Efficient remote memory for parallel and distributed data analytics
Abstract/Contents
- Abstract
- This thesis describes Clamor, a functional cluster computing framework that adds support for fine-grained, transparent access to global variables for distributed, data-parallel tasks. Clamor targets workloads that perform sparse accesses and updates within the bulk synchronous parallel (BSP) execution model, a setting where the standard technique of broadcasting global variables is highly inefficient. We show that this restriction of workloads is powerful, enabling efficient transparent remote memory access using techniques from distributed shared memory that are known to be inefficient for general workloads. These restrictions further enable novel features in Clamor, including a dynamic distributed serving mechanism that takes advantage of the functional programming model to cache and serve data for the duration of a parallel task, and lineage-based fault recovery at a finer granularity than existing systems that do not account for sparsity in the access pattern. Clamor can integrate with existing Rust and C ++ libraries to transparently distribute programs on the cluster. We show that Clamor is competitive with Spark in simple functional workloads and can improve performance significantly compared to custom systems on workloads that sparsely access large global variables: from 5× for sparse logistic regression to over 100× on distributed geospatial queries.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2022; ©2022 |
Publication date | 2022; 2022 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Thaker, Pratiksha Ranjit |
---|---|
Degree supervisor | Zaharia, Matei |
Thesis advisor | Zaharia, Matei |
Thesis advisor | Levis, Philip |
Thesis advisor | Ousterhout, John K |
Degree committee member | Levis, Philip |
Degree committee member | Ousterhout, John K |
Associated with | Stanford University, Computer Science Department |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Pratiksha Thaker. |
---|---|
Note | Submitted to the Computer Science Department. |
Thesis | Thesis Ph.D. Stanford University 2022. |
Location | https://purl.stanford.edu/tt346kh1748 |
Access conditions
- Copyright
- © 2022 by Pratiksha Ranjit Thaker
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...