🤖 AI Summary
This work addresses the challenge of data inconsistency or loss in transactional key-value operations across heterogeneous storage media—such as DRAM, SSD, HDD, and NVM—by proposing and implementing a decoupled, multi-tier transactional key-value store. The system uniquely enables unified read-write access within a single transaction to replicas of the same key-value pair distributed across distinct storage pools, supporting flexible configuration of replication or tiering strategies to jointly optimize performance and durability. Built as an extension of FoundationDB, it integrates diverse heterogeneous storage resources while guaranteeing strong transactional consistency. Evaluation through microbenchmarks and real-world eBay production workloads demonstrates that the open-source implementation achieves excellent performance and practical viability.
📝 Abstract
A stash is a storage medium such as Dynamic Random Access Memory (DRAM), Solid State Disk (SSD), Hard Disk Drive (HDD), or Non-Volatile Memory (NVM). This paper presents a disaggregated transactional key-value (KV) store, DiStash, that governs KVs cross pools of stash types. It enables an application to use a single transaction to read and write different copies of one or more key-value pair across the different pools of stashes. It simplifies the application logic by (a) preventing undesirable race conditions that may cause copies of data across different stash pools to reflect different values and/or (b) failures that may result in loss of key-value pairs. A configuration of DiStash may use a pool of stashes as either ephemeral or durable storage. The application dictates whether the content of its participating stashes are inclusive (replicated) or exclusive (tiered). We implement a DiStash by extending FoundationDB. We quantify the tradeoffs with its design decisions using microbenchmarks and eBay's production workload. We open source our implementation at https://github.com/ebay-USC/DiStash.