Federated Data Tiering for Edge Computing Cache Misses
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current edge computing systems face inefficiencies in data management due to the need for each edge server to independently decide which data to keep and flush, leading to costly and time-consuming cache misses, diverting resources from improving user experience.
Innovation Solution
Implementing a federated machine learning approach that uses a global model to predict and proactively tier data, storing the most sought-after data at edge servers, while less sought-after data is retained in central systems, using a decentralized data tiering system that anticipates user requests and reduces cache misses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If each edge server independently manages its own cache data, then local data access autonomy is improved, but computing resources are wasted on cache maintenance and cache misses occur more frequently
Solution Approach 1:
The patent introduces a hub system as an intermediary between edge servers and central data centers. The hub system coordinates data tiering decisions across multiple edge servers, reducing redundant cache maintenance while maintaining local access autonomy. The hub receives data access patterns from edge servers and orchestrates data placement decisions, eliminating the need for each edge server to independently manage its cache.
Solution Approach 2:
The hub system serves multiple functions: it acts as a data tiering coordinator, a caching layer for frequently accessed data, and a learning system that aggregates access patterns from multiple edge servers. This multi-functional approach improves overall system efficiency while maintaining the autonomy of individual edge servers.
2Speed
If all data is stored at edge servers to improve access speed, then data retrieval time is reduced, but storage capacity and computing power at edge servers become insufficient
Solution Approach 1:
The patent segments the storage system into three hierarchical tiers: edge servers for hot data (frequently accessed), hub systems for warm data (moderately accessed), and central data centers for cold data (rarely accessed). This segmentation allows data to be stored at the appropriate level based on access frequency, optimizing both retrieval speed and storage utilization.
Solution Approach 2:
The system proactively pre-loads data from the hub or central data center to edge servers based on predicted access patterns. By anticipating future data requests and pre-positioning data at the edge before it is actually needed, the system reduces latency while maintaining efficient use of edge storage capacity.
3Speed
If frequently accessed data is kept at edge servers, then user access speed is improved, but cache misses still occur requiring expensive regeneration processes
Solution Approach 1:
The system implements feedback loops where edge servers report data access patterns to the hub, which aggregates this information and adjusts data tiering decisions accordingly. This continuous feedback mechanism allows the system to adapt to changing access patterns, keeping frequently accessed data at the edge and reducing cache misses.
Solution Approach 2:
The hub system uses aggregated access pattern data to proactively identify data that will likely be needed soon and pre-loads it to edge servers before actual requests occur. This preliminary action prevents cache misses by ensuring data is already available at the edge when users request it.
Data Source
AI summary
Systems and methods for tiering data in distributed data networks. A global model is developed based on federated learning where edge servers are able to train a model. The learning from the edge servers are collectively applied to the global model. This process can be repeated until the global model is ready for deployment. The global model allows data to be tiered. This may include pushing data from a datacenter to edge servers or cleaning edge servers of data that does not satisfy the global model. The model can be retrained and can be used to proactively push new content out to the edge servers.