Cache Migration Selecting Long-Term Utility Entries

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The migration of virtual machines (VMs) between hosts results in performance loss and prolonged migration times due to the handling of large storage caches, as discarding the cache leads to repopulation delays and migrating the entire cache consumes excessive network bandwidth.

Innovation Solution

Selecting only cache entries with long-term utility indicators for migration, along with their metadata, while dynamically adjusting the cache size to optimize the operating cache size and reduce the amount of data to be transferred.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the whole cache is migrated with the virtual machine, then the VM performance is maintained after migration, but the migration time is prolonged dramatically and network bandwidth is consumed for an extended period

Engineering Contradiction:
ImproveVM performance after migrationVSAvoidmigration time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The cache is divided into two segments: frequently accessed cache entries that are migrated with the VM to maintain performance, and less frequently accessed entries that remain on the original storage device. This segmentation allows partial migration that balances performance requirements with reduced migration time and network bandwidth consumption.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Only the essential portion of the cache (frequently accessed entries) is extracted and migrated with the VM, while the remaining cache entries are left behind. This extraction approach maintains the critical performance benefits of cache migration while avoiding the drawbacks of migrating the entire cache.

Inventive Principle:
Principle #2Taking out (Extraction)

2Loss of time

If the cache is discarded and only the virtual machine is migrated, then the migration time is reduced, but the VM suffers a loss of performance following migration and the cache takes a long time to repopulate

Engineering Contradiction:
Improvemigration timeVSAvoidVM performance after migration
Core Design Contradiction:
Loss of timeVSReliability

Solution Approach 1:

Instead of migrating the entire cache (excessive action) or no cache at all (insufficient action), the invention migrates only the necessary portion of the cache - specifically the frequently accessed entries. This partial action is sufficient to maintain VM performance while avoiding the drawbacks of full cache migration.

Inventive Principle:
Principle #16Partial or excessive action

3Reliability

If the whole cache is migrated, then performance is maintained, but network bandwidth is consistently consumed for an extended period

Engineering Contradiction:
ImproveI/O throughputVSAvoidnetwork bandwidth consumption
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The cache migration is segmented to transfer only frequently accessed entries over the network, significantly reducing network bandwidth consumption. The less frequently accessed entries remain on the original storage device, eliminating the need to transfer them across the network while maintaining I/O throughput for the migrated portion.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS9285999B2Cache migration
Publication Date: 2016.03.15 VMWARE INC
  • US9285999B2 patent drawing
  • US9285999B2 patent drawing
  • US9285999B2 patent drawing

AI summary

Exemplary methods, apparatuses, and systems determine that a cache is to be migrated from a first storage device to a second storage device. Each cache entry within the cache includes a first indicator to indicate whether or not the cache entry has long-term utility. Only a portion of all cache entries are selected to be migrated and the portion is selected from cache entries with the first indicator set to indicate long-term utility. The selected cache entries and metadata for cache entries that were not selected are migrated from the first storage device to the second storage device.