Hardware-Assisted Bare Metal Migration via Dirty Page Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Bare metal servers (BMS) lack the capability for online migration, necessitating hardware fault avoidance and load balancing without service interruption, which is crucial for high-performance computing environments.
Innovation Solution
An online migration method and system for BMS, utilizing a hardware card to manage virtual machine managers, record dirty memory page locations, and migrate memory pages between servers in real time, reducing server burden and ensuring seamless transitions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If online migration is implemented for BMS, then service continuity and load balancing are improved, but device complexity increases due to hardware card and virtual machine manager involvement
Solution Approach 1:
The hardware card serves as an intermediary component between the source and destination BMS, managing the migration process by coordinating with the virtual machine manager to track dirty memory pages and facilitate data transfer, thereby enabling service continuity without requiring direct complex interactions between the servers themselves
Solution Approach 2:
The system creates a copy of the virtual machine manager on the hardware card to handle migration tasks independently from the main server operations, allowing the migration process to proceed without interrupting the running services on the source or destination BMS
2Reliability
If hardware card manages migration process, then migration reliability is improved, but server resource consumption increases during migration
Solution Approach 1:
The migration management functions are extracted from the server's main processing tasks and relocated to the hardware card, which handles dirty memory page tracking and coordination independently, thereby reducing the computational burden on the server during migration operations
Solution Approach 2:
The hardware card performs preliminary actions by pre-tracking dirty memory pages and preparing migration data structures before actual migration occurs, allowing the migration process to proceed more efficiently with reduced real-time resource consumption
3Measurement precision
If virtual machine manager tracks all memory pages, then migration precision is improved, but processing time increases
Solution Approach 1:
Instead of tracking all memory pages, the system implements partial tracking by focusing only on dirty memory pages (pages that have been modified), which are the only pages that need to be migrated, thereby reducing processing time while maintaining sufficient precision for successful migration
Solution Approach 2:
The hardware card receives feedback about dirty memory page locations from the virtual machine manager and uses this information to optimize the migration process, selectively transferring only the necessary pages rather than scanning the entire memory space
Data Source
AI summary
An online migration method and system for a bare metal server, where a first hardware card receives a migration command for a first bare metal server, and where the first hardware card is inserted into the first bare metal server. The first hardware card notifies, based on the migration command, the first bare metal server to start a virtual machine manager in the first bare metal server, where the virtual machine manager records first dirty memory page location information that is for memory of the first bare metal server and that is generated by the first bare metal server, and sends the first dirty memory page location information to the first hardware card. The first hardware card online migrates a dirty memory page of the first bare metal server to a second bare metal server based on the first dirty memory page location information.