Storage-Optimized Data-Atomic Algorithms for Handling Erasures and Errors in Distributed Storage Systems

@article{Konwar2016StorageOptimizedDA,
  title={Storage-Optimized Data-Atomic Algorithms for Handling Erasures and Errors in Distributed Storage Systems},
  author={Kishori M. Konwar and N. Prakash and Erez Kantor and Nancy A. Lynch and Muriel M{\'e}dard and Alexander A. Schwarzmann},
  journal={2016 IEEE International Parallel and Distributed Processing Symposium (IPDPS)},
  year={2016},
  pages={720-729}
}
Erasure codes are increasingly being studied in the context of implementing atomic memory objects in large scale asynchronous distributed storage systems. When compared with the traditional replication based schemes, erasure codes have the potential of significantly lowering storage and communication costs while simultaneously guaranteeing the desired resiliency levels. In this work, we propose the Storage-Optimized Data-Atomic (SODA) algorithm for implementing atomic memory objects in the… 

Figures and Tables from this paper

ARES: Adaptive, Reconfigurable, Erasure Coded, Atomic Storage
TLDR
This work presents the first erasure-code based atomic algorithm, called ARES, which allows the set of hosts to be modified in the course of an execution, and provides bounds on the latency of read/write operations and analyze the storage and communication costs of the ARES algorithm.
A Layered Architecture for Erasure-Coded Consistent Distributed Storage
TLDR
This work proves liveness and atomicity of LDS, and also compute performance costs associated with read and write operations, and uses a class of erasure codes known as regenerating codes for storage of data in the back-end layer.
Storage-Efficient Shared Memory Emulation
TLDR
The design of storage-efficient algorithms for emulating atomic shared memory over an asynchronous, distributed message-passing network is studied and an extension of the single-writer algorithm to a multi-writer multi-reader environment is proposed.
Fast Lean Erasure-Coded Atomic Memory Object
14 In this work, we propose FLECKS, an algorithm which implements atomic memory objects in a multi-writer 15 multi-reader (MWMR) setting in asynchronous networks and server failures. FLECKS
RADON: Repairable Atomic Data Object in Networks
TLDR
This paper provides erasure-code-based algorithms having the additional ability to perform background repair of crashed nodes and shows how a slightly stronger network stability condition $N2$ can be used to construct algorithms that never violate atomicity.
CassandrEAS: Highly Available and Storage-Efficient Distributed Key-Value Store with Erasure Coding
TLDR
An erasure coding-based protocol that implements a key-value store with atomicity and near-optimal storage cost is proposed, namely Cassan-drEAS (Cassandra + Erasure-coding Atomic Storage).
Multi-Version Coding—An Information-Theoretic Perspective of Consistent Distributed Storage
TLDR
An achievable multi-version code construction via linear coding and a converse result that shows that the construction is asymptotically tight are provided.
Self-Stabilizing and Private Distributed Shared Atomic Memory in Seldomly Fair Message Passing Networks
TLDR
To the best of the knowledge, this work is the first to address privacy and self-stabilization in the context of emulating atomic shared memory in networked systems and uses a bounded storage during asynchronous executions.
Fundamental Limits of Erasure-Coded Key-Value Stores With Side Information
TLDR
This paper studies a system where a node acquires side information of the versions propagated to some other nodes based on the network topology and shows that the side information can result in a better storage cost as compared with the case where the nodes do not exchange side information for some regimes.
Fragmented Objects: Boosting Concurrency of SharedLarge Objects
TLDR
This work defines the notion of fragmented objects: concurrent objects composed of a list of fragments that allow operations to manipulate each of their fragments individually, and provides an implementation of a distributed file system that utilizes coverable fragmented objects (i.e., files).
...
...

References

SHOWING 1-10 OF 24 REFERENCES
Using erasure codes efficiently for storage in a distributed system
TLDR
A new approach to maintain ensure-encoded data in a distributed system that allows the use of space efficient k-of-n erasure codes where n and k are large and the overhead n-k is small is proposed.
Optimistic Erasure-Coded Distributed Storage
TLDR
It is shown that tolerating asynchronous periods does not increase storage overhead during synchronous periods, and the used storage is lower in synchronousperiods, which are considered common in practice, as compared to asynchronous periods.
A coded shared atomic memory algorithm for message passing architectures
TLDR
An atomic shared-memory emulation algorithm that is called Coded Atomic Storage, which uses erasure coding methods and has a bounded storage cost, and an algorithm known as the Communication Cost Optimal Atomic Storage (CCOAS), which achieves a smaller communication cost than CAS and CASGC.
Space Bounds for Reliable Storage: Fundamental Limits of Coding
TLDR
The results show that the space complexity of providing reliable storage in asynchronous distributed systems is Theta(min(f,c)D), which implies that the asymptotic storage cost is either as high as with replication, namely O(fD), or as high under concurrency as with the aforementioned code-based algorithms, i.e., O(cD.
PoWerStore: proofs of writing for efficient and robust storage
TLDR
This paper implemented PoWerStore, a robust and efficient data storage protocol, and shows its improved performance when compared to existing robust storage protocols, including protocols that tolerate only crash faults.
RAMBO: A Reconfigurable Atomic Memory Service for Dynamic Networks
TLDR
An algorithm that emulates atomic read/write shared objects in a dynamic network setting that guarantees atomicity for arbitrary patterns of asynchrony and failure, and satisfies a variety of conditional performance properties.
Low-overhead byzantine fault-tolerant storage
TLDR
An erasure-coded Byzantine fault-tolerant block storage protocol that is nearly as efficient as protocols that tolerate only crashes and achieves throughput within 10% of the crash-tolerance protocol for writes and reads in failure-free runs.
Efficient Replication of Large Data Objects
TLDR
A new distributed data replication algorithm tailored especially for large-scale read/write data objects such as files that guarantees atomic data consistency, while incurring low latency costs.
Optimal Resilience for Erasure-Coded Byzantine Distributed Storage
TLDR
The first simulation of a multiple-writer multiple-reader atomic read/write register using erasure-coding in this setting that achieves optimal resilience and minimal storage overhead is provided.
Dynamic atomic storage without consensus
TLDR
It is discovered that dynamic R/W storage is solvable in a completely asynchronous system: DynaStore, an algorithm that solves this problem is presented.
...
...