AWS re: Invent ENT 204- From Science Fiction to Reality- NASAs Journey into the Cloud

I am now so turned on and that’s not a pretty picture thank you all for coming and for having us stand between you and drinks since you shouldn’t drink we’re going to take forever Oh we’d have a drink with you so this is quadra shams he he’s now a rocket scientist I am Tom Soderstrom I’m the chief technology officer at NASA’s Jet Propulsion Laboratory we’re also Caltech were also federally funded research center just ask me and will spin will be anything you want us to be but what we are is we do lots of computing and what we wanted to talk to you about today is this is this is the thesis right here what was science fiction yesterday is evolving into today it’s now reality on the edge of reality but what is it going to be tomorrow and cloud has everything to do with this so that’s the journey we’re going to take so we will start by talking about science fiction yesterday then how we’re evolving where is the edge of reality and where it’s going and we will leave time for questions so you can ask all your technical questions then what we wanted to do is lift it up a little bit it’s not cloud it’s not just about servers it’s not just about storage it’s a whole lot more it is telling taking us into the future did you see the keynote this morning do you want to drive that Rover on Mars quad I can get you there so quadra let’s let’s talk about mobile science fiction yesterday all right so high level description of science fiction yesterday we always try to draw a parallel between us and night in 1880 so imagine yourself prior to any time between before 1885 and then put yourself in the shoes of a small business owner and a small business owner whose business relies on electricity to generate the prod that you need to sell and in this business owners mind if you were given a large in 1880’s you were actually you know generating your own electricity if you wanted to run the small business and putting yourself in the shoes of the small business owner if you were given a large order right around Christmas time you had two options ok you can go invest make a capital investment in a larger generator wait for the generator Toby arrived wait for the installation to be done wait for the calibration to be done maybe go even out for a little bit you know have some outage to do the installation process or you can refuse the business right because technically you’re going to make this capital expenditure and then this generator is basically gonna sit idle for the rest of the year until hopefully next Christmas you’ll get the same competant amount of business again now in 1885 Thomas Edison transformed the way we approached electricity Thomas Edison opened the gates in eighteen eighty five to the Pearl Street power station and with the power station he allowed us to think of electricity as a consumable commodity he allowed us to think of it as something that we can get on demand as much as or as little as we need and to get and pay for as much as we use right it is the same model he basically did modern plumbing for electricity and it allowed us to think of electricity the way we think of modern plumbing cloud computing to NASA to us is that same model apply to your compute storage and network needs it is the ability to provision capacity on demand get it as quickly as you need it turn it off and stop paying for it when you’re done with it that’s what cloud computing is to JPL so just like that was science fiction in 1880 in 35 years ago when we launched what spacecraft at this Voyager all right we got some old people in here it it was science fiction and in fact it broke within a few years so what do we do we reprogram the thing over and over and over and just extend its life so that’s part of the story here is to how do we keep evolving but it entered a new dimension where is it is it everybody who thinks it’s outside the solar system raise your hand everybody who thinks it’s on the border raise your hand so I helped my hand off a boat it’s entering a new dimension and that had less IT on it than you have in your pocket in fact thousands of times less so going towards the new reality science fiction ten years ago reality as we saw today it’s driving on Mars now what science fiction next but spacecraft is this why am i showing it because we’re

Enterprise IT and just to put this just to put this in context three years ago Tom and I met with the team the Cassini team which is basically an orbiter around Saturn at this point and we started describing to them how we can provision capacity on demand and we can make our servers we can have you know once server right now and 5,000 the next moment and the scientists project scientists looked at us and exclaimed wow that’s science fiction these are guys that are running spacecraft on Jupiter and they’re calling cloud computing science fiction that’s really where you know we’re coming from here so so let’s start the journey that was science fiction we’re going to create this spacecraft called Enterprise IT how many people are from an enterprise here or more let’s say more than 500 people in your company Wow okay so this is going to be very relevant to you it’s also going to be relevant to all your startup’s how many startups do we have in here because you’re the ones who’s going to make it happen you’re going to ride on top of that large infrastructure that you can create and dismantle in seconds but you’re the ones who’s going to create the new businesses that’s that is the new reality so let’s take a look at evolution JPL is all about evolving we are nestled in the foothills of California’s you heard this morning we’re into robotic spacecraft everything that doesn’t have a human on it is a robot to us would it be an arm or a spacecraft you heard about curiosity that was just the last of many missions that we’ve done this year it’s been a very very busy year fact 10 times busier than normal so we’ve had a lot of successes the next successes will come from faster innovation less budget more IT so here’s how we do business today this is an IT crew very fine-looking IT crew they we’ve done the budgeting we assemble them and now they’re ready to go to work to build that infrastructure we need to transform that into a way where we what you create the infrastructure as we walk on it and what is that come on cloud computing so we also have to transform how we do business it’s no longer about coming up with the ideas getting them vetted getting the budget build them it’s about rapid prototyping it’s about trying things and failing Thomas Edison was asked how come you failed so many times and he said I didn’t fail at all I learned 1,000 ways of not creating a light bulb and that’s the mentality we have to do and cloud computing will help us get there so let’s take a look at today’s reality let’s see what was science fiction just a few years ago and where we are and this is all about cloud computing our cloud journey started with Jim Rinaldi who’s the CIO JPL on my boss said I’m tired of buying computers I just want to rent them he said it in these nicer words but that’s what he meant how can we do that and jonathan chang raise your hand was one of the initial victims he’s a developer and he said okay Jonathan you can’t buy any more servers you have to rent them from the cloud and Jonathan grumble and eventually he got used to it and it’s been very good so let’s take a look at the experience quadra how can we try cloud computing immediately and securely and this is for years ago by the way when we said okay we’re going to do this so Tom and I work with a variety of missions you work with a lot of missions across NASA and some agencies actually beyond NASA as well and in our journey we come across a lot of prevalent myths in terms of cloud computing and the number one admit that we hear is that cloud computing is not secure you can’t run sensitive applications on the cloud and we started evaluating this myth a while ago in 2008 actually and we learned some lessons in this process we read first of all from Bruce Schneier that the biggest risk in cloud computing the biggest fear in cloud computing that the public holds is having to trust a third-party entity but Bruce Schneier very profoundly asked in that same blog who we already trust do we not trust Microsoft with our operating systems do we not trust Cisco with our firewalls it is an important question to ask it’s not something radical to have trusted third-party entity what did we do when we did when we trusted operating systems with Microsoft we evaluated the risk we evaluate the cost of you know not doing this and writing our own operating systems and some people would have argued at the time that that was the right thing to do but then we realized that we can get a strategic advantage by using these capabilities and we could mitigate the risks sufficiently to take advantage of these new capabilities so you know that gave us some inspiration a few years ago and we started looking at cloud computing so we started trying cloud computing in a very open environment we have a lot of public data our Charter is to get our data out to you as quickly as you possibly can so we put we built a bunch of public outreach applications and we put them across multiple cloud vendors so we put them on Microsoft Google Amazon as well as some of our private cloud solutions and we learned a lot we were able to train our developers on the new cloud paradigm while learning about the security

implications and how to effectively secure applications in the cloud and very soon we realized that you can actually do secure computing in the cloud and Tom will actually go on the record saying that in many cases cloud computing can actually be more secure than your internal infrastructure and you can tweet that and the reason why it can be more secure they enter in your than your internal infrastructure is because of the unprecedented transparency that cloud provides you over your infrastructure it is the ability to list all of your instances with a single API call it is the ability to respond to incidents by API calls and saying you know plug this firewall hole it is the ability to define dynamic to dynamically define subnets they enable collaboration across different network it’s the ability to isolate machines as soon as you think that they’re compromised and it’s the ability to uniformly deploy lots of machines with a single am I so the other thing about cloud computing where so I mean enterprise IT I’m here to make your job better we have to make it simpler I mean the future is easy to find easy to buy easy to use and easy to lose that’s not infrastructure so what we did at JPL is we haven’t single on-ramp you type in cloud computing it in your browser and it brings up the chasm the cloud application suitability model which tells you based on the quest answers you give the questions where it that application best lives so it makes it easy and you can charge a normal charge back model that we do at JPL that’s important if you’re in corporate IT or insecurities if there is a security breach guess who’s responsible anybody yell it out the CIO the local CIO is responsible whether he or she knew about it or not so that’s why I T needs to be the enabler but make it easy to use and this is one way of doing it and when the choice points to a commercial cloud we also have a catalog of hardened images that our IT security team has built which allows us to foster adoption of secure the best security practices it allows us to IT security to provide oversight and guidance as to how to secure your applications rather than our developer going off and installing linux on a desktop under his cubicle and running it as a server and by the way that will never get past I promise so Tom how can you try mission cloud computing for mission critical operations well kwacha there’s a myth that the cloud is not reliable so what does that really mean let’s examine that man would you so I think actually cloud computing allows us to get as much reliability and availability as we need and at JPL our approach is to actually define a spectrum along which we we we can control the amount of ale ability for our application so let me explain the spectrum to you if you are running thousands of machines you will see every once in a while in instance failing this will happen in your own data centers this can happen in the cloud but there’s a lot less likely to see an entire data center entire availability zone fail it’s even less likely to see multiple availability zones fail but that has happened right so if you want the ultimate security you can get but you can build an application with multi region availability and the reason why cloud computing is helping you here is because you’ve got a uniform interface to deploy these applications across data centers around the world and if you’re really good if you’re as good as Netflix you can actually have a chaos monkey that’s a very surly monkey according to according to URI who only works from nine to two and randomly turns off your machines to ensure that the developers and netflix adhere to the best availability practices and their infrastructure is always available and why not why shouldn’t we at NASA be thinking like that as well there are physical aspect of the world that we just can’t control here in the foothills of Pasadena which means that we’re in fire County so this is the station fire which came very close to JPL within inches of JPL which meant the whole city was actually evacuated for a while now the scientists running our applications from Cornell or around the world in Spain don’t really care that JPL is on fire they want to get access to their data because they’ve got a paper deadline right you don’t you guys remember the angry professors know you you don’t wanna mess with them so we’ve also started using cloud computing to augment the availability and durability for application in fact the Mars exploration Rovers are archiving dozens of terabytes of data all the data that we ever collected and sent to Mars to the modulation rollers is backed up today on the cloud through a orchestration process powered by swf and storage durable storage provided by s3 but backup is easy recovery is hard so we actually also audit these capabilities by routine by randomly picking files recovering them and checking if the if they have changed on the system where if you have lost anything more importantly in this

process we have built a distributed cronjob system anybody dealt with cron jobs recently alright so the cron jobs are evil because they’re so simple you can set them up and say okay every day at five run this particular task and soon enough in a mission-critical environment people will come to rely on this file that you’re delivering a 5 p.m. every day and then before you know it you’ll retire that machine and all send this file stop showing up and you have no idea why right this machine went off well you’re not going to put the two and two together because you’re panicking somebody’s calling and saying hey this file isn’t here eventually you’ll figure it out and then you’ll go well I don’t have to script anymore what do I you know so with swf we have actually built distributed cron jobs that run across multiple machines it allows us to have this job occur even if the particular machine is down it also allows us to get a centralized dashboard to see which jobs are failing or timing out so that we’re not running across 50 machines Oh which cron job is doing this and you know what if the order in which things are happening you can actually very succinctly express complex workflows and see them in a centralized dashboard and that’s what we really like and what I really like is there is no people harmed in this process which means it’s inexpensive it’s word so we love Netflix for coming up with ad of the chaos monkey because you can build reliable systems out of unreliable components it’ll change the way we write code that’s all right you guys are coders you can figure that out more importantly it’ll think make things faster less expensive and a whole lot more reliable so for instance how do we go to secure uplink planning in November 2010 we roll the dice with Jon Callas who’s the project manager for Mars exploration Rovers and how did we roll the dice one of the things that we often run into is people said oh it’s all right for you NASA guys you have tons of money we can’t afford to do cloud and that is so backwards first of all we don’t have lots of money secondly all you can afford it so what we did in this case Mars exploration Rovers were supposed to last 90 days eight years later all the IT systems were full so we said okay Jon Callas we can do it inside or outside either way it’s going to cost money what do you want to do and we did it outside in the cloud and we saved one order of magnitude on cost and we gained to order magnitudes on response time for mission critical operations it’s encrypted it’s all that good stuff that you would hope we would do but it’s it really showed us the value that we can use it on mission-critical computer that application by the way has got one hundred percent availability since 2010 which is something that we’re very proud of because we’re using very highly available services in the cloud that sustain you know AZ failures very very effectively without us having to do a lot of engineering around it what about high performance computing what about supercomputing reserved for the rich no way that college student could use it right quarter I’m not true recently NASA headquarters are funded us to actually evaluate the efficacy of cloud computing for high-performance computing and we ran some benchmarks to realize that we can run about you know for thirty-five dollars an hour we could run wait alright for thirty-five dollars an hour we can actually run a you know 12.7 teraflops of compute capacity put this in context a top 500 computers about 50 teraflops so in today’s spot market prices you can actually provision a top 500 supercomputer for about a hundred dollars an hour right now put yourself in the shoes of a graduate student who’s got a paper deadline with armed with the armed with you know basically top 500 supercomputer at his disposal with a couple hundred dollars pocket it’s going to transform the way we do science today and it’s crowdsourcing of the best kind all of a sudden we can get everybody to participate to build better government or your own company whatever happens to be so Tom cloud computing should be as simple as using a credit card rate so in fact in 2008 it was that’s when we started doing this and what we discovered is that we are Enterprise IT not so easy we can’t do credit cards we do charge backs we are pretty big organization so how do we deal with that well that’s easy anytime we need a contract and to get a contract we argue a lot anybody got a lawyer at the visitor is anybody lawyer in here oops no that’s good that’s good is there an auditor in here oh I’m so sorry I used to say it’s their 90 security person in here and people will go okay I said everybody else hugged that person because that is your future the auditors and lawyers they are our future to really take advantage of this so thank you for being here so we did agree on it it took over eight months and we learned a lot we learned about what matters and what doesn’t matter and we happy to share those stories sls are almost meaningless for us if we lose science data that cost millions of dollars to get back and we get ten cents on the

dollar back that’s meaningless three bikes and their app and we’ll go to a different cloud provider that’s how we look at it so for instance we had one hundred percent availability on Mars rovers it was only advertised as an SLA of 98% if we were to go and say that we’re going to get a vendor that has twenty-eight percent of 98% we couldn’t do it so we have to think differently so what we recommend is that you big IT enterprise IT create a cloud computing commodity board and it’s the the commodity is the important part cloud computing is like electricity you can just buy it if you have a big enough contract you don’t have to argue every single time and that’s what we did so we actually have several different cloud vendors course we love amazon but we have other cloud vendors to all right they’re not playing how can we use cloud computing for really hard things quadra it all depends what the next slide is um so well the biggest miss that we have come across as cloud computing is only meant for storage and processing and that’s also not true sure we have used cloud computing for radar data processing where a scientist came to us with a paper deadline and said well I started this process and it’s going to take a week and a half for it to finish on my local infrastructure can you please help me make it you know finish faster so we coat it overnight and then we launched a machine a set of 80 machines in the cloud and specifically in Gulf cloud in this case in four hours later the scientists had all of his results yeah three gig of wonder gigabytes of input data four terabytes of output data that he needed to examine just a little part of we did this overnight we wrote the code four hours later we executed the code it was done and the scientists got his data and of total cost of 256 dollars that was a real empowerment first like and basically deal with a big strategic advantage to our scientists was able to actually make his paper deadline but again that’s the processing all right that’s processing and storage why what would cloud do other than processing and storage well the fact the matter is that the cloud vendors have been in this business of making highly available systems highly distributed applications and highly performant applications for a long time right and they have basically started to create an ecosystem that will streamline the development of your applications so that you don’t have to go figure out how to run a high availability load balancer so that you don’t have to go figure out how to run your own orchestration service and operate it and you don’t have to go figure out how to run your very highly available databases so for the Mars Science Laboratory we use a slew of services to expedite our processing needs so we use swf for the orchestration ec2 for processing s3 for the storage and delivery of the system simple db2 index these plans and images cloud wash to monitor our instances and get a text message anytime something funky happens to our system I am identity access management to control access to our sensitive resources and load balancers to develop and deploy highly available applications now I imagine if you have to figure all that out from scratch imagine the time and was taken our developers to go figure this out and by the way we would have figured it out but it would have taken us a lot longer and we probably would have made a lot of mistakes that we don’t have to anymore and then in addition to that we would have had to maintain that code make it building something is cheap and easy the lifecycle maintenance is what really cost money so this is this is the future not us not having to develop it here’s another use case of you know so some things are perfect for the cloud some things just don’t fit and that’s what we’re about we’re not putting the right thing in the right place this is the right thing in the right place it’s a spacecraft that once a month we’ll go over the North Pole and look for carbon lots of compute power needed for that flight that aircraft will fly it’ll land 30 days later it’ll fly and land so we can spin up lots of computers spin them down and it saved two-thirds of the IT costs doing it that way over what we normally do this is the future this is truly the future the hybrid cloud so the hybrid cloud means many things to many different people to us what it means is that we can spin up a small set of servers and compute tasks in our own environment they spin up their equivalents in a cloud environment like Amazon’s they do the job using all that infrastructure we’re renting temporarily they spin down and we have it inside as a backup should that cloud disappear so today it’s amazon its many different services of amazon it could be other clouds too so when you think about your architecture think about the hybrid cloud and we tend to believe that maybe we have a private cloud today but maybe private cloud so I’m not going to be needed maybe they will all be secured by you all and run on a large vendors infrastructure because after all it’s just water it’s just electricity and if you’re a developer itching to convince your management to try out cloud computing imagine the use case here in case the cloud isn’t available it would be no worse off than if you were just relying on the internal infrastructure but for the days that cloud is highly available which is almost every day you can augment your internal capacity and

pay only for what you use and get your results a lot faster alright so we’ve already talked about that one so this little thing called curiosity you heard about it this morning and quadra is not going to get quite set quite as excited this time but he’s going to talk about what happened i’ll talk about this part our world turned upside down we had two weeks curiosity has been in planned for years and years and years we think of everything that could go wrong we try to mitigate it there’s two weeks before it said okay this is your Olympics this is your super bowl you have to stream it to the world all right and by the way this is two weeks or actually 10 days before the rover actually landed so every developer at JPL was basically working on every developer on my team was basically focusing on making curiosity work and completely or subscribe and bread can you raise your hand Brett shore stepped in and worked very diligently for the next 10 days to develop develop and deploy a system that could sustain the bandwidth limitations that we had the bandwidth constraints and the bama de quirements that we had and we worked very closely in this case to deploy this cook this application we worked very closely with folks at amazon in fact jami kenney and myles ward for amazon were at JPL 4am the night before landing making sure that all the pieces were in place and that we could actually sustain this kind of bandwidth the paint wasn’t even dry yet um so what do we do we started with cloud formation stacks that had layers a two-layer fan out cashing architecture tied together with an elastic load balancer in each of these stacks could do 25 gigabits per second of traffic all right well we add it we need to do 50 we added their staff we had it to do 75 and keep going so now we’ve got five stacks that can do 125 gigabits per second of traffic and in total but how do we tie all these together well we simply use RAF 33 and used weighted round-robin to actually load balance across the load balancers and the fundamental advantage here was that we can actually get multi-region support for all load balancers if you wanted to go really advanced if he had 11 days instead of 10 we would have done latency based routing mix with weighted round-robin so that the people in LA go to the elastic load balancer in San Francisco while the people in New York go to the elastic load balancers in Virginia right you can do all of that in a matter of a you know a few hours just by making you know reading a couple articles and making a few clicks on your console and you’re ready to go imagine what this would have taken prior to having this kind of a dns service in the past meanwhile elastic load balancer is also monitoring the health of every individual node and taking them out of commission as soon as they start responding stop responding to the health checks furthermore auto-scaling groups are then protecting the number of nodes that are available in each cloud formation stack so that if you know we had out of auto scaling groups with minimum and maximum number of nodes to be exactly the same so we weren’t using it to actually scale we were using it to maintain so if you know one of the 40 notes goes down it just replaces it with another 40 so we ended up deploying thanks to Brett’s heroic work we ended up deploying a scalable architecture that could sustain the failure of the entire East Coast and actually many other data centers and still deliver video so we ended up doing more than 150 gigabits per second of traffic any thousand requests per seconds in a matter of a couple hours we served over 100 50 terabytes to the public and we have some very happy customers at the end that’s a Pete I singer and Richard cook the project manager and deputy project manager of curiosity to the smartest people we’ve ever met now remember all of this in breadth raise your hand again Brett will be the first one to admit he had a lot to learn he’d never even have a clue how to do any of this two weeks before so it was a little nervous but thanks to amazon it all worked out and to Brett’s hard work this is curiosity now on Mars the self-portrait and it’s healthy it’s looking for signs of life and we’re looking forward to see if it finds it so cloud computing is not part of everything that we do four years ago it was cloud computing the experiment today it’s the pervasive cloud it’s part of everything that we do so what will the future look like so these may not be having a drink on Mars but it’s really important and how do you do that we all talk about disaster recovery it’s damn expensive so most of us do we hand wave at it well all of a sudden you can do it you can build it into your enterprise architecture and you can build you can use glacier and you can use things like Pandora I see the penciler CEO here where you make it really that we do use that at JPL so it looks like it’s local storage you make it simple and it puts it in the least expensive and most appropriate place so this is an actual use case cost modeling

of a real specific use case of storing 70 terabytes of data somewhere for disaster recovery purposes the red line is glacier now this is where you developers and smart people come in it isn’t just that you put it there and bring it back you really have to think it through it depends on how fast you need it back so really go through the use case and think about it so you can fine tune it and there’s different strokes for different folks glacier is really good for long-term storage and it’s really good but if you bring it back all the time you pay a lot of transfer chargers so architecture think it through when you couple that with the new gum cloud we can have a really secure and I tar export control environment so we’re quite excited about that now here’s another one what about finding any information at any time any information and what does cloud have to do with it so we built a prototype and the prototype the science fiction is this you want to be able to find we’re interested ion propulsion you want to be able to find wherever somebody said the words ion propulsion whether it’s spoken or in a video or typed so was that possible we partnered with Microsoft they had something in the back shelf we tested it and they commercialized if we can use it and we ran it through Amazon to get it started in the cloud and now we built a prototype an asset to it’s just for internal private very sensitive videos now where we’re going to store it so we were waiting for the infrastructure to spin up and eventually we just put it into gov cloud overnight and it sits there and it works and all of a sudden we can find those information the other benefit anybody heard of 508 compliance anybody from the government 508 compliance is the hindrance to putting videos on and we know from YouTube generation that videos is the future short snippets of videos that tell you everything you need but they have to be transcribed so this does automatic transcription it just comes as part of it so all of a sudden by using gub cloud it’s secured by using cloud computing we can put it there and by using actually two clouds in this case Asher to Amazon gub cloud it’s completely transparent and completely automated so that is real key so what about the workforce how can you have an unlimited workforce but for us it’s about not having to build it inside but being able to have an environment where we can bring in lots of other partners what does that look like anybody participated in the international space apps challenge all right thank you one person give him a hero’s applause thank you thank you anybody ever participated in hackathon okay one two three and a half four that’s really powerful and what do they all have in common they’re all volunteer they all use cloud computing and they all use the latest technology usually non-structured databases etc that’s our future we posted 60 problems and at the same time on all seven continents there were people programming over the weekend came up with a hundred solutions that we’re not gonna use all right what does SaaS mean say it louder what does it mean to us put it Tom and I stumbled upon a little secret students as a service um we mentor dozens of students you know as part of our job to identify the future leaders and identify the next generation workforce and as part of this cloud computing involves a lot of prototyping sweet are able to hire these students and actually one of our students is here in the audience Ben Schwartz when you’re through Maryland and these students are actually providing really compelling results to us in terms of you know proving the efficacy of cloud computing and then building real applications that were then immediately integrating into mission-critical applications so on the on the screen here you’ve got andres rio frio and alex curran alex curran was a senior in high school he worked on the mer backup project that you saw the data archiving project andres rio frio freshman at UC santa barbara worked on the carved project which is the aircraft that flies once a month now these interns are able to provide real proof to us that we can actually put these applications in the cloud we get to mentor them they get to give us real code and they’re really cheap so everybody wins but so so this is a case of these high school student and college student giving a presentation we do IT innovation seminars to all of JPL not just to the other students and they just rock the house it was fantastic so students as a service don’t ester underestimate what they can do and you can’t afford it yeah not to demean our developers but we actually learn a lot from the students because they have such high levels of energy and just get been started on talking about anything and

just see what you learn a fun bit so so Tom how can we collaborate better in the future no thank you for asking let me turn it right back at you back how can the scientists collaborate better in the future alright so NASA just wanted us to actually build a collaborative environment powered by the cloud the typical open source model for scientific computing today is that scientist and by the way these aren’t computer scientists are getting open source science software in as a in the form of source code which means that this geologists physicist chemist is going to download a binary source code file that’s zipped and then they’re going to try to compile it and oh wait there’s dependencies oh wait the dependencies are dependencies oh my god I got the wrong version and they’re all of a sudden wasted three days trying to figure out how to install this software and in this process they might have actually broken the other dll files and the SOS that they have and they really don’t want to be dealing with this so what is cloud computing doing to help us there well instead of difference emanating this software as a source code we’re now able to disseminate the software in terms of virtual machine images this means that the applications can be executed in the same environment that the developer tested and developed in this means that we have streamline access to these algorithms and different versions of the algorithms this means that we have co-located the data with virtually limitless capacity so that scientists are no longer waiting for the terabytes of data to transfer from Virginia back to pasadena it means that they can get started fast they can access date a fast and they can store the results fast and they can collaborate better than they could ever before if you tie that in with a crowdsourcing and the students that are getting their PhD somewhere getting access to the data that is today’s science fiction it’s going to be tomorrow’s reality well we will have many many more inventions innovations scientific results but it’s not limited to science it’s also travel dollars are expensive video conferencing is that we are at least my generation not you young people we have relationships face-to-face that’s how we create trust so video conferencing is really key problem is there’s so many of them so how do you deal with it you buy the MC use and that do the translation or you use the cloud to do it for you all of a sudden it’s MC you as a service I just made that up but you don’t have to pay for it anymore or you have pay for it but you don’t have to buy the big infrastructure the other way but you’re still tied to a wall for video conferencing so let’s take a look at what we think the future looks like and unfortunately you won’t hear them I’ll talk over them but these are to JPL could people working at once yeah they are to JPL working from wherever they’re working and they’re driving the robots they can see and people can see them and they are driving to meetings and they’re participating people login whether we had some people who were at sick they could log in they can drive around themselves they’re a little bit of problem with stairs but they’re light so somebody can help them we’re working on a quadcopter model but it’s it’s very effective in fact any consultants in here raise your hands okay the consultants have found they get more face time with their customers when they come in through the robot well imagine coming in via via skype chat where it says hey Tom hey I need attention right now so if we do it right we could our collaboration with videoconferencing and robots maybe one day we could get some help from the locals and being helpful neighbors we would help the locals right back disclaimer these images are not real remember the robot arm taking pictures uh huh huh now how is it easy so how can you easily access it from any device anywhere of course we’re talking about mobility and do we have any mobile developers in here all right the future will truly belong to you and we want you we build games for everything we do but it’s it’s not the real time games it’s to keep you informed about what NASA does but we also use things like we’ve got large-scale high-rise images satellite images from Mars and the moon and the earth even and instead of requiring the scientists to download the entire 100 gigapixel or multi-hundred gigapixel panoramas and models of the terrain we actually tile these models into small tiles at every resolution so the scientists are downloading exactly what they’re can get much like the way google map works on your browser or your iphone and we’re able to do all this processing in the cloud very rapidly without having to worry about you know the time it takes so now we’re talking about science fiction we’re talking about the future architecture and the next 20 seconds will show you what it looks like so the way this came about with we had one scientist who’s wanted to buy an ipad and i said nope not gonna let you and he goes what i said i’ll buy you an ipad if you can create something mission-critical on it and of course he

did so what what does it mean to think about it so all the data i’m going to show you all the data that’s stored on them about the moon pictures from the 60s and on there in one place in a cloud guess what’s cloud Amazon’s Cloud and they will access it from any device a mobile device anything Android’s iphones etc and they will take advantage of the capabilities of that device pinching swiping all the heavy duty processing happens in the cloud when they need more sensitive data they log in on that device through VPN into our private cloud and all of a sudden you now have the same application serving two different purposes what happens if they lose that device use security people in here do we care now we buy a new one the data doesn’t live on the device is just display that so technically the data isn’t stored in amazon’s cloud it’s stored in JPL’s cloud with an extension into external data centers so he’s smooth he’s the next CTO so here he’s pinching and swiping on the ipad he wants to look at it in 3d this is actually looking at where the phosphorus is on the moon to imagine the astronaut walking on the moon with an any device and that’s very powerful and it is that fast because of we’ve using Hadoop to slice the images beforehand and you can get access to every single image ever acquired for the moon from an orbiter or a lander by going to LMK MP nasa gov can try it on your browser so now the final one can we go beyond software and of course we can 3d printing as a service this was two days before this press conference he had no idea he’s going to have to show a model of the heat shield of MSL so he we had our mechanical engineers design it will take the model and email it to a 3d printing house that sent it back so within two days they had several copies of this so now what we have is we have cheap 3d printers where they can explore they can come up with ideation and then we send the ones that are promising to somebody was a more expensive printer so all of a sudden for two hundred dollars we can get a really good 3d printed copy another version is augmented reality and who has an iphone or an iOS device in here raise your hand keep your hand up if you have a spacecraft 3d downloaded ok very few some do I will show it in the break right after it’s quite amazing and it’s our way of reaching the public but it’s much more than that it is again the future it’s how we will do engineering it’s how we will build our models see if they match what was prescribed be able to interact with it see what parts came from Larry what the failure rates were etc etc it ties in big data it ties in cloud and it ties in the mobile device so this was doing the spacecraft 3d 3,000 miles apart where the tag was using video conference so he’s holding Hubble here at Goddard so now we’ve taken the journey from science fiction to evolving and what’s going to be tomorrow’s possibilities so what we want to do is what do you do after this what are y’all going to do go to happy hour right so at one point in the future we invite you to happy hour with the locals again not real it is our most capable Rover yet underground their ice wine so thank you for listening to us and we will be here for a while we have time for a few questions I hope yes