NASA Nebula, Cloud Computing In a Container

1sockchuck writes "NASA has built its Nebula cloud computing platform inside a data center container so it can add capacity quickly, bringing extra containers online in 120 days. Nebula will provide on-demand computing power for NASA researchers managing large data sets and image repositories. 'Nebula has been designed to automatically increase the computing power and storage available to science- and data-oriented web applications as demand rises,' explains NASA's Chris Kemp. NASA has created the project using open source components and will release Nebula back to the open source community. 'Hopefully we can provide a good example of a successful large-scale open source project in the government and pave the way for similar projects in other agencies,' the Nebula team writes on its blog."

Comments

Is that something like what Google has?

[Taco Cowboy]

Water Cooled Data Center In A Container?

Re:Is that something like what Google has?

[Anonymous Coward]

they rediscovered the rack-mountable server. and much media celebration ensued. yawn. wake me up when something genuinely interesting happens, save this Cloud Computing crap for the buzzword crowd.

hahahah

[Anonymous Coward]

yes but does it run linux?

Re:hahahah

[NoYob]

Yes. And it's all written with PHP.

About Time

[Mortaegus]

We need more investment in open-source projects like this. Such works are building a better future for us and I, for one, thank them for it. Though I doubt we will see any major commitment pushing for open sourced works from private operations, so the quote is a little optimistic, IMHO.

120 Days?

[captaindomon]

Doesn't that still seem like quite a long time? Four months? I guess that includes the "Planning Cycle", whatever that means.

Re:120 Days?

[worip]

This is the aerospace industry we're talking about, there is A LOT of red tape surrounding each and every movement, change, purchase, etc.

Re:120 Days?

[RockDoctor]

120 Days?

Doesn't that still seem like quite a long time? Four months?

Have you ever tried wiring up the interior of anything built into a shipping container? It seems like a quite reasonable build speed to me, having been involved with the building/ rebuilding of a half-dozen containerised laboratories over the years.

Firstly, if you're using a standard shipping container to simplify movement issues, then you've got to stick to the dimensions, weights and stiffnesses specified for such containers. Otherwise you also need to specify special transport arrangements, unusual crane lifts. Bang goes your "standardisation" justification.
That requirement limits the number, size and area of doors you can use to access the interior. Sure, you can build it on an open frame, then put it on the back of a lorry, ship it, and spend the next 6 months ripping it apart on site to find out what got damaged in transit.
That in turn constrains the number of technicians that you can get working inside the unit at one time. Personal experience says that you can't realistically get more than 4 people working in one at once before elbows clash enough to slow things down. If you need more people on the job, dump the 8-hour working day for two or 3 shifts.
That these units contain lots of standardised 19" racks will help in one respect - you can build and test the racks in the workshop, haul them into the unit, wire them up to the test loom (don't "save time" here, please ; "short cuts make long delays") and if they pass, connect them up fully. But this would be counteracted by the high rack density, and the need for cooling pipework (gas or liquid).
The plumbing is going to be "fun". For suitably small values of "fun". I'd be strongly tempted to plan on building the plumbing and commissioning it before anything more than basic power and lighting goes into the shell. which begs the question of building and testing the interconnects as the first (hardest to remove) racks go in. You really don't want to pour water all over the back racks when testing the penultimate racks. Details depend on the technology chosen.

The timescale doesn't sound too unrealistic to me ; I've done similar jobs, and while they're not as bad as scraping up guinea-pig shit, they're harder work than my present job.

Of course they will

[NoYob]

In addition, the Nebula platform itself will help facilitate the adoption of open source software across the Government.

That won't be sole the reason. As departments have to cut budget's in the near future, they'll be looking more and more to F/OSS to save money. Nebula is proving the low budget F/OSS solution as viable.

Re:Of course they will

[Mortaegus]

It doesn't matter what the reason is, the fact that they are doing it at all is what matters to me. I'd accept a company bragging and boasting about themselves for doing something good in a heartbeat, as long as they did something worth doing. Likewise, I don't care what the motivating factor is in this instance, as long as it gets done, and really is open-sourced.

Re:Of course they will

[TheKidWho]

What a fanatical viewpoint.

I don't care what it is, as long as it's open source!

If it's not open source, then it's BAD.

Re:Of course they will

[Mortaegus]

Not necessarily. I wouldn't want military weapons systems to use open-sourced software, and I believe that any work that takes time, effort, and an investment in resources should be properly rewarded, so not releasing the source-code in such instances is entirely forgivable. Provided, of course, that it is released within a reasonable time frame. I'll freely admit to stealing other people's work in the past and adapting it to my own ends. The difference being that I understood what was done and was capable of doing it myself if I had the time and resources to invest in making it, but without their original work I would have lacked the foundation upon which my own is built. It's called evolution, and it applies to more than physiology. Keeping code closed forever is like making a species extinct. Something else might come in and fill the void, eventually, but it has been removed from the gene-pool. But maybe that's a bad example.

Re:Of course they will

[smoker2]

What a fanatical viewpoint.

Yes you have. Get a fucking clue.

If it's not open source then it's another kind of software, software isn't BAD or GOOD, it's how it's used that's important. Or was that a troll ?

Re:Of course they will

[TheKidWho]

It was sarcasm at the GPs viewpoint.

quick?

[pak9rabid]

NASA has built its Nebula cloud computing platform inside a data center container so it can add capacity quickly, bringing extra containers online in 120 days

4 months is quick?

Re:quick?

[Anonymous Coward]

4 months is quick?

Compared to 2 years (from TFA)? Yes, quite quick. :)

Re:quick?

[pak9rabid]

It may be quicker, but that's by no means quick.

Re:quick?

[ArsonSmith]

Welcome to government run programs.

Re:quick?

[sensationull]

It's the public sector, in government terms NASA has just invented the deployment warp reactor.

wrong use of "cloud?"

[Lord Ender]

Like so many promising high-tech ideas, the word "cloud computing" is being over-used. Cloud computing means being able to get virtual hosting with a few clicks, and automatically scale up and down as demand changes, all while being billed by resources actually used.

Not every cluster of servers or supercomputer deserves to be called "cloud." Not everyone who runs VMware deserves to be called "cloud."

Re:wrong use of "cloud?"

[JWW]

I believe that Nebula is deserving of the cloud moniker. From what I've seen of it, it is very similar to Amazon's EC2 "cloud" services.

Its got virtual hosting and can scale and provides many of the same components as EC2 plus has disk sharing across instances.

Nebula is really a cloud.

Re:wrong use of "cloud?"

[Anonymous Coward]

And EC2 and Nebula are really just mainframes. Welcome to the 1960s, you dipshits. You're "rediscovering" what IBM has been offering for nearly 50 years now.

Re:wrong use of "cloud?"

[MrTester]

Your looking at it wrong.
NASA is standing up a generic service for data processing. Various programs with data processing needs at NASA can access this, and from their perspective they get "virtual hosting with a few clicks, and automatically scale up and down as demand changes, all while being billed by resources actually used."

Re:wrong use of "cloud?"

[Yvanhoe]

At one moment there was entanglement between "grid" and "cloud" computing. I'm not sure people who use them usually understand their underlying meanings. I try to make people more specific, I don't like these words which are too close to buzzwords, in my humble opinion...

Re:wrong use of "cloud?"

[wagnerrp]

You're looking at it wrong. NASA's batch supercomputers have operated in this fashion for decades. One division gets a large supercomputer, and various projects get allotted CPU quotas. They can run as small or as large of tasks as they please on the first free nodes that open up, all billed to their quota. The fact that they put it on the internet for use with web servers does not make it anything new or amazing.

Re:wrong use of "cloud?"

[ArsonSmith]

everything old is new again.

With a new name too.

NASA computing in a cloud?

[Subm]

A nasa computer in a nebulous cloud?

I think we all know where this is going -- V'ger wants to bond with its creator and instead gets a bald girl and Decker to create a new life form.

Is NASA suffering from mission creep?

[wisebabo]

Look, I'm a big fan of the space program and everything but with the country at (two) war(s), hocked to the hilt, economically stuttering, NASA (like the rest of the government) needs to be focused on its "core competencies" (no I'm not a PHB). Where does building data centers fit into NASA's mission statement?

I realize that there are tremendous amounts of data that needs to be captured, analyzed and archived (the Terra satellite sends a terabyte of data a day alone I think) but isn't this something that can be done more efficiently by private industry (Google?). Maybe it can be even outsourced providing it is not of a sensitive nature, I mean isn't the data for all mankind?

Re:Is NASA suffering from mission creep?

[belthize]

A lot of their information is considered sensitive, particularly engine modeling and telemetry.

The question of mission creep is certainly valid. In this case though I suspect it's simplest/best for NASA to do it themselves. Frequently outsourced cloud computing at this level looks good financially but simply isn't practical when you translate white paper speak:

extensible, flexible, segmented, secure, etc into reality

use our API and like it, you can have 8 or 16GB mem/node, somebody else is using the cluster you get these 0 to N nodes now, sorry about your data getting out

The right approach here (and one that NASA appears to be taking) is sharing of knowledge on how best to build a flexible cluster environment. It's OK for everyone to have a wheel, we just need to quit deciding if it should be round or not.

I have to say I cringed reading some of the Nebula pages. They're definitely written by somebody trying to sell the project to the rest of NASA. That's the real danger, not that NASA develops it's own cloud but that NASA is so departmentalized that nobody uses it and each sub-section develops their own.

Re:Is NASA suffering from mission creep?

[zeldor]

This is not so much a "data center" as it is going to be a generic compute resource.
Given the computer modeling and simulations that NASA needs to do to support its
"space" missions this kind of thing is very much related to its core mission statements.

Re:Is NASA suffering from mission creep?

[vrmlguy]

NASA is *always* under-budget. Out-sourcing a one-off design to private industry *always* costs more than doing something in-house. Do the math.

Re:Is NASA suffering from mission creep?

[DerekLyons]

NASA (like the rest of the government) needs to be focused on its "core competencies" (no I'm not a PHB).

If you aren't a PHB, then why complain using PHB/Marketdroid fuzzy buzzwords?
 

Where does building data centers fit into NASA's mission statement?

NASA does all manner of things that aren't launching things into space because without doing those things, the things they do launch into space might as well be chunks of firewood. For example, operating a large communications network, or operating a considerable chunk of data processing horsepower. But, you need to read TFA - they didn't develop a data center, they took commercially available hardware and deployed the [open source!] NEBULA cloud managment application on that hardware.
 

I realize that there are tremendous amounts of data that needs to be captured, analyzed and archived (the Terra satellite sends a terabyte of data a day alone I think) but isn't this something that can be done more efficiently by private industry (Google?). Maybe it can be even outsourced providing it is not of a sensitive nature, I mean isn't the data for all mankind?

Actually, given the byzantine nature of Federal procurement, it may not be more efficient to be done by an outside contractor. Doubly so given the even more byzantine web of privacy, access, and security requirements.
 
Yes, the science data is for all mankind, but there is usually a 'hold back period' of a year or two where only the science team (usually from outside NASA) has access to it. This is only fair, as they're the guys who fought for funding for the instrument, designed it, developed it, tested it, operated it - and spent years of their lives doing so. (We're used to talking about 'NASA satellites', but in reality NASA is often just the bus driver and road crew for instruments from outside of NASA.)

Re:Is NASA suffering from mission creep?

[Angst Badger]

but isn't this something that can be done more efficiently by private industry

I know that there is, in some areas, a belief in the unicorns-and-rainbows magical power of private industry to perform more efficiently -- a view that makes one wonder how many of its adherents have actually held real jobs in private industry -- but the evidence is simply lacking, particularly where government contracts are concerned. Ever seen the miracles private industry works with DoD contracts?

The vital difference between the public and private sectors that market ideologues always fail to take into account is that private businesses have to turn a profit. That's an expense that not-for-profit activities don't incur. Given the choice between scientists who care primarily about the outcome of the project and private contractors who care primarily about billing for more than the cost of the project so they can make a profit, it's far from clear that the latter will always do the best job.

Re:Is NASA suffering from mission creep?

[qw0ntum]

It's important to remember NASA consists of more than just Johnson and Kennedy (space centers). Each center focuses on certain areas that are important to NASA's overarching mission. Ames, actually, does a lot of computer science and software engineering research, so this makes sense to do here.

Additionally, NASA deals with a lot of data, for example from their earth-monitoring and climate science missions. There is a huge need for computing infrastructure, and a resource like this that can be shared across NASA centers is a great benefit to researchers across the agency.

Regarding sensitive information: the mindset at NASA (at least among the PHB's) is to be super paranoid about any possible data breaches. PR is extremely important to NASA, and if there were any major embarrassments due to data breach that could result in decreased funding, which as we all know is already incredibly tight. So, by providing services like this in-house, it's much easier for teams at other NASA centers to get access to the resources they need without jumping through hoops to get clearance. I will note that much of this feeling is a result of ITAR, so NASA's concerns are definitely well-grounded.

Re:Is NASA suffering from mission creep?

[AMuse]

To follow up on this (Disclaimer: I am a NASA employee), NASA and other federal agencies are prohibited by policy and law from transmitting or storing many of our data types on non-government owned hardware and networks. (Transmitting of course can be done if it's tightly encrypted). Processing our data on private servers is strictly prohibited in many cases.

The most frequently cited laws and policies which dictate this are FISMA and OMB M-06-16, but there are many others. Employees are even prohibited from doing team collaboration with things like Google Docs, because information which is not yet deemed to be sensitive (say, an immature design for a propulsion system) might become very sensitive, and once it's "out" it is out for good.

Like it or not, there's a lot of other countries with developing missile programs, communications programs and many other technologies which have dual civilian and military use, and NASA is charged by congress with keeping technology that may have military applications out of foreign hands.

If Nebula is able to perform as well as clouds such as EC2 and the like, and allow NASA and other federal agencies to do cloud style processing within the government sector, it could save HUGE amounts of taxpayer money that's otherwise legally obligated to be "Wasted".

Re:Is NASA suffering from mission creep?

[AMuse]

There's another important factor in the paranoia about data breaches and risk that's often VERY overlooked.

As part of the chain of responsibility, the CIO community (the individual CIOs at the 11 NASA centers, and the federal CIOs in general) are very risk-averse. Why might that be? Well, in addition to the normal slamming your agency has to endure if there's a data/privacy breach, the CIOs and decision makers may also be civilly or criminally liable for negligence if it can be shown that they were permitting workplace practices that went against federal regulations. A few CIOs that I know are actually carrying personal liability insurance (out of their own pockets) to cover themselves in case such accusations are leveled.

Now, imagine you're the person tasked with pushing the envelope technologically (Hey, it's what NASA does) but the only thing your bosses ever remind you of is that it's your ass on the line if anything is ever breached, inappropriately stored or transmitted, etc -- and that fines and jail time aren't out of the question. That's enough to make someone pretty risk-averse!

Re:Is NASA suffering from mission creep?

[mcgrew]

I realize that there are tremendous amounts of data that needs to be captured, analyzed and archived (the Terra satellite sends a terabyte of data a day alone I think) but isn't this something that can be done more efficiently by private industry (Google?). Maybe it can be even outsourced providing it is not of a sensitive nature, I mean isn't the data for all mankind?

If you outsource it, you have all the expenses you now have, plus the civilian PHBs' outrageous pay, perks, and bonuses. I don't understand where people get the idea that government always fumbles. Maybe it's the corporations that run the news outlets?

Why would private industry be more efficient? Garrison Keilor covered this in his column today:

Re:Is NASA suffering from mission creep?

[mcgrew]

Crap. Here's the link.

The big lie that Republicans have inflicted on us, starting with St. Ronald, is that government is a morass of inefficiency, and private enterprise is the Enlightenment. (Republicans have practically disappeared from the Snow Belt. I just point this out.) My own experience is that when I go to get a new driver's license in St. Paul, or deal with the city inspector when a sewage line breaks, or walk into a post office to mail letters, or talk to the police when our house alarm goes off, I find public employees to be cheerful and competent and highly professional; and when I go for blood draws at Quest Diagnostics, a national for-profit chain of medical labs, I find myself in tiny, dingy offices run by low-wage immigrant health workers who speak incomprehensible English and are rude to customers and take forever to do a routine procedure. An hour in a Quest office will ruin your whole day.

If the government took over this miserable operation, paid the people decently and trained them to smile and speak softly to the clientele, civilization would be advanced. If we simply extended Medicare to anyone who wished to sign up for it, the vast Kafkaesque bureaucracy of for-profit insurance would come crashing down, and the public would be healthier.

Re:Is NASA suffering from mission creep?

[ArsonSmith]

"a view that makes one wonder how many of its adherents have actually held real jobs in private industry"

Worked in both and private industry is many many times more efficient in every way.

Re:Is NASA suffering from mission creep?

[ArsonSmith]

Random anecdotes do not make data. Health-care is already over 50% run by government and it has steadily gone down hill as that number has gone up. Please look into the budgeted costs of what a lightly regulated company does compared to a heavily regulated health-care company or a completely public program.

Re:Is NASA suffering from mission creep?

[mcgrew]

Knowlege is more than just data, and some data are anectdotal. I don't need a double blind study to know that my city-owned electric company's rates are the lowest in the state, or that the privately owned Amerin's are the highest. I don't need a study to know that I was, in fact, well treated at Illinois' DMV and was treated like dogshit by the cable company.

CWLP has the best customer service, lowest rates, and best uptime in the state for a reason -- if they screw up, the Mayor loses his job. Amerin, on the other hand, is not beholden to its customers; it's not like they can go down the street and find another electric company. They are only beholden to the stockholders.

When I buy drugs, I only pay the co-pay. The drug's price varies from pharmacy to pharmacy, but since I have no financial stake in where I buy them, I go to the closest one -- which is the most expensive.

My employer is my insurance company's customer, not me. With health care, the patient is completely out of the loop.

As to what Congress is doing now, I'm not for it. It's nothing but welfare for the rich once again; the insurance companies will continue to screw you over, and idiots will still think it's free market and better than England's or Canada's and will say "see what government-run health care did?" when it won't really be government-run. Just like they say there were WMDs in Iraq, and that Iraq had something to do with 9-11.

When you put people who think government is always the problem in charge of government, it will be.

Re:Is NASA suffering from mission creep?

[lennier]

"Yes, the science data is for all mankind, but there is usually a 'hold back period' of a year or two where only the science team (usually from outside NASA) has access to it."

Because it takes this long for Majestic-12 to photoshop out all the aliens and boy, are their mouse fingers tired.

"This is only fair, as they're the guys who fought for funding for the instrument"

'I need a hundred million dollars to count all the left-handed spiral galaxies. It will revolutionise science!'

'No I need TWO hundred million dollars to count all the RIGHT handed spiral galaxies. The leftists must not triumph!'

'Gentlemen, you both make good points. Shall we decide this by asking the person whose money it actually is? The US taxpayer?'

'....'

'Haha, no of course not. The usual procedure it is, then... rocket pistols at dawn, Pad 39-B.'

Re:Is NASA suffering from mission creep?

[ArsonSmith]

Calling both power and cable companies private and or free market is kind of a stretch.

Re:Is NASA suffering from mission creep?

[smoker2]

Your comment was fine up until the last paragraph, then you made assumptions which you have no evidence for. Please make your assumptions fit the actual truth. Even if the truth doesn't fit, maybe it's something worth aiming for. Instead of the crap you have been used to up 'till now. This is what govt. was designed for...

<Now for a ramble> I've been drinking in a pub which has a new landlady. She's got nice tits, but this is irrelevant. One of the first things she does is put all the condiments in to buckets, for each table. I tried to point out that this meant MORE work for her staff, but she prefers to see it as "doing more for the customer". I cannot understand this. Sure we get to decide how much yellow crap we squirt on our food, but we can no longer guarantee what the yellow crap contains. Anybody can sneeze or cough on the bottles and it will never get cleaned off. At least with the individual sachets, you used it and no-one else did. But apparently, because I haven't done a 2 year management course, my opinion is worthless. Never mind that a) I'm a customer , and b) I've been drinking in pubs for the last 30 years. 2 Years management course rules the world ! Kids today !!!
We'll never get any credit when it all goes to shit and they have to come crawling for advice, will we get any notice ?

No, the world would rather die than listen to advice these days. As usual, it says more about you than it does about us. The world today is all about instant gratification and fuck the future. In case you haven't noticed, your future is our future too. Don't fuck with MY future ! [ or I'll fuck with yours ! - I mean it ]

I don't get it.

[filesiteguy]

So you put some servers in a shipping container.

What does that bring you aside from the ability to move said servers around? If you need to access a remote data center you simply dial it up.

I can't wait until they put one of these in a double-wide and the next tornado spots it.

Re:I don't get it.

[drinkypoo]

Shipping containers are stackable. Remember when blade servers were introduced, and many people said "no one will want those, you can't even install per-node peripherals"? These are the blade server equivalent of data centers. Also, all ISO shipping containers have convenient mounting points, so you can protect these from heavy weather by sinking piles and just bolting them down. They even have the potential to be watertight, so if you built them cleverly enough, with air intakes and exhausts bussed through them and with hoods on the roof of the top unit, they could even be flood-resistant in such a scenario.

I have no idea if any of that applies to this system, but there's good reasons to use containers.

Can You Imagine a Beowulf Cluster of These?

[vrmlguy]

Hello, old meme, long time no see.

They're using Eucalyptus

[Xouba]

The best thing about this is that they are, allegedly, using Eucalyptus: http://nebula.nasa.gov/blog/2009/nov/how-eucalyptus-enables-ec2-compatibility-with-nebu/

HOT WHEELS!

[abbynormal brain]

We'll know they *really* made it when we see the die cast replica.

Nothing new?

[Anonymous Coward]

Am I missing something here or is this not what Sun has been offering for a while now?

http://www.sun.com/products/sunmd/s20/

...Cloud computing?

[MathiasRav]

Huh? Oh well, better get back to my astronomy assignment. Err, I mean Cloud Study.

Why a modular data center? Yes!

[rlh100]

Do you know how long it takes to build a data center? In most cases it is years and years. And you have to build it to your eventual maximum size. Guess to small and you have to build another one. Guess to large and you have a lot of expensive floorspace going to waste.

With a data center in a shipping container you and build it in less than 6 months from first thinking of need and having it up and running. Yes it will take another 3 months to plan it, get the funding, find a place to put it and install it.

You can buy your data center in 3200 sq. ft. increments. Maybe have a plan to add one every six months.

You can also depreciate it as equipment rather than facilities.

And finally you can move it on a whim. Tornado in Florida, send it overnight to the Midwest. Flooding in the Midwest send it to New Mexico. To hot in New Mexico, move it to Utah. Cheaper power in Washington state, move it again.

It would also make sense to containerize your chillers as well so you can install/move them both. Then all you need to move it is a flat secure location, power and Internet connectivity.

Of the containerized data centers the sgi (Rackable Systems) ICE Cube Modular Data Center is, IMHO, the best design because they deal with power efficiency in the rack with DC/DC power supplies and they talk about containerized chillers as either part of the container or as a separate container.

Take a look at:
http://www.sgi.com/products/data_center/ice_cube

RLH

Why is NASA doing this?

[Animats]

NASA needs to focus. They act like they're the National Science Foundation. They shouldn't be doing general R&D.