At the recent Consumerization of IT Conference in San Francisco, I was in a session presented by Tom Soderstrom, IT Chief Technology Officer, NASA Jet Propulsion Laboratory.
Many people were attracted to his catchily titled presentation, A Multi-Cloud Model and the Mars Landing. I was one of them. NASA is no stranger to cloud computing, as Openstack was conceived at NASA and based on its Nebula Cloud.
Nebula Cloud
Tom Soderstrom
It has been said that many smaller companies embraced public cloud computing earlier than large companies; that is the enterprise market. NASA JPL is not super large (5,000 people and $1.5B in annual contracts), compared with Fortune 100 companies, but is not too small, either. Soderstrom’s presentation, which consisted of a few use cases at research facilities like JPL, was a good tool to show how cloud computing can be exploited more in many market segments and sizes.
One thing he mentioned about future trends was interesting. It is summarized in his post as SMACK.
SMACK stands for:
Social (social networking moving into the enterprise)
Mobile (needs a mobile-first point of view)
Analytics (actionable part of Big Data)
Cloud (where we will run the service or application that we’re building or buying)
Key disruptors (will a given technology change our lives?)
Soderstrom went on to discuss how almost all the devices we use are connected to the Internet now and more will be later, and how that would increase the amount of data generated and exchanged. NASA also rides that trend and needs to deal with it, as shown in the next slide.
One example is that the Lunar Laser Communication Demonstration, a laser-based communication system, is installed, making it possible to send and receive much more data.
Soderstrom then talked about clouds. NASA is an early adopter of cloud computing. They have reviewed and used the cloud computing solutions shown in the following slide. It would be nice to get information for those who follow them about how they found each cloud solution.
He pointed out some inhibitors to the adoption of cloud computing. There are some myths about cloud computing’s security and reliability that slow adoption.
Security: To date, the number one reason for slow adoption of cloud in the enterprise market is security worry. Soderstrom, like other experts, dismisses this. Cloud providers like Amazon, Google, and Microsoft spend millions of dollars to tighten the security of their data centers and the cloud services they provide. At NASA, he classified data into several categories in terms of sensitivity and tried cloud computing on them gradually. He joked that cafeteria menus do not have to have the same level of security measures as top secrets.
Reliability: Amazon’s AWS and other clouds crashed several times in the past and some people insist that cloud computing is not reliable. Cloud computing is distributed computing with many moving parts. Each part could fail independently of others, and the failure could propagate to other parts for further failure. That makes it fundamentally unreliable. If cloud computing is executed without this in mind, it is unreliable. However, cloud providers have redundancy embedded into their computing infrastructure so that, as a whole system, their cloud computing is pretty stable and reliable.
Soderstrom refers to Netflix as an example. Netflix has made their software open source, and it is quite robust because it uses Chaos Monkey to make it very reliable in the face of unforeseen problems. While it is running, Chaos Monkey can disrupt any portion of its operations randomly. Software is designed and constantly modified to recover from such unforeseen disruptions.
Two more interesting uses for cloud computing at NASA are backup and hybrid clouds.
Backup: A cloud can also be used for backup, and backed up versions can be compared with local copies for verification.
Multiple clouds: In addition to public cloud, people have started to talk about private and hybrid clouds. Soderstrom talked about how hybrid cloud was used when the Mars landing was video-streamed. The number of viewers fluctuated at different stages of the landing. When the number shot up, a lot of computing, storage, and networking resources were required. But it was not so at other times. So NASA used a hybrid cloud system to handle the load fluctuation, as shown in the following picture. They used Amazon AWS for this purpose.
Use of cloud computing for power industry
I won’t repeat the definition of cloud computing here, but cloud computing can play a great role in the power industry. The basic computing can be outsourced to different kinds of clouds, depending upon security and availability needs. In addition, with the advent of smart grid, more data are generated and stored. Some data may not be useful now but may prove to be important later, when combined with other data that were collected from different sources. Analytics is still in its infancy, and to make the matter worse, we need to tailor analytics to each case that differs vastly from others.
