Ask HN: Forensic analysis for 50 year-old tape? | Hacker News Reader

Ask HN: Forensic analysis for 50 year-old tape?

57 pointspasser_byer33 comments

Im posting for a friend who moved recently. He found a magnetic tape that was written over 50 years ago. He says, if he recalls correctly, the tape would have contained 2 file, EBCDIC encoded, written by an IBM utility using a tape sub-system attached to a s/360 processor running MVS.

Assume the IBM guys have this setup in some dusty basement in upstate New York.

This is a 2 part question. What is the best method for attempting to read this mag tape? That’s the main question. Second, what is the probability of success here? Assume the tape was kept in a climate controlled home.

View on HN

I think you should reach out to Jason Scott over at the Internet Archive. He's been involved with helping people get old stuff off of old media before and if he doesn't have the tech and know how, chances are high he knows who to ask.

He's very easy to reach on Twitter, @textfiles.

Great tip, thank you. Easy is relative. Lucky for me, I have kids who can get me going with a twitter account.

email jason at textfiles dot com

Thank you!

This probably isn't going to help the poster specifically but it's an excuse to share a good presentation from Vintage Computer Fest West 2020 re: magnetic tape restoration: https://www.youtube.com/watch?v=sKvwjYwvN2U

I believe the presenter's tape was also from an IBM 360. I guess maybe, if the rig the presenter used to read his tape still exists, this might be of use to the poster. The presenter had a tape from his college days in the 1960s, stored in less than ideal environmental conditions, that he wanted to read.

The presentation explores using software to digitize and analyze the analog signal generated by the flux transitions on the tape to reconstruct the data. Essentially, it's moving the digital portion of the tape drive into a software domain and doing signal processing. Software-defined radio kind of stuff, to some extent. Very similar to the efforts to preserve floppy disks (which are kinda like tape drives for tiny circular tapes, basically) by recording the flux transitions and processing them.

The presentation gets into some work applying these techniques to reconstruct tapes recovered from the Whirlwind project, too.

Thanks, that's an interesting talk. It certainly makes sense to do as much of the signal processing in digital as possible but you still need a functioning analog front-end and their approach required access to an at least partially working 9 track tape drive, which I don't think have been manufactured in about 20 years.

That's why you should reach out to the presenter.

>50 yr old tape< = circa:1972 vintage Tape

search: >'magnetic tape restoration'< @DDG : <https://html.duckduckgo.com/html/?q='magnetic tape restorati...>

The chances of reading the data will depend on several factors;

  1. The tape-stock (base and binder stability), the stored temperature, humidity and magnetic field exposure.

  2. A working and aligned reader.

Background.

The IBM System/360 (S/360) : https://en.m.wikipedia.org/wiki/IBM_System/360

Is a family of mainframe computer systems that was announced by IBM on April 7 1964, and delivered between 1965 & 1978.

Extended Binary Coded Decimal Interchange Code (EBCDIC ) : https://en.m.wikipedia.org/wiki/EBCDIC

Is an eight-bit character encoding used mainly on IBM mainframe and IBM midrange computer operating systems.

Also See;

search: >'NASA call for Moon tapes'< @DDG : <https://html.duckduckgo.com/html/?q='NASA call for Moon tape...>

I’m guessing you also have a book recommendation on the history of the 360 architecture. Excluding Watson’s bio and “Mythical Man Month”, what book should I read next?

Not GP, but 'IBM's 360 and Early 370 Systems', Pugh, Johnson, Palmer is solid.

The geophysical industry used 9-tracks for a long time and there are still services that will recover your legacy tapes and reformat to any modern output.

I have never used these guys, OnTrack(0), but they claim to do 9-track recovery.

Everything we did in the industry for decades lived on 9-track tapes. EBCDIC headers are a standard feature of SEG-Y output datasets.

Here is one who deals with geophysical data, Data Strategies Interchange(1).

[0] (https://www.ontrack.com/en-us/tape-services)

[1] (https://go-dsi.com/seismic-data-transcription/)

Good luck. Sounds like a fun project. I wrote a SEGD/SEGY reader for geophysical data a long time ago using tcl/tk. It was surprisingly easy. It used data from disk files though. I got rid of my SCSI 8mm and DLT drives a long time ago and went to pure disk IO. No more issues spinning tapes after that. In the industry though we did have some really good tape reading/writing/checking routines. Haven't done that since the mid-90's myself.

Email George Blood LP (https://www.georgeblood.com/), they have support to read like 100+ old magnetic formats and are wonderful pros.

Another thank you! I have choices

Aside from the physical encoding, note the conventions for IBM tape file sequence and labeling:

https://www.ibm.com/docs/en/zvse/6.2?topic=SSB27H_6.2.0/fa2m...

You can bypass default label assumptions and label processing using JCL directives like BLP, NL, etc.

When I was in mainframe operations and at times ran the Production Control Center (triage), I rescued some systems from being rerun due to clobbered labels by altering JCL.

This is very helpful. I’m guessing you have a copy of the Brown/Blue => Blue/Brown book.

I'm not familiar with that description. I worked as a 370 series (370/3033/308x/309x) OS/MVS computer operator in the 80s, before programming.

I worked as an 370 ACP/TPF Assembly Language Programmer in the 80s. I think this is referring to Gary DeWard Brown who wrote books about assembly language and JCL for example. Don't remember if they were blue though.

Gary deWard Brown is the author and hence the name. If I recall, he also describes utilities like IEBGENER, IEBCOPY and so on.

One copy of the book on Amazon was over $100.

Contact this guy: https://www.youtube.com/c/CuriousMarc

If he does not have a machine capable of reading the tapes in his house/museeum, he will know someone who has.

(and as a bonus, the recovery might end up as a entertaining youtube video)

There is a pretty marvelous woman, named Marianne Bellotti. She probably couldn't help you directly, but she's a total "old computer" nerd.

https://www.marianne-bellotti.com

I do not have contact info for her, and she has no idea who I am, so you're on your own...

I'm guessing that would be one of the old reel to reel tapes. I think the hardest part (assuming the tape is intact) would be finding a working tape reader. From pictures I've seen I think those were real beasts, in cabinets as large as the mainframe itself. The Computer History Museum in Mountain View CA is one of the few places that might have one.

Otherwise you might be able to build your own tape reader. You could probably make it a lot smaller and simpler than the original machines since throughput wouldn't really be an issue. Still it would probably be quite an engineering project.

I have a couple of 4U 9 track tape drives with centronics SCSI interfaces. Pulled from 90s Sun gear from ex Federal deployment. Government stuck with tape for a long time.

One of them ran in 1999, at least, but they've been in the barn since.

rebuilding that to functional might be harder than making some hacked up sensor to run tape through.

Now that would be a fun hack, no? I bet with the collective knowledge from this thread I might be able to pull it off.

I’m still working for the man, though.

I recall a time that I wrote a program that called for a tape mount and a file to be read where this file spanned 2 volumes. There was a 3270 terminal next to the tape reader. After the job is submitted, I was amazed at the velocity of the read.

The all time favorite raised floor data center story: I worked for a commercial software vendor back in the day where sufficient physical security for the data center was a lock on the building door, but no locks to the raised tile floor area.

One day I walked through the doors of the data center and everything was covered in a fine ash-like powder. Another employee brought their young child with him to the data center. This child pressed the red button labeled “Haylon Dump”.

I was told by the operators on watch that everyone was lucky to find their way out safely.

New locks were installed the next and badge swipes were required for entry.

Amazingly, all the hardware mostly survived.

Yep, that is why Halon is used in data centers. It stops the combustion reaction at relatively low concentrations and it is a gas so it does not damage electronics. However, breathing it in is not really recommended.

I am surprised at your last line, if there was not a fire, why did some hardware not survive?

Maybe there'd at least be a way to salvage the basic mechanical structure, replace worn out mechanical parts with 3D printed or CNC'd equivalents and replace the sensor head with some kind of modern magnetometer.

I'd bet they'd operate with a cleaning. maybe some caps replaced. Any rubber belts are gone but i think these are direct drive.

1) be careful! The magnetic coating will flake off and you will lose the data. Research a bit: this is a solved problem. There are chemical treatments to render the oxide coating more flexible for an attempt at reading it.

2) is there a way to avoid forcing physical contact with rubber capstans as you attempt to pass it over a magnetic play head? Is there an alternative magnetic sensing technology you can use to extract the data vs a proximity based typical tape playback head?

I'm curious if there is something valuable on this or if you are just doing this for the challenge.

Get yourself a TEAC/Tascam 8 track reel-to-reel and digitize the tracks, then take it from there on a DAW.

Sadly ibm used 7-9 track tapes, so there's only a 33% chance that'd work, assuming all sold the same

You can offset the tracks and run it again until you have a good image. It should be fairly trivial to sync up repeat takes. But better of course if you can gain access to the original gear in running condition.

IBM’s dusty tape lab used to be in Tucson. Recovered the space shuttle tapes iirc.

OP here. These are some really great ideas here and I thank you.

Reply on news.ycombinator.com