Email or username:

Password:

Forgot your password?
Jason's posts Post Back to profile
Jason Lefkowitz

Founded in 1974, Tandem Computers was a leader in high-availability computing. For decades, if you had an application that absolutely could not tolerate unplanned downtime -- a bank, a stock exchange, a telephone network -- Tandem's "NonStop" computers were aimed at you.

NonStop machines achieved reliability through massive hardware redundancy. A NonStop computer was a cluster of computing modules, each with its own processors, memory, disks. A failure in one couldn't affect the others.

Redundancy made Tandem. So naturally, when they made some coffee mugs, it was important that they feature redundancy too.

en.wikipedia.org/wiki/Tandem_C
12 October at 2:11 | Open on vmst.io
21 comments
Jason Lefkowitz

You can still buy a NonStop-branded machine today, though thanks to 30 years of mergers and acquisitions you'll be buying it from HP.

hpe.com/us/en/compute/nonstop-

12 October at 2:12 | Open on vmst.io
Mark Eckenwiler

@jalefkowit When I think of HP in its current form, what comes to mind is Sick Sigma.

12 October at 2:29 | Open on thepit.social
JP

@jalefkowit lol. I once had a one in our dev shop. It was a rack with a NSK machine with a dec alpha unix workstation as its console with a proliant server as the Unix workstations console. Running windows. To get to the NSK you ran reflections on the windows machine and it auto connected you to the alpha and then onto the serial port on the NSK

Was shipped from the vendor like that as an integrated solution.

12 October at 3:03 | Open on aus.social
Chris Petrilli

@jalefkowit they also put out some absolutely epic tech reports.

12 October at 2:15 | Open on hachyderm.io
JP

@jalefkowit and then there was a tcp/ip bug which crashed the whole thing hard.

12 October at 3:01 | Open on aus.social
C++ Wage Slave

@Jplonie @jalefkowit
You're not joking. I used to work for a company that made redundant systems. (It wasn't Tandem.) They were pompously advertised as having infinite MTBF, because marketing people are like that and they plug their ears whenever Engineering talks to them.

What actually happened was that a bug took out node 1, and so node 2 took over node 1's workload and immediately crashed in the exactly the same way. Unsurprisingly, adding more nodes (beyond 2) didn't make the system any more reliable: they just produced a bigger and less predictable boot loop.

Not every failure would cascade across the cluster like this, but enough bugs did to make the infinite-MTBF claim laughable.

Edit: typo.

@Jplonie @jalefkowit
You're not joking. I used to work for a company that made redundant systems. (It wasn't Tandem.) They were pompously advertised as having infinite MTBF, because marketing people are like that and they plug their ears whenever Engineering talks to them.

What actually happened was that a bug took out node 1, and so node 2 took over node 1's workload and immediately crashed in the exactly the same way. Unsurprisingly, adding more nodes (beyond 2) didn't make the system any more...

12 October at 9:39 | Open on infosec.space
Harris👍Trump👎

@jalefkowit And then Compaq bought them and they died. Even renaming Compaq “HP” didn't help.

12 October at 3:49 | Open on mastodon.radio
Zeki Çatav 🤔 ☕ 🕯️🎶
Harris👍Trump👎

@catavz @jalefkowit

We lived in the Boston area in the late 80s … when DEC had a fleet of helicopters to fly people between their various campuses … *sigh*

12 October at 5:17 | Open on mastodon.radio
Large Format Projectionist

@jalefkowit

I saw my old employer's last Tandem. I wish I could've been there for the decom, it must be quite a procedure to actually turn one off.

The other division still operates Stratus, which makes no sense to me. Why? They ALSO have mainframes, which is what Stratus and Tandem were imitating.

Politics, and inertia.

12 October at 4:13 | Open on mastodon.social
Rogan Dawes

@Benhm3 @jalefkowit lots of banks here in ZA had Stratus or Tandem for their ATM switches, and mainframes for their actual banking transaction processing. Not sure if that is still the case.

12 October at 12:17 | Open on infosec.exchange
Large Format Projectionist

@RoganDawes @jalefkowit

Precisely what my old employers did, until they got rid of the Tandems around 2010. But that division they own has kept their Stratuses well beyond any rational limit.

Although reading the Wiki page on Tandem, I get that this isn’t just some “let’s cloud it” situation.

12 October at 22:12 | Open on mastodon.social
Steveg58

@jalefkowit
Back when I was working for WA Newspapers our biggest rival for the Saturday newspaper market was the Western Mail. We had a clunky old Merganthaler Linotype System 5500 and they had a shiny new Tandem NonStop. But they were so bad a managing it that it used regularly fail because they would not review the logs and take action when one module failed. Instead they would ignore it until the other module of the pair failed and the whole machine went down.
Eventually their parent company bought us and put us under their control ... you can imagine how thrilled we were.

@jalefkowit
Back when I was working for WA Newspapers our biggest rival for the Saturday newspaper market was the Western Mail. We had a clunky old Merganthaler Linotype System 5500 and they had a shiny new Tandem NonStop. But they were so bad a managing it that it used regularly fail because they would not review the logs and take action when one module failed. Instead they would ignore it until the other module of the pair failed and the whole machine went down.
Eventually their parent company...

12 October at 4:14 | Open on aus.social
Patrick Middleton

@jalefkowit Some great stories on this thread!

12 October at 5:08 | Open on sfba.social
Debbie Goldsmith 🏳️‍⚧️🏳️‍🌈

@jalefkowit IIRC when Tandem was in their heyday their main corporate location was where Apple Park is now.

12 October at 6:39 | Open on mastodon.social
Tim Bray

@jalefkowit In the eighties I was building a T1/ISDN multiplexing system and we booted Tandem from the platform vendor competition because their developer tools were utterly abysmal. Instead we bet on Xenix and it worked out ok.

12 October at 7:34 | Open on cosocial.ca
Steve's Place

@jalefkowit I might still have a promo shirt from them with this phrase on it: "So nice, so nice, we do it twice."

12 October at 11:05 | Open on mastodon.stevesworld.co
Volker Buzek
Stephen Hoffman

@jalefkowit Around time of the failed ports of NSK to Alpha and then to Itanium, the redundancy switched from being hardware lockstep to firmware. (None of the commodity processors support lockstep.)

The NSK product line and most of the HP “server” business (and whatever little was left of DEC) was eventually ceded to HPE.

Here is some info on the post-Tandem-hardware commodity designs, from 2008:

availabilitydigest.com/public_

12 October at 14:30 | Open on infosec.exchange
LeoBurr :leoburr:

@jalefkowit I was at Compaq during the Tandem merger.

I thought we all were getting free or discounted dual stick popsicles.

12 October at 18:26 | Open on tiggi.es
Sonikku

@jalefkowit
I encountered Tandem during my 6+ year tenure in the banking industry. Some banks in my country still use these systems, because of the high reliability required.

12 October at 21:10 | Open on oldbytes.space
Go Up