Erasure Coding Vs. Replication: A Quantitative Comparison

By Hakim Weatherspoon, John D. Kubiatowicz. International Workshop on Peer-to-Peer Systems, 2002

1,049 citations recorded by the source corpus.

Subfields: Computer Networks and Communications

View Paper: https://www.semanticscholar.org/paper/25e5d5a046afa5fcde7be23d087ae69f4b438e13

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

Citations in the corpus

Availability in Globally Distributed Storage Systems (2010, methodology): “In [33], replication is compared with Reed-Solomon with respect to storage requirement, bandwidth for write and repair and disk seeks for reads.”
A Solution to the Network Challenges of Data Recovery in Erasure-coded Distributed Storage Systems: A Study on the Facebook Warehouse Cluster (2013, background): “In [8], the authors perform a theoretical comparison of replication and erasure-codes for peer-to-peer storage systems.”
MR-PDP: Multiple-Replica Provable Data Possession (2008, background): “These include birth/death models [11, 29], replica turnover or “churn” [27], and the lifetime of data [22].”
Erasure Coding in Windows Azure Storage (2012, background): “The advantage of erasure coding over simple replication is that it can achieve much higher reliability with the same storage, or it requires much lower storage for the same reliability [7].”
Pond: The OceanStore Prototype (2003, result): “A detailed analysis confirming this intuition can be found in our earlier work [36].”
A Survey on Network Codes for Distributed Storage (2010, background): “To realize the increased reliability of coding however, one has to address the challenge of maintaining an erasure encoded representation.”
Designing a DHT for Low Latency and High Throughput (2004, background): “First, it determines a block’s availability when nodes fail [46].”
Using redundancy to cope with failures in a delay tolerant network (2005, background): “Encour-agingly, some basic studies show the advantages of using erasure coding with replication [23, 16].”
Survey of research towards robust peer-to-peer networks: Search methods (2006, background): “Storage Consistency & Replication [101-112] Eventual consistency Trade-offs Distribution [39, 42, 90, 92, 113-131] Epidemics, Bloom Filters Fault Tolerance [40, 105, 132-139] Erasure Coding Byzantine Agreement Locality [24, 43, 47, 140-160] Load Balancing [37, 86, 100, 107, 151, 161-171]”
Glacier: highly durable, decentralized storage despite massive correlated failures (2005, background): “[42] show that erasure codes can achieve mean time to failures many orders of magnitude higher than replicated systems with similar storage and bandwidth requirements.”
High Availability in DHTs: Erasure Coding vs. Replication (2005, result): “While previous comparisons exist [4, 7, 17 ] they mostly argue that erasure coding is the clear victor, due to huge storage savings for the same availability levels (or conversely, huge availability gains for the same storage levels).”
Network Coding for Distributed Storage Systems (2007, methodology): “Redundancy management strategies for such systems have been evaluated in [ 28 ], [3], [2], [20], [27], [5], [25], [11].”
Erasure-coding based routing for opportunistic networks (2005, background): “Erasure codes have recently been discussed and applied to many network applications, including achieving efficient distribution of bulk data in overlay networks [1] and P2P networks [10], coping with unreliable transmissions in wireless sensor networks [13], and achieving reliability in large-scale ”
Rethinking erasure codes for cloud file systems: minimizing I/O for recovery and degraded reads (2012, background): “The performance tradeoffs between replication and erasure coding are well understood and have been evaluated in many environments, such as peer-to-peer file systems [43, 50] and open-source coding libraries [37].”
Efficient Replica Maintenance for Distributed Storage Systems (2006, background): “Most DHT evaluations consider whether data would survive a single event involving the failure of many nodes [8,40].”