In previous years, Werner Vogels CTO at AWS said somewhat jokingly that it would “fix the speed of light” as the new world of the cloud removed the limitations in computing. During his keynote address at the 2021 re: Invent Conference, he said that this task remained on his to-do list, but that there were immediate prospects at hand.
Vogels revisited some of his comments about innovation from Re: Previous Invention Conferences to frame how past limitations have hampered the ability to evolve and grow. “Before the cloud, you either had to make a huge investment, you had to buy hardware, you had to hire IT people — things that had nothing to do with actually building a product.”
What Cloud did, he said, was to take whatever was tied up—all the pieces of hardware—and make it all programmable. “Suddenly, accessing capacity was just the click of a button.”
Data centers and other physical resources, for example, have become virtually programmable. Vogels said the early days of Amazon Elastic Compute Cloud (EC2) saw a simple interface, and since that time EC2 has been ambitious in developing instances — virtual servers in the cloud — to meet user demand.
“Almost every day you would ask us for a different family,” he said. “You want optimal storage. You want optimal computation. You want to have large memory copies where you can run your SAP HANA.”
He said the first cases were based on a limited hypervisor, noting that massive investment and innovation in data centers are needed to ensure AWS can bring next-generation computing platforms into the hands of customers. At this year’s conference, Vogels announced the new Mac EC2 M1 instances, which should offer even greater efficiency. “Apple claims for M1 Mac instances there is a 60% cost improvement over the platforms they have used before.”
Although the evolution of the cloud has removed some of the limitations on innovation and development, he said there is more that can be done from a digital and political perspective. “There are a number of laws that are constantly being constraints that we have to deal with,” Fogels said. “Whether there is a delay in bandwidth…or the law of the land.”
The latter is easier to address by working with regulators to educate them about the cloud, its capabilities, and the potentially higher level of security for data in the cloud than in private data centers, he said.
The possibilities for scaling in the cloud can reach far beyond terrestrial innovation. Payam Banazadeh, CEO and founder of Capella Space, a developer of surveillance satellites, joined Vogels to talk about the opportunities that the cloud offers for working with data transmitted from space.
“Sensors from space have a really unique advantage,” Banzadeh said. “We need it in order to have a connected world.” He said that organizations need continuous global access to information in order to make decisions on a global scale. “For this, space is the missing link,” Banzadeh said. He said the satellites from Capella Space can capture images and other changes on the planet’s surface down to millimeter resolution. “It gives us reliability and visibility of our planet in all conditions. This is a prerequisite for access to real-time monitoring.”
Banazadeh said he expects the Capella satellites to accumulate more than 500 petabytes of data in the coming years that must be transmitted to terrestrial computers. “In order to lay the foundation for integrating our sensors with other sensors, in real time with low latency and low interaction, we had to build our business on a very different foundation than what has been traditionally experienced before,” he said. Capella needed a distributed network, complete automation, real-time data processing, and instant scalability. The satellite developer chose AWS in 2020 to serve the ground station.
Vogels gave a summary of AWS’ extensive geographic cloud expansion and how this allowed companies to choose which regions to use based on latency and cost to reach their intended customers. The increasing cloud footprint has helped Amazon expand its services.
“There are applications we wouldn’t have built if these areas weren’t closer to you,” Fogels said. “In 2006, 2009, or even 2013, Alexa won’t be possible.” Slow response time is critical to Amazon’s voice-controlled virtual assistant, which makes regional cloud availability even more important. “If you don’t get a response from Alexa within a second, it won’t feel like a normal conversation,” he said.
Advances in hardware, GPUs, and Alexa-enabled machine learning capabilities, Vogels said, but that latency still needs to be reduced. He said AWS currently maintains 25 territories across seven continents. “There are nine more areas that we plan to start working on in the next two years.”
Nasdaq CEO at AWS re: Invent Talks Cloud’s Impact on FinTech
Cloud computing is like a day in a chocolate factory for IT managers
Capella Space is working with AWS to handle satellite downlinks