Server running without unplanned downtime since 1993

Originally published at:


I have a Stratus FT server running in my development lab that’s been up for several years. Only time it’s come down was due to building power issues and an office move. Stratus is a technology partner of ours as well as a customer - great company.

We deal with a lot of banking and credit card processing companies and if you’re looking for some serious uptime, nothing beats a NonStop system (formerly known as Tandem Computers). A significant chunk of my company’s business comes from this very particular industry niche.

When “mission critical” truly means life or limb (think E911 systems) - you’re not going to trust anything coming out of Redmond.


I’m sure that there is a place of honor reserved for this system in the datacenters of Valhalla; for system that stands fast against entropy and corruption for longer than most product lines remain alive possesses might matched only by its devotion to duty.


Is that server made out of stacks of beige 1993-era tower PCs? Colour me impressed.


The flickr geolocation tags claim that the photo was taken somewhere in France; and the article refers to what sounds like an American company; so I doubt that it is an image of the actual hardware; but from what I was able to google up; the Stratus and Stratus-derived fault tolerant systems don’t look all that exotic, externally. Considerable amount of special attention and custom hardware to do things like run CPUs in lockstep; but unless you look closely there isn’t much to distinguish a super-redundant system running two physical CPUs to provide one fault-tolerant logical CPU from a two-socket server; which is about the most banal object in your average datacenter.

I don’t doubt that the guts are really impressive; but the era of Serious Hardware also looking totally badass appears to be substantially over. You still get the occasional cool bezel or seriously nice wiring job; but the Crays and Connection Machines of the world have mostly been replaced by rack-mounted rectangles. Pragmatic; but not elegant.

1 Like

This is what my FT 6200 looks like - basically a double stacked 4u pizza box with a common backplane. It’s the hardware drivers that allow it to be hot-swappable. I can pull the I/O, CPU and power enclosures without interruption and the system will failover automatically. Not all that exotic and it uses basically commodity parts and mounts in a standard rack. I think mine is running Xeon chipsets.


Wow, I’m amused when I get to over 90 days of uptime on my desktop / home server. I’ve been playing with some settings lately, so I’m no where near that right now!

woof@garm ~ $ uptime 11:50:54 up 4 days, 17:25, 2 users, load average: 1.12, 1.37, 1.32

So … about 8762 days short? (Yeah, I know, it wasn’t up continuously, but still …)

So, how often do they have planned downtime?
Also, if they just hot swap out components, at what point does it become a different server?

“This ax has been in my family for hundreds of generations. A distant relative replaced the dull iron head with a steel one. My great grandfather put a new handle on when it broke…”


Great, now I want one for, um, totally justifiable reasons that I will think of shortly. Must restrain gadget lust.

1 Like

I thought this. My Windows laptop hasn’t had any unplanned downtime in about 3 years, but then it typically only gets used an hour a week max these days.

My phone, on the other hand, often has uptimes of months. Phone technology is rather impressive these days.

1 Like

Ahh windows 10 - it gives very little but takes everything.


And I thought those Data General systems running Meditech were stable.

This topic was automatically closed after 5 days. New replies are no longer allowed.