Optimizing Cloud Resources + Reducing Your Carbon Footprint with TimeBox

At Cisco Engineering, innovation isn’t just something we do; it’s a way of life.

With tens of thousands of developers churning out an equal number of cutting-edge solutions at high velocity, Cisco truly is at the helm of technological innovation.

For context, Cisco has a vast amount of DevOps activities that are associated with development and these require significant resources for running workloads. The resources encompass storage, compute, memory, and associated ancillary costs such as real estate footprint, electricity, and others. Moving to the cloud does not change the fundamentals of this challenge,  even cloud workloads at the end of the day need to run on compute (and consume electricity). This landscape created the perfect opportunity for Cisco internal engineering to innovate.

Enter TimeBox.

Born out of a Cisco-fueled engineering hackathon and with roots in our Kanata, Ontario, Research & Development Centre, TimeBox is an award-winning made-in-Canada solution. With two filed patents, it is taking cloud resource optimization at Cisco to new heights.

As a data-driven resource optimizer, TimeBox:

1. Understands intent.

2. Provides recommendations.

3. Monitors and heals workloads on auto-pilot.

4. Provides insight into workloads.

5. Is a one-stop-shop to discover your Total Cost of Ownership (TCO) footprint, directly mapping to financial costs.

Here is the recipe:

Through machine learning, TimeBox understands the intent of historic workload computations, then uses those to make recommendations for a better schedule. Once tweaked, this schedule gets re-trained for subsequent, more sophisticated Artificial Intelligence (AI) driven recommendations. It also works as a smart assistant, automatically answering frequently posed questions and challenges encountered by our Cisco engineers. These include:

1. Determining the optimal resources required for a given workload.

2. Autonomous monitoring and healing of aborted workloads.

3. Total Cost of Ownership for a given workload.

4. Preventing accidental hoarding of resources.

In a nutshell:

Scheduling and optimizing cloud resources is not a new idea, but using genetic modelling-based AI to solve for it may just be. TimeBox can be pervasive, with applications across any industry where the efficiency of resource allocation is critical. Where there are resources that undergo periodic consumption, there is a need for optimal capacity planning, workloads with large variety, and associated variable characteristics.