Context:
The ever-growing number of services and Internet of Things (IoT) devices has resulted in data being distributed across different locations (regions and countries) and different storage tires. Additionally, data exhibits different usage patterns, including cold data (written once and never read), stream data (produced once and consumed by many), and hot data (written once and consumed by many). Furthermore, these data types have different performance and dependability requirements (e.g., low latency for data streams).
To ensure the reliability and improve the performance of data-intensive applications, data are either replicated or erasure-coded and distributed across different storage tiers, while frequently accessed data are stored on high-speed devices close to end users (i.e., cached). While much work has investigated data caching, data placement strategies (i.e., deciding what to cache), data movement, cache partitioning, cache eviction [1–8], and cost-efficient data redundancy techniques in caching systems [9], few efforts have focused holistic caching and data management when caches are distributed across heterogeneous platforms (from Edge to Cloud), utilize storage devices with varying performance and cost characteristics, and simultaneously serve diverse workloads, including traditional data services, serverless workflows, and data streaming.
The goal of this engineer position is to study, implement, and evaluate novel cross-tier and distributed caching strategies, alongside supporting data management techniques, for hierarchical multi-tier storage systems. The engineer will work closely with a PhD student on this topic.
References:
[1] Asit Dan and Don Towsley. 1990. An Approximate Analysis of the LRU and FIFO Buffer Replacement Schemes. SIGMETRICS Perform. Eval. Rev. 18, 1 (apr 1990), 143–152. https://doi.org/10.1145/98460.98525
[2] Marek Chrobak and John Noga. 1999. LRU is better than FIFO. Algorithmica 23 (02 1999), 180–185. https://doi.org/10.1007/PL00009255
[3] Blankstein, Aaron, Siddhartha Sen, and Michael J. Freedman. “Hyperbolic caching: Flexible caching for web applications.” 2017 USENIX Annual Technical Conference (USENIX ATC 17). 2017.
[4] Cristian Ungureanu, Biplob Debnath, Stephen Rago, and Akshat Aranya. 2013. TBF: A memory-efficient replacement policy for flash- based caches. In 2013 IEEE 29th International Conference on Data Engineering (ICDE). 1117–1128. https://doi.org/10.1109/ICDE.2013.6544902
[5] Orcun Yildiz, Amelie Chi Zhou, Shadi Ibrahim. 2018. Improving the Effectiveness of Burst Buffers for Big Data Processing in HPC Systems with Eley. Future Generation Computer Systems, Volume 86, 2018, Pages 308-318, ISSN 0167-739X, https://doi.org/10.1016/j.future.2018.03.029.
[6] G. Aupy, O. Beaumont and L. Eyraud-Dubois, "Sizing and Partitioning Strategies for Burst-Buffers to Reduce IO Contention," 2019 IEEE International Parallel and Distributed Processing Symposium (IPDPS), Rio de Janeiro, Brazil, 2019,
[7] ZHANG, Yazhuo, YANG, Juncheng, YUE, Yao, et al. {SIEVE} is simpler than {LRU}: an efficient {Turn-Key} eviction algorithm for web caches. In : 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24). 2024. p. 1229-1246.
[8] Juncheng Yang, Ziming Mao, Yao Yue, and K. V. Rashmi. GL-Cache: Group-level learning for efficient and high-performance caching. FAST’23, pages 115–134, 2023.
[9] RASHMI, K. V., CHOWDHURY, Mosharaf, KOSAIAN, Jack, et al.{EC-Cache}:{Load-Balanced},{Low-Latency} cluster caching with online erasure coding. In : 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16). 2016. p. 401-417.
