Integrating parallel file systems with object-based storage devices

Devulapalli, A.; Dalessandro, D.; Wyckoff, P.; Ali, Nawab; Sadayappan, P.

doi:10.1145/1362622.1362659

Cited by 28 publications

(18 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The T10 standards [8], [19] were proposed to store data and attributes as objects. Based on T10 standards, Seagate, IBM (ObjectStore [7]), and Panasas (PanFS [25]) implemented prototypes [19] and demonstrated the capability of object-based storage systems. RADOS [24]], as part of Ceph [23], is a scalable and reliable object storage service for petabyte-scale storage clusters.…”

Section: Related Workmentioning

confidence: 99%

Toward Transparent Data Management in Multi-Layer Storage Hierarchy of HPC Systems

Wadhwa

Byna

Butt

2018

2018 IEEE International Conference on Cloud Engineering (IC2E)

View full text Add to dashboard Cite

Abstract-Upcoming exascale high performance computing (HPC) systems are expected to comprise multi-tier storage hierarchy, and thus will necessitate innovative storage and I/O mechanisms. Traditional disk and block-based interfaces and file systems face severe challenges in utilizing capabilities of storage hierarchies due to the lack of hierarchy support and semantic interfaces. Object-based and semantically-rich data abstractions for scientific data management on large scale systems offer a sustainable solution to these challenges. Such data abstractions can also simplify users involvement in data movement. In this paper, we take the first steps of realizing such an object abstraction and explore storage mechanisms for these objects to enhance I/O performance, especially for scientific applications. We explore how an object-based interface can facilitate next generation scalable computing systems by presenting the mapping of data I/O from two real world HPC scientific use cases: a plasma physics simulation code (VPIC) and a cosmology simulation code (HACC). Our storage model stores data objects in different physical organizations to support data movement across layers of memory/storage hierarchy. Our implementation sclaes well to 16K parallel processes, and compared to the state of the art, such as MPI-IO and HDF5, our object-based data abstractions and data placement strategy in multi-level storage hierarchy achieves up to 7× I/O performance improvement for scientific data.

show abstract

Section: Related Workmentioning

confidence: 99%

Toward Transparent Data Management in Multi-Layer Storage Hierarchy of HPC Systems

Wadhwa

Byna

Butt

2018

2018 IEEE International Conference on Cloud Engineering (IC2E)

View full text Add to dashboard Cite

show abstract

“…The SCSI framework that we use to prototype our SSD, tgt [10], has been used in several research projects [8,7]. SCST [6] is a similar framework, but it runs in the kernel rather than in user mode.…”

Section: Related Workmentioning

confidence: 99%

Prototyping a high-performance low-cost solid-state disk

Budilovsky

Toledo

Zuck

2011

Proceedings of the 4th Annual International Conference on Systems and Storage

View full text Add to dashboard Cite

We present a design for a high-performance low-cost solid-state disk (SSD). Ignoring garbage-collection costs, our SSD performs only 1 + ε physical accesses to NAND flash pages for every request of a page-size block by the host, for some small ε. This is true for all access patterns, including random writes, which are usually slow on low-cost SSDs. Garbage collection in all SSDs is determined primarily by how full the SSD is, and its cost is similar in most SSDs. The unique feature in our design is that it achieves high performance even with when the SSD contains only a small amount of RAM. In most SSD designs, this would imply low performance; in ours, it does not. A small RAM lowers the cost of an SSD with a given flash array. Our design achieves high performance with a small RAM using two innovative ideas. One is the use of a clever mapping data structure. The second is a host-assisted hinting mechanism that uses RAM on the host to compensate for the small amount of RAM within the SSD. This mechanism is implemented as an enhanced SCSI driver (kernel module). Our prototyping methodology is also a significant contribution. We simulate the SSD in software, using files to represent the flash array, but the resulting prototype is a working SCSI device that file systems can be mounted on.

show abstract

“…The Ceph and PVFS object storage abstractions were described in detail in Section IV. Devulapalli and Ali have also investigated integration of T10 compliant object storage in the PVFS file system [26].…”

Section: Enterprise Storagementioning

confidence: 99%

Object storage semantics for replicated concurrent-writer file systems

Carns

Ross

Lang

2010

2010 IEEE International Conference on Cluster Computing Workshops and Posters (CLUSTER WORKSHOPS)

View full text Add to dashboard Cite

Abstract-High-performance I/O is a key requirement for many of today's critical computational science applications, and parallel file systems are being driven to progressively larger scales to keep pace with demand. One cost-effective way to meet this demand is through the deployment of commodity storage hardware in conjunction with file systems that provide software resiliency. This requires a re-evaluation of the core components of parallel file system architecture, however. In addition to interacting with resilient protocols, parallel file systems must also take into account unique HPC workloads that include bursty, highly concurrent access to large shared files. Such workloads are traditionally a challenge for software replication algorithms, in part because the underlying storage does not provide convenient semantic building blocks. In this work we isolate a common component of many parallel file systems, the object storage abstraction layer, and propose the introduction of semantic properties that will enable it to better serve as the building block for resilient HPC storage architectures. The properties that we have identified are atomicity, explicit versioning, and commutativity. We outline how these properties can be used to simplify software replication protocols for highly concurrent workloads. We also demonstrate that these properties can be implemented portably while still maintaining high performance on both commodity and enterprise-class storage platforms.

show abstract

Integrating parallel file systems with object-based storage devices

Cited by 28 publications

References 14 publications

Toward Transparent Data Management in Multi-Layer Storage Hierarchy of HPC Systems

Toward Transparent Data Management in Multi-Layer Storage Hierarchy of HPC Systems

Prototyping a high-performance low-cost solid-state disk

Object storage semantics for replicated concurrent-writer file systems

Contact Info

Product

Resources

About