here in Czech Republic, and it's per year. I don't think I am nostalgic enough yet to pay for that. :-)
Anyway, thanks for info.
I wonder whether part of the problem might be that it's so awkward to replicate mainframe style init on desktop operating systems. Desktop OS init tends to be very pale against a mainframe. I'm working on a project related to this at the moment.
This is important when working with ESD-sensitive electronics.
That's 1500USD. I'm sure 900USD is possible if you keep an eye out, seems like a pretty sharp deal though.
I haven't found any with shipping (to Denver CO) more than ~$100. $2k sure is a bit steep....
And he got it for 237$, good deal :)
"And I had someone from IBM run the serial number for me,
and the original price tag was $350,000."I've only done a small bit of MF work, but the old-school IBM mainframers I've met (usually at conferences) have always been exceptionally helpful and friendly. It's a fairly small community. Young blood is often quite welcomed.
And those beasts will likely be around so long as there's mains power.
That said, man I wish I had the space for one!
*An incredibly lame joke playing off the I/O throughput of mainframes, and...oh, never mind.
It's interesting now, I would have never thought of designing machines (off the shelf bought parts) for low power usage before I moved out.
When I did move out, power was 30 cents per kilowatt-hour and I definitely noticed the difference moving from NVidia to AMD (for a PC that ran 24 hours a day). That PC is now back on NVidia because a single GTX 970 was more efficient than my 2x6870s. I really wish they'd allow me to turn off the dedicated GPU and go back to the Intel iGPU when I'm only running Visual Studio. Lucid Virtu could have done this but it's really buggy.
These days I just make sure Wake On Lan works properly (a challenge in Windows 10 with commodity hardware) and run mostly ARM boxes otherwise.
I don't think I'd even consider running old mainframe equipment in Australia (unless I had 2+KW of solar on the roof)
Of course with electrical heating most of the year, running power hungry hardware 24/7 is arguably more efficient than just straight burning the electricity for heat...
And as my (bank's) house has mostly electric heating I somehow still don't care about power consumption of various computers and electronics much.
OS/390, z/OS, and other ESA or z/Architecture operating systems are definitely licensed to a particular machine. Therefore, in practice you cannot run any classic ESA or z/Architecture operating system on your PC unless you can obtain a license from IBM allowing you to do so. It is believed that there are, however, four ways you could run z/OS, z/VM, z/VSE, OS/390, VM/ESA, or VSE/ESA under Hercules using currently available licenses:
1. Running under Linux on the Pentium processor of a P/390 which is licensed to run the OS.
2. Running under Linux/390 on a mainframe which is licensed to run the OS.
3. Running under the terms of a disaster recovery provision of the OS license (but I really don't recommend depending on Hercules to be your disaster recovery solution!).
4. Using the IBM OS/390 or z/OS DemoPkg, which is available only to IBM employees and IBM Business Partners
Note that I'm only speculating here. I'm not a mainframer, just a guy who's had a passing interest in these systems and when I was reading about them it appeared that the difficulty of gaining access to these environments was a significant obstacle to learning about them, whether for professional or hobbyist reasons, that despite the fact that there appears to still be quite a bit of demand for mainframe programmers.
Non-commercial is not an absolute defence not a requirement for Fair Use by it is a factor and with the educational/research facets ...
Pro tip: when someone offers money for using something about which you don't care, say "Sure, how much do you think you need to get started?"
My aged professors in college would joke about their interactions with mainframes. Some stories included making them vibrate at the right frequency such that they could move around. Apparently as a grad student one got in trouble for making them bump into each other. Kinda like having two Nokia phones vibrate in a "battle" against each other.
I'm sure he'll learn about principles of effective presentation i.e. "don't read the slides" in due time.
:P
The disability argument doesn't really apply here - what about deaf users, why say anything that's not on the slides? Because that's what the transcript is for. Usability is orthogonal to what I'm saying.
I would guess you can give a z890 a good match with a raspberry pi 3... Unless it's IO bound...
A quick google says: 2451 mips on a 4-core raspberry pi 3 against 9000 mips on a 32-core z990... hmmm.
They are inefficient for daily use, but great for lab work. I spin them up, run whatever distributed task I need on 8 cores with 15K SAS drives, usually distributed Sony Vegas renderer, and shut them down. They also make a great lab for virtual machine work, so that does mean bringing the cloud in-house. :)
The X366 is a serious piece of hardware that later got renamed to 3850. Their current incarnation is http://shop.lenovo.com/us/en/systems/servers/mission-critica...
My apartment is on the second floor, so a) I would not be able to get it upstairs and b) the floor would probably collapse under the weight. Plus, it is pretty big and would make my living room kind of ... crowded. And then, there is the electricity... And even if I somehow could solve all these problems, I would still need to attach a storage server to it, which kind of has the same problems.
So I envy this person, a lot, but realistically, I would not want to buy one for myself.
(IIRC, during the S/390 days, IBM built a 4U mini-mainframe one could rack-mount. There also was the P/390, which was a single-board implementation that came with one CPU, a channel controller and 128 or 256 MB RAM, all on a PCI card - I think it was meant for developers. Sadly, IBM has given up on small-ish mainframes...)
More: https://en.wikipedia.org/wiki/PC-based_IBM-compatible_mainfr...
These days there's always the Hercules emulator to get your IBM fix.
Me? I'm looking for a Micro A, the desktop version of the Burroughs B5000 mainframe.
$ ssh root@73.250.101.107
Still up, and it does prompt for a password. But he said the Chinese had been trying to get in for 2 weeks and failed, so it's probably not something obvious.
Why is he not using public key authentication? Or even better, limiting access with a firewall.
"The Chinese" (botnets) probe my ssh daemons all the time, but only with common, poor passwords.
You can get old SPARC servers super cheap on eBay too. I'm super tempted to get one. I was also trying to find an Itanium the other day but those are darn impossible to get. :(
I remember being so impressed at the size of the monitor and the crispness of the display...and plus the fact that the workstations only seemed to be in offices that had "clout."
though i got rid of my e4500 a while ago, and was glad to see that one go.
In terms of pure processing power though, still no idea
However I am wondering, what could it be used for nowadays?
I would do some benchmarking, I don't know, e.g. LINPACK or maybe even distributed.net projects, BOINC ones, etc.
Other than that? Maybe you could do some 3D renderings? could you put it head to head with a GPU doing something embarrassingly parallel? probably not with anything like a Titan X, but maybe it could give a run for its money to something less powerful?
I am very intrigued by what you could achieve.
You could run a couple of hundred or thousand Linux VMs for fun.
It's an IBM mainframe...it's not about how fast the CPU is, it's about how fast the whole system is (cpu, memory, i/o).
I've read that a lot of what made zSeries "special" was that they had special purpose chips for a lot of things, from I/O to encryption and other stuff (not really well versed in the area so please correct me if I'm wrong).
Not sure how many of those special features you can use if you use Linux instead of z/OS, but could it be in principle something tractable or not even close?=
These days it's a somewhat inefficienct electric heater that also can do some computing tasks.
Computers do this in the form of sound, fans, magnetic fields and network connections.
Of course, that's a very small amount of energy.
A tiny bit of energy is stored in the bits of permanent storage. That can be kept for (presumably) indefinite time, and not released as heat until the bits are erased.
The network connection energy is mostly lost as heat in the wire, but to the extent that it transmits information elsewhere that energy is also retained.
Hence very, very close, but not identical, unless you have the pathological condition of never storing any data or causing any other process to store data. To do this you'd have to compute away and no-one could ever learn the results.
Edit: I may be wrong about the net heating due to storage, since setting storage bits implies erasing their old state which releases heat into the room. Anyone with better thermodynamics training than me want to give a better answer?
Or maybe I should get back to work...
Also, after watching the very impressive video about many of the things he learned from the whole process (with hardware that's only 12 years old!) it turns out that since he's not putting any real load on it the power consumption isn't really that high.
Awesome project though and he presents well.
Very impressive kid.
It's largely random which submission of a story gets traction on HN. We plan to make the dupe detector more sophisticated. One of our goals is to privilege the first submitter more often, because we hope that will incentivize people to submit more unique things, which should benefit the community. In the meantime, though, it's a lottery in any one instance—but it does even out in the long run, the more good stories you submit.
I didn't mean to be offensive towards the user that posted this entry, but just highlight this "issue".
My main "wtf" is due to the fact that in the past I submitted links that were already in the system and I got redirected to the comments page, as usual. Considering this is the exact case and there are eighteen hours between submission, i guessed this is a noticeable glitch.
So please, stop dowinvoting me.
Yes, it's probably a glitch in the HN dupe detector, but people didn't click through to this article to read about that.
These things happen sometimes, you say something and it gets interpreted in a way you didn't intend. I wouldn't worry about it too much.
Probably a better version of this comment would be something like: "I actually submitted this 18 hours before this one went up. It's funny how sometimes they catch and sometimes they don't. Super interesting video though, glad it's being appreciated!"