Cloud evolution: The era of intent-aware clouds

As the cloud technology has evolved from a centralized to a distributed cloud approach, it has also adapted to industry transformation e.g. enabling innovative Mobile Edge Cloud and Industrial Cloud solutions. While such technologies are key to many of tomorrow's telecommunication, industrial and consumer applications, the cloud evolution must also be accompanied with a deeper understanding of cloud administrators' and users' business perspectives. To realize new opportunities, it's critical that any increase in scale and complexity is balanced by a focus on making the cloud simpler, more flexible and aware of high-level business and user intent.

Evolving the cloud to focus on high-level intents

By designing cloud management systems which learn, execute and adapt based on intent-aware behavior, cloud systems can intuitively align with cloud operators' and users' high-level objectives, such as requesting an optimized service deployment or requiring maximum service availability.

In the context of an industrial cloud, a business intent could relate directly to an expected outcome, specifying which operational goal should be met, instead of how it should be achieved. Imagine a fully-autonomous logistics chain which features tracking containers, pallets and individual items working together with industrial robots in real-time – all interacting with manufacturing equipment assisted by context-sensitive information. By evolving the industrial cloud towards a design based on intent-aware behavior, the cloud would be able to act on industrial business objectives – irrespective of scale or complexity.

A 5G connected intent-aware cloud for telecommunication, industrial and commercial enterprises

The intent-aware cloud

Building a cloud system that interprets high-level intents represents a new set of research challenges. While cloud management systems typically focus on providing advanced orchestration capabilities to simplify cloud-related operations, an intent-aware cloud is meant to bring a more holistic approach to cloud management, largely inspired to also become a potential business contributor. By promoting a better understanding of cloud administrators' and users' perspectives, several new key system attributes must be considered:

• Business objectives
  This represents a new aspect of cloud management, enabling capabilities to interpret business language to specify cloud-related operations inspired by high-level intents. As business objectives are specified in terms of measurable goals, business-related key performance indicators (KPIs) are identified, and an intent-monitoring service is dynamically built to assess the specified objectives in real-time. Intents can be specified by both cloud administrators and users by providing concrete operative goals, from improving cloud infrastructure resources efficiency by a certain percentage, to optimizing a service performance while minimizing operational cost.
• Self-organizing services
  Self-organizing services are assumed to interact with each other directly, reacting quickly to changes of an execution environment, while allowing for efficient cognitive cooperation for advanced autonomous behavior. This is considered a key aspect of an intent-aware cloud, as high-level intents would typically involve the deployment of dynamically composed services for intent monitoring purposes. And with focus on cloud simplification, resource optimization and performance, it is imperative to provide self-configuration, self-management, self-optimization, self-healing and self-protection capabilities.
• Intelligent awareness
  This reflects an understanding of the applications and services deployed on a cloud system, as well as their interactions and interworking characteristics. It also implies a proper perception of cloud administrators' and users' perspective regarding their intents for cloud infrastructure resources, services and applications. And in combination with real-time cloud data analytics, Machine Learning (ML) and Artificial Intelligence (AI) algorithms, a cloud management system has the means to enforce high-level business intents.

Self-organizing distributed services

With increasing interest for real-time and high-volume data services, e.g. driven by industrial automation, automotive, real-time logistics, entertainment and AR/VR, services are expected to always be provided in the most efficient way and at the most appropriate location across a network of distributed sites. The intent-aware cloud will allow businesses to optimize service management, provisioning, performance and overall cloud system efficiency, while minimizing latency and system resources.

Towards the vision

Leveraging on several years of contributions to different research projects, each one enlightening us on different innovative solutions and technologies, we have gradually started to materialize the vision of an intent-aware cloud.

One example is our contributions to the EU-FP7 UNIFY and the 5G Exchange projects, where we worked on advanced cloud orchestration solutions for end-to-end service composition and delivery automation, applied to the concept of anything-as-a-service (XaaS), as well as experimented with the notion of intent-based specifications and interfaces.

As another significant research project paving the path towards the intent-aware cloud, there is the Calvin platform, which we developed to completely abstract an underlying cloud infrastructure, letting software developers focus on their specific application's requirements and capabilities through declarative statements describing their intent. The platform thereby automatically and dynamically partitions and orchestrates the software based on those rules. Find out more in our conference paper, Calvin – merging Cloud and IoT, and our presentation on Serverless IoT Deployment.

The next step in cloud evolution

In this era of industry transformation, cloud infrastructure are quickly becoming larger and more complex, evolving towards highly distributed and real-time interworking architectures. To sustain this growth opportunity, there's a need to simplify cloud-related operations with innovative technologies. For cloud providers, this could mean managing highly scalable and distributed cloud systems according to their high-level cloud infrastructure business objectives. While for service providers, this could mean simply focusing on their high-level service delivery aspirations.

Leveraging a high-degree of autonomous behavior, self-organizing distributed services, and an advanced intent-aware framework, the intent-aware cloud is designed to actively contribute to high-level operational and business objectives, as inspired by intent. By evolving cloud technology towards an intent-aware behavior, we empower cloud systems to continue evolving as needed, while keeping cloud operations as simple as possible.

Find out more

Read our evolved cloud collaboration blog post to explore how distributed architecture can improve efficiency and quality of video conferencing platforms.

Read more about Ericsson edge computing