« Back to Blog
BLOG POST

Running Elasticsearch on a Supercomputer from the 1960s? Here’s what you need to know.

By Freon Smuggler

Posted on: April 1, 2019

We all know that Elasticsearch is a powerful, flexible database, but we often hear the question “How can we make Elasticsearch really cool?”

The answer is — Freon.

Specifically, the Freon used to cool the CDC 6600, designed by Seymour Cray and considered the world’s fastest computer from 1964 to 1969.

But how do you run Elasticsearch on a computer released the year that Daredevil first appeared in comics and the Beatles first came to the United States?

(Sidebar: Why did those two media properties never cross over? “Daredevil in… A Hard Day’s Knight” — this practically writes itself.)

Here are six helpful tips to get you started with ES on a CDC 6600:

1. Don’t do it

One-hundred percent, do not do this. If you really need to do something retro, consider getting chickens.

2. Decide if the CDC 6600 is really the right computing technology for you

The CDC 6600 has a lot going for it — for instance, it has a Fortran compiler — but maybe you don’t trust silicon transistors and want something more time-tested, like vacuum tubes.

Of course, you’re probably thinking of inventing a time machine and grabbing ENIAC.

But have you considered something before even vacuum tubes? Have you considered running ES on a Jacquard loom? That would be years of fun — decades, even! (Even though a Jacquard loom is really more suitable for Redis.)

3. Get a CDC 6600

I don’t know where, maybe eBay?

(Note: If your project manager has the resources to do an Ocean’s 11-style heist to get a CDC 6600 from a museum, I call dibs on the Brad Pitt role, since, like him, I enjoy eating in every scene.)

4. Get a weight belt to support your back

You’ll see why.

5. Download ES

The current version of ES comes in at around 109MB.

So, assuming an 80-character punch card, we’re talking about 1362500 cards, give or take.

Luckily, since I also assume you sprung for the fancy card reader (1200 cards/minute), loading ES should only take about 19 hours.

Better get some more Freon just to be safe.

6. Oh, I forgot a step

Your CDC 6600 might work with Fortran, but ES — and the Lucene library behind it — is written in Java, so it’s time to roll up your sleeves and start rewriting.

Or you could do something really useful, like wander the land like the Alan Lomax of programming and collect recordings of the last living speakers of Cobol.

Those six simple steps will have you up and running ES on a CDC 6600 supercomputer in no time at all (from a geological point of view).

Or you could get an ES instance right now with our Beta platform.

Next: How to run ES on a generation starship — you know, for searching the universe for habitable planets without faster-than-light travel.

Happy April Fool’s day, y’all!

 

References:
https://www.computerhistory.org/revolution/supercomputers/10/33
http://ed-thelen.org/comp-hist/CDC_6600-Brochure-JitzeCouperus.pdf
http://ed-thelen.org/comp-hist/6600-humberd-slides.pdf
http://ygdes.com/CDC/cdc6600.html

eBook: Everything you need to know about DBaaS

Learn why DBaaS is booming and get a checklist of questions to ask vendors to find the right fit for your business.

Learn More →

DBaaS for Elasticsearch JetpackeBook: DBaaS for Elasticsearch JetpackEverything you need to know about Elasticsearch with Kibana. Plus, learn how DBaaS can save you time and money.Learn More →

Keep in the know!

Subscribe to our emails and we’ll let you know what’s going on at ObjectRocket. We hate spam and make it easy to unsubscribe.