How Cloud Systems Automatically Assign Virtual Machines to Servers

A method for cloud services to automatically choose the best server for a task by filtering out impossible options and then ranking the remaining ones.

What it covers

This patent describes an automated system for managing cloud infrastructure, such as assigning virtual machines or databases to physical servers. It uses a two-step filtering process: first, it applies 'hard constraints' to disqualify any server that cannot physically or logically handle the task (like a server that is already full or lacks the required hardware). Second, it applies 'soft constraints' to rank the remaining servers based on performance preferences, such as spreading out workloads to avoid bottlenecks. This ensures that redundant services are placed on different physical machines, preventing a single hardware failure from taking down an entire application.

What it doesn't cover

• —Does not cover manual placement of objects by a human administrator.
• —Does not cover systems that rely solely on a single-step selection process without distinguishing between hard and soft constraints.
• —Does not cover the internal logic of how a virtual machine operates once it is placed on a host.
• —Does not cover hardware-level circuit switching or physical network cabling.

The clever bit

The innovation lies in the strict separation of 'hard' (binary pass/fail) and 'soft' (optimization-based) constraints, which allows the system to quickly prune the search space before performing more complex calculations on the remaining candidates.

Why it matters

As cloud computing grew, manually deciding where to put thousands of virtual machines became impossible. This patent provides a structured, scalable way to manage large-scale data centers, which is essential for the reliability of modern cloud platforms like Microsoft Azure. It helps ensure that services remain online even when individual physical servers fail.

Real-world examples

• 1.Microsoft Azure resource scheduling
• 2.Automated virtual machine load balancing
• 3.Data center capacity management software