On the service capacity region of accessing erasure coded content

Abstract

Cloud storage systems generally add redundancy in storing content files such that K files are replicated or erasure coded and stored on N > K nodes. In addition to providing reliability against failures, the redundant copies can be used to serve a larger volume of content access requests. A request for one of the files can either be sent to a systematic node, or one of the repair groups. In this paper, we seek to maximize the service capacity region, that is, the set of request arrival rates for the K files that can be supported by a coded storage system. We explore two aspects of this problem: 1) for a given erasure code, how to optimally split incoming requests between systematic nodes and repair groups, and 2) choosing an underlying erasure code that maximizes the achievable service capacity region. In particular, we consider MDS and Simplex codes. Our analysis demonstrates that erasure coding makes the system more robust to skews in file popularity than simply replicating a file at multiple servers, and that coding and replication together can make the capacity region larger than either alone.

Load Balancing Performance in Distributed Storage with Regular Balanced Redundancy

“…Storage schemes we consider fall into the class of combinatorial batch codes [15]. The second approach asks the question other way around and seeks to find the set of all object demand vectors that can be supported by a system with a given storage scheme, namely the system's service capacity region [16], [17]. Our treatment of load balancing falls into this second approach.…”

“…Our treatment of load balancing falls into this second approach. [16] and [17] only address the case where each node stores a single object, their approach being to find the system's complete service capacity region. However determining this region with multiple objects at each node appears to be a largely intractable problem.…”

Service Rate Region of Content Access from Erasure Coded Storage

“…The two file case is considered in [9]. The situation becomes increasingly complex depending on the number of files K in the system.…”

“…In [9], the achievable service rate region was found for some common classes of codes, such as maximum-distanceseparable (MDS) codes and simplex codes. That paper also determined the service rate region when K = 2, with arbitrary numbers of systematic and coded nodes.…”

On Distributed Storage Allocations of Large Files for Maximum Service Rate

“…Most of the work on data access in DSSs is concerned with the download latency (see e.g., [1]- [4] and references therein). It has recently been recognized, that another important metric that measures the availability of the stored data is the service rate [5], [6]. Maximizing service rate (or the throughput) of a distributed system helps support a large number of simultaneous system users.…”