"Chaotic Architecture" At NASA's Jet Propulsion Laboratory 69
New submitter CarlaRudder writes: NASA's Jet Propulsion Lab (JPL) is ditching old, rigid, legacy tools and adopting a much more flexible approach that allows people within the company to pick and choose the technologies that help them do their jobs better. CIO Jim Rinaldi and IT Chief Technology Officer Tom Soderstrom are calling it "chaotic architecture," and they are using it to better prepare for change and to attract the next generation of IT talent to JPL.
Moar Struts! (Score:4, Funny)
They probably just signed a contract with Jebediah Kerman's Junkyard and Spaceship Parts Co. [kerbalspaceprogram.com]
Open Parametric CAD Standard (Score:5, Insightful)
What the Mechanical Engineering world badly needs is an open Parameteric CAD Standard. Right now it's horrible. Each company uses it's own proprietary file that cannot be easily shared with other software. There are some portable formats but you basically give up all of the engineering data. A CAD file should be an engineering document not just a model of what the perfect part should be. It should contain all of the important parametric data and the tolerances, GTOL's, surface finishes, fabrication notes, etc. It is amazing that this still doesn't exist and the costs of dealing with it are astronomical.
Re: (Score:3)
The Military has been working on it. The most recent rev to Mil-Std-31000A in 2013 is getting closer. But it's more one way, along the lines of a .pdf where each software writes to a common format that can be read universally.
Re: (Score:1)
ISO 10303, commonly referred to as STEP, is the standard neutral format for CAD data. The latest version, 10303-242 supports the kind of Model Based Definition (MBD) data you describe including tolerances and notes associated to 3D geometry and part features.
Re: (Score:2)
It's really hard. You could take a look at adapting Collada.
Re: (Score:1)
On a related note. Is it really true that even today you can find American engineers not using metric units? That must a royal pain for things such as collaboration, documentation, supply chain management, reducing cost, competition etc.
Sounds scary, but it makes sense. (Score:5, Insightful)
1) On every project, (individual) people have been the critical success factor, not process.
2) While you will always need process, process is not a replacement for good people. Most common IT processes attempt to ensure that errors made by poor performers are caught, but they also ensure that your best people will not be operating at peak performance. This is sometimes called "predictable mediocrity"
This Chaotic Architecture thing sounds like a step in the right direction... putting trust (with oversight) in people rather than an ivory tower dictating company-wide policies. The real trick is how to organize that oversight without ending up with the same dictatorship by corporate architects. This requires effective management at all levels; daring to delegate and trust rather than dictate... but I've noticed a bad shortage of such Leaders in the places I've worked the last few years.
Re:Sounds scary, but it makes sense. (Score:5, Interesting)
I agree with everything you said. However, there is the other side of it too:
'chaotic architecture' could just as easily be the state where users are given control and IT has to support whatever nonsense users want. We've all seen it. Company goes "BYOD" and "chaotic architecture" follows... every piece of crap random consumer grade device gets brought in... half of it doesn't run the business critical apps properly, centrally managed A/V isn't possible, virus infections run rampant and IT finds itself working on some twits $300 Sony Vaio with 1GB RAM and Vista Home Basic... torrent software consumes all bandwidth. Some nimrod installs an inkjet color printer that's only compatible with XP, then buys a Windows 8 laptop and wants IT to make it work...
IT needs to facilitate users getting the tools they need, WITHOUT letting it get TOO chaotic. :)
Re: (Score:2)
Re: Sounds scary, but it makes sense. (Score:4, Interesting)
I've worked with software that came from a "chaotic" environment. When it came to fixing bugs, there was a lot more moaning about "why the hell did he do that?!" compared to software developed against corporate standards, however the time and cost of fixing a bug were similar for both types of software. When it comes to adding enhancements, I found that the success factor was not having a company standard software architecture, but a good one. Software developed by a lone but clever developer entirely doing his own thing turned out to be easily enhanced due to outstanding software architecture. Software developed to company standards was equally maintainable if the standards were good enough, however in many cases one would find that the standards were applied poorly, or were themselves incomplete, leading to poorly structured and hard-to-maintain software. Again, in the end it comes down to people
Re:Sounds scary, but it makes sense. (Score:5, Insightful)
You need a good process if you having things break is unacceptable.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
From 20 years in aerospace:
You need a good process if you having things break is unacceptable.
Security of the CAD files and engineering the ability to take a lightning strike on the project would be a plus. From a grandkid of a generation that gave up on it 50 years ago.
Re: (Score:2)
How do they defend against Ruby and NoSQL? (Score:1)
It's one thing to allow IT and devs to choose their own tools, but how are they going to ensure that the chosen tools are actually any good? For example a few years back Ruby on Rails and NoSQL were all the rage. The hype was intense, and a lot of CIOs and managers bought into it without actually thinking it through. The problem is that Rails and NoSQL were pushed by many Rubyists who, well, didn't have a fucking clue as to what they were doing! If you thought Java software from the early 2000s was bad, the
Re: (Score:2)
Yes. One project I worked on around then included a large chunk written by a Ruby fan, who was sure it was the best possible tool for the job.
It was only when their stuff was complete and took all the CPU and still ran like a slug that they finally admitted that they'd have to toss Ruby out and write it again in another language. For all I know, it might have been the best possible tool for the job on a desktop machine, but not a sub-200MHz ARM.
Re: (Score:2, Interesting)
JPL doesn't use a lot of Ruby and NoSQL.. most of what we develop for spacecraft is in plain old C and some C++. Science data processing is often FORTRAN, because there's an enormous legacy base of processing codes in FORTRAN.
More python recently. Some LISP in days gone by. Some Java.. some enterprise apps (payroll, etc.) are Java against an Oracle back end.
Re: (Score:2)
The push to open source to lower costs really threw productivity down the cliff...
That theory would seem to conflict with what the JPL guys said, but don't let me stop your frothing, Coward.
The real architecture is chaotic too (Score:3)
Anyone who has ever worked or spent much time at JPL knows that the real architecture is chaotic too - a maze of buildings built over decades, and (like MIT) described only by arbitrary numbers.
By "chaotic" they mean "modern" (Score:1)
Seriously? Because that's how the rest of us have been doing it for at least decade or so now. How does some dinosaur CIO thinking that our new-fanged "interpreted languages" and "distributed version control" are less organized than his precious mainframe programmed in assembly with a bit of C here and there make news? Oh ya, I forgot, this is slashdot.
Disclaimer: No, of course I didn't RTFA.
Re: (Score:3)
Eh, you're in danger of being tarred with the dinosaur brush yourself unless you recognize that mainframes at JPL died 30 years ago and the systema being replaced are probably Oracle Applications and Microsoft Exchange.
Re: (Score:2)
Maybe remind somebody of the new policy :)
Re: (Score:2)
And I didn't mean to denigrate your main point, however your point would be stronger if stated accurately without hyperbole.
Re: (Score:2)
For that matter, sysops should not call themselves engineers unless they actually are.
Oh please (Score:2)
"Enterprise Architect". Just another feel good title for some idiot who wouldn't feel so important if he was called Facilities Manager. And what the fuck exactly do those people know about computer system architecture? Answer: Nothing. Stay out of IT decisions and stick to signing the cheques.
Re: (Score:2)
Aaah, of course, the "Solution Architect". Silly me.
Since we're playing buzzword bingo, presumably you'll be able to tell me if he leverages enterprise vision milestones in a collaborative centralised framework with a total quality cross sector commitment to excellence while delivering win-win best of breed paradigms?
Re: (Score:2)
"Your attitude is a perfect example of why senior management are happy to outsource IT"
And your obsession will stupid terminology for age old jobs is why no one takes people in management seriously. "Solutions architect". Stupid bloody name. As if no one else in the company comes up with solutions to problems in their particular area. Try "Operations manager". Or doesn't that come with such a highbrow sound? Whats your cleaner called, "Detritus removal architect"?
"Or that your view of the business is limite
Re: (Score:3)
Put IT people in charge? Oh , you mean like at Google for example? Or Microsoft?
Plenty of IT people become businessmen, but you never hear of an MBA graduate becoming a master programmer.
Re: (Score:2)
Look up who started the 2 companies. It wasn't MBAs. They just get hired when some unimaginative grunt is needed to do the day to day running of the company.
And why would IT people take the lead in non tech firms? Its well known that business types are afraid of IT because they don't understand it and so just promote their own, the CTO never stands a chance. Hence all the accountants in charge in the UK and marketing types in the USA.
Oh btw, erlang is a programming language, not a filesystem. HTH.
Re: (Score:2)
Heres one for you pal:
http://professionalsuperhero.c... [profession...erhero.com]
Who's Brin and Gates? Well since you apparently can't use Wikipedia Brin studied comp sci and maths.
Gates spent most of his formative years coding and devising algorithms.
"This thread is one for the wall."
If thats the idiot A/C wall then yeah, you're right up there with the best my friend. Its rather tell you posted entirely as A/C so no one can look up your moronic drivel in years to come.
JPL Isn't a Company (Score:2)
> that allows people within the company
JPL is a NASA center managed by CalTech. Neither is a company.
Article Summary (Score:2)
Quick summary: we stay the fuck out of the way of the engineers so they can install and use the tools they prefer in the way they want.
Kerbal Space Program (Score:2)
Remember the Mars Climate Orbiter loss? (Score:2)
Chaotic Architecture, brought to you by NASA. The organization where one team uses metric and the other English units of measure [cnn.com].
Re: (Score:1)
They're ditching telemetry data as well (Score:2, Interesting)
Apparently they are also ditching old, rigid, legacy telemetry data from the Apollo missions. They "lost" the tapes. All 14,000 of them!
NASA is a bunch of liars. Slashdot's fortune is apt: "The trouble with a lot of self-made men is that they worship their creator."
Certainly, not the creator of the firmament that the shuttle is (possibly) designed to penetrate.