BIOS Memory Range Reservation for Server Reset
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current server management in datacenters faces significant challenges with prolonged initialization times during physical resets, which can lead to extended downtime and violate the 99.999% uptime requirement due to the need for full system resets, especially during firmware updates or fabric resets, resulting in costly and frequent outages.
Innovation Solution
The implementation of a streamlined physical reset process that involves the BIOS reserving a memory range during initialization, allowing the host operating system to bypass reloading the kernel or hypervisor, and using a resume vector to transfer execution control, thereby reducing the time required for a physical reset and minimizing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a full system reset is performed, then the server can be restarted, but the initialization time increases to over three minutes
Solution Approach 1:
The reset process is segmented into two distinct paths: a fast path for firmware soft resets that preserves kernel state, and a slow path for full physical resets that reinitializes everything. This segmentation allows the system to choose the appropriate reset depth based on the specific need, avoiding unnecessary full resets and reducing initialization time when only firmware updates are required.
Solution Approach 2:
The system performs preliminary actions by preserving the kernel and hypervisor states before a physical reset occurs. By maintaining these critical components in memory and marking them for preservation, the system avoids redundant reloading during the reset process, significantly reducing the initialization time while still achieving the necessary firmware updates.
2Reliability
If a kernel soft reboot is used, then session disconnection is minimized, but it cannot handle firmware updates requiring fabric resets
Solution Approach 1:
The reset mechanism is made dynamic by allowing the system to adapt its behavior based on the reset type. A resume vector is introduced that dynamically directs the boot process to either the preserved kernel (for soft resets) or reinitialization (for physical resets with firmware updates). This dynamic approach enables the system to maintain session continuity when possible while adapting to handle firmware update requirements when necessary.
Solution Approach 2:
The resume vector acts as an intermediary mechanism that mediates between the reset trigger and the kernel initialization process. It provides a controlled transition path that allows the system to preserve kernel state for soft resets while still enabling full reinitialization when firmware updates require it, thus bridging the gap between session continuity and firmware update capability.
3Reliability
If full system reset is performed for firmware updates, then fabric reset is achieved, but client sessions are disconnected and reconnect time is added
Solution Approach 1:
The reset process is segmented into preservation of critical states (kernel, hypervisor, device tree) versus reinitialization of peripheral components. By preserving the fabric controller state and other critical components, the system can perform firmware updates without requiring full system reinitialization, thus avoiding session disconnects and reconnect delays while still achieving the necessary fabric reset for firmware updates.
4Loss of time
If BIOS initializes within a fixed memory range, then physical reset time is reduced, but memory location constraints are imposed
Solution Approach 1:
The BIOS performs preliminary action by pre-defining and reserving specific memory ranges for its initialization operations. By establishing these memory constraints in advance, the BIOS can efficiently initialize within the reserved ranges during physical resets without needing to search or allocate memory dynamically, significantly reducing reset time while the memory location management complexity is contained within the BIOS firmware itself.
Data Source
AI summary
Apparatuses, methods and storage medium associated with streamlined physical reset are described herein. In embodiments, an apparatus for computing, including streamlined physical reset, may comprise one or more processor cores; memory having a plurality of memory locations; and a basic input/output system (BIOS) to provide basic input/output system services, wherein the BIOS stays within a range of memory locations during each initialization of the BIOS, including an initialization of the BIOS that is part of a physical reset of the apparatus, to streamline the physical reset. Other embodiments may be described and/or claimed.