Cloud computing, which provides services including data storage and video streaming over the Internet, was already growing before the COVID-19 pandemic occurred. During the pandemic, the technology took off even more as social distancing became essential, lockdowns became frequent and remote work became the norm in multiple industries, including health care, education and commercial, in many countries. In a post-pandemic world, cloud use is likely to continue to grow. According to Gartner, more than half of enterprise IT spending in key market segments will shift to the cloud by 2025, overtaking spending on traditional IT.
However, the question remains as to how a business can maximize value from its transition to the cloud.
Cloud computing remains a fragmented industry. In other words, it lacks transparency in terms of its costs and values, customization (appropriate matching between customers and service) and scalability. The technology also has redundant offerings. For example, without a clear sense of the market in terms of cost and the types of services different companies offer, it will be time-consuming process for a business to try out a cloud computing service, possibly determine that the fit is not perfect and then try the services of another company. It is a bit like trying to find the best rate of car insurance by spending hours on the phone with different companies. In addition, the transfer of assets—in this case, data—remains challenging, impeding the collaboration between groups working with different clouds. Moreover, the growing concern over data safety and privacy, as well as the ensuing increase in regional government regulation, will present additional challenges for efficient computing and analytics across different cloud services, countries and continents.
Therefore, there is a need to reduce the cost of using cloud computing while increasing the compatibility and security for a broader application of the technology to more applications, including autonomous vehicles and remote health care.
The Sky Computing Lab, launched recently by UC Berkeley, is a research lab that aims to accelerate the transition to cloud computing. Its endeavor is certainly ambitious. Not content with just finding ways to incrementally increase the adoption of cloud computing, the lab aims to build a new infrastructure for interconnected cloud computing to address many problems facing the industry, such as high costs, low compatibility and low usability.
Sky Computing wants to shift the existing fragmented market of cloud computing to a utility market that will allow cloud computing to become as connected and easy to use as the Internet, eventually evolving into Sky Computing.
“If you look at the Internet, under the hood, it’s a collection of different networks and technologies. But the Internet ties them together. When you communicate, you do not know what different technologies and networks are under the hood, right? You just send packets from A to B,” said Ion Stoica, the lead scientist at Sky Computing Lab, “Sky plans to do similarly for the cloud.”
The researchers at Sky Computing Lab envision that sky computing having several components. First, mediator services, or “intercloud brokers,” will act as the linchpins in the more interconnected cloud computing. Each mediator will have a catalog that holds information about the cloud’s services, compatibility, pricing and performance. Meanwhile, a job intake interface will collect descriptions of the user’s needs so that an optimized user-specific plan can be created based on what clouds to use and when.
In addition, Sky Computing will improve compatibility because broker services can move or replicate data to the designated locations, allocate resources and execute and troubleshoot a computational plan across clouds. Moreover, these services will have the capability to verify the identity of users, ensuring data security.
In the future, Stoica expects cross-cloud agreements to enable the free exchange of data and the connection of many disparate clouds.
Eventually, Sky Computing will expand to reach a greater number and range of consumers interested in using cloud services, according to Stoica, making it easier to use the hardware that’s right for a user’s needs, regardless of which cloud originally holds that application. Users will be able to aggregate resources across clouds to run large workloads that they cannot access on individual cloud accounts. Furthermore, Sky Computing will help companies comply with data sovereignty rules in different countries and protect their data more easily.
The vision of Sky Computing is becoming a reality. The team at Sky Computing Lab is testing an early version of an intercloud broker prototype for machine learning workloads. The researchers have spelled out their vision in a white paper titled “The Sky Above the Clouds.” “We should be clear that what we are proposing is nothing short of a revolution in an important technology ecosystem,” said Stoica. “We don’t ask that readers accept Sky Computing as inevitable—merely as not impossible. And we invite interested readers to help us on this quest.”