Týr: Blob Storage Meets Built-In Transactions

Matri, Pierre; Costan, Alexandru; Antoniu, Gabriel; Montes, Jesús; Pérez, Marı́a S.

doi:10.1109/sc.2016.48

Cited by 8 publications

(9 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Týr is a blob storage system with support for transactions; it provides blob storage functionality and high access parallelism [64]. Measurements show that many applications do not require most of the functionality provided by full-fledged file systems but instead only use a subset that can be provided by blob or object storage systems [63].…”

Section: Storage and File Systemsmentioning

confidence: 99%

Survey of Storage Systems for High-Performance Computing

Lüttgau

Kuhn

Duwe

et al. 2018

JSFI

View full text Add to dashboard Cite

In current supercomputers, storage is typically provided by parallel distributed file systems for hot data and tape archives for cold data. These file systems are often compatible with local file systems due to their use of the POSIX interface and semantics, which eases development and debugging because applications can easily run both on workstations and supercomputers. There is a wide variety of file systems to choose from, each tuned for different use cases and implementing different optimizations. However, the overall application performance is often held back by I/O bottlenecks due to insufficient performance of file systems or I/O libraries for highly parallel workloads. Performance problems are dealt with using novel storage hardware technologies as well as alternative I/O semantics and interfaces. These approaches have to be integrated into the storage stack seamlessly to make them convenient to use. Upcoming storage systems abandon the traditional POSIX interface and semantics in favor of alternative concepts such as object and key-value storage; moreover, they heavily rely on technologies such as NVM and burst buffers to improve performance. Additional tiers of storage hardware will increase the importance of hierarchical storage management. Many of these changes will be disruptive and require application developers to rethink their approaches to data management and I/O. A thorough understanding of today's storage infrastructures, including their strengths and weaknesses, is crucially important for designing and implementing scalable storage systems suitable for demands of exascale computing.

show abstract

Section: Storage and File Systemsmentioning

confidence: 99%

Survey of Storage Systems for High-Performance Computing

Lüttgau

Kuhn

Duwe

et al. 2018

JSFI

View full text Add to dashboard Cite

show abstract

“…In this section we detail the implementation of the above principles in a prototype file system optimized for small files, that we name TýrFS. We implement it as a thin layer atop the Týr transactional, distributed object storage system [27]. The system is composed of a data service described in Section V-A, directly mapped to Týr native semantics.…”

Section: Týrfs: a File System Optimized For Small Filesmentioning

confidence: 99%

“…• We propose a novel file system architecture optimized for small files based on consistent hashing and dynamic metadata replication while not sacrificing performance for other workloads (Section IV). • We detail the design and implementation of TýrFS, a file system implementing the aforementioned principles (Section V), built atop the Týr object storage system [27]. • After introducing our experimental platform (Section VI), we leverage TýrFS in various applicative contexts and factually prove its applicability to both HPC and Big Data workloads (Section VII).…”

Section: Introductionmentioning

confidence: 99%

TýrFS: Increasing Small Files Access Performance with Dynamic Metadata Replication

Matri

Pérez²,

Costan³

et al. 2018

2018 18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)

Self Cite

View full text Add to dashboard Cite

Small files are known to pose major performance challenges for file systems. Yet, such workloads are increasingly common in a number of Big Data Analytics workflows or largescale HPC simulations. These challenges are mainly caused by the common architecture of most state-of-the-art file systems needing one or multiple metadata requests before being able to read from a file. Small input file size causes the overhead of this metadata management to gain relative importance as the size of each file decreases. In this paper we propose a set of techniques leveraging consistent hashing and dynamic metadata replication to significantly reduce this metadata overhead. We implement such techniques inside a new file system named TýrFS, built as a thin layer above the Týr object store. We prove that TýrFS increases small file access performance up to one order of magnitude compared to other state-of-the-art file systems, while only causing a minimal impact on file write throughput.

show abstract

“…Týr [32], which is an object store targeted at coping with highly concurrent I/O patterns. It features a lightweight transaction protocol that natively enables atomic operations such as appends to be performed at high speed.…”

Section: B Evaluated Persistence Backendsmentioning

confidence: 99%

SLoG: Large-Scale Logging Middleware for HPC and Big Data Convergence

Matri

Carns

Ross

et al. 2018

2018 IEEE 38th International Conference on Distributed Computing Systems (ICDCS)

Self Cite

View full text Add to dashboard Cite

Cloud developers traditionally rely on purposespecific services to provide the storage model they need for an application. In contrast, HPC developers have a much more limited choice, typically restricted to a centralized parallel file system for persistent storage. Unfortunately, these systems often offer very low performance when subject to highly-concurrent, conflicting I/O patterns. This makes difficult the implementation of inherently concurrent data structures such as distributed shared logs. Yet, this data structure is key to applications such as computational steering, data collection from physical sensor grids or discrete event generators. In this paper we tackle this issue. We present SLoG, a shared log middleware providing a shared log abstraction over a parallel file system, designed to circumvent the aforementioned limitations. We evaluate SLoG design on up to 100,000 cores of the Theta supercomputer: it demonstrates high append velocity at scale while also providing substantial benefits for other persistent backend storage systems.

show abstract

Týr: Blob Storage Meets Built-In Transactions

Cited by 8 publications

References 17 publications

Survey of Storage Systems for High-Performance Computing

Survey of Storage Systems for High-Performance Computing

TýrFS: Increasing Small Files Access Performance with Dynamic Metadata Replication

SLoG: Large-Scale Logging Middleware for HPC and Big Data Convergence

Contact Info

Product

Resources

About