Abstract
A key component in large scale distributed analytical processing is shuffling, the distribution of data to multiple nodes such that the computation can be done in parallel. In this paper we describe the design and implementation of a communication middleware to support data shuffling for executing multi-stage analytical processing operations in parallel. The middleware relies on RDMA (Remote Direct Memory Access) to provide basic operations to asynchronously exchange data among multiple machines. Experimental results show that the RDMA-based middleware developed can provide a 75% reduction of the costs of communication operations on parallel analytical processing tasks, when compared with a sockets middleware.

Abstract
Deduplication is now widely accepted as an efficient technique for reducing storage costs at the expense of some processing overhead, being increasingly sought in primary storage systems [7, 8] and cloud computing infrastructures holding Virtual Machine (VM) volumes [2, 1, 5]. Besides a large number of duplicates that can be found across static VM images [3], dynamic general purpose data from VM volumes allows space savings from 58% up to 80% if deduplicated in a cluster-wide fashion [1, 4]. However, some of these volumes persist latency sensitive data which limits the overhead that can be incurred in I/O operations. Therefore, this problem must be addressed by a cluster-wide distributed deduplication system for such primary storage volumes.

Abstract
The ordering of events is a fundamental problem of distributed computing and has been extensively studied over several decades. From all the available orderings, total ordering is of particular interest as it provides a powerful abstraction for building reliable distributed applications. Unfortunately, deterministic total order algorithms scale poorly and are therefore unfit for modern large-scale applications. The main contribution of this paper is EPTO, a total order algorithm with probabilistic agreement that scales both in the number of processes and events. EPTO provides deterministic safety and probabilistic liveness: integrity, total order and validity are always preserved, while agreement is achieved with arbitrarily high probability. We show that EPTO is well-suited for large-scale dynamic distributed systems: it does not require a global clock nor synchronized processes, and it is highly robust even when the network suffers from large delays and significant churn and message loss.

Abstract
NoSQL databases opt not to offer important abstractions traditionally found in relational databases in order to achieve high levels of scalability and availability: transactional guarantees and strong data consistency. In this work we propose pH1, a generic middleware layer over NoSQL databases that offers transactional guarantees with Snapshot Isolation. This is achieved in a non-intrusive manner, requiring no modifications to servers and no native support for multiple versions. Instead, the transactional context is achieved by means of a multiversion distributed cache and an external transaction certifier, exposed by extending the client's interface with transaction bracketing primitives. We validate and evaluate pH1 with Apache Cassandra and Hyperdex. First, using the YCSB benchmark, we show that the cost of providing ACID guarantees to these NoSQL databases amounts to 11% decrease in throughput. Moreover, using the transaction intensive TPC-C workload, pH1 presented an impact of 22% decrease in throughput. This contrasts with OMID, a previous proposal that takes advantage of HBase's support for multiple versions, with a throughput penalty of 76% in the same conditions

Abstract
The ability of NoSQL systems to scale better than traditional relational databases motivates a large set of applications to migrate their data to NoSQL systems, even without aiming to exploit the provided schema exibility. However, accessing structured data is costly due to such exibility, incurring in a lot of bandwidth and processing unit usage. In this paper, we analyse this cost in Apache HBase and propose a new scan operation, named Prepared Scan, that optimizes the access to data structured in a regular manner by taking advantage of a well-known schema by application. Using an industry standard benchmark, we show that Prepared Scan improves throughput up to 29% and decreases network bandwidth consumption up to 20%. © 2017 ACM.

Abstract
Deduplication has proven to be a valuable technique for eliminating duplicate data in backup and archival systems and is now being applied to new storage environments with distinct requirements and performance trade-offs. Namely, deduplication system are now targeting large-scale cloud computing storage infrastructures holding unprecedented data volumes with a significant share of duplicate content. It is however hard to assess the usefulness of deduplication in particular settings and what techniques provide the best results. In fact, existing disk I/O benchmarks follow simplistic approaches for generating data content leading to unrealistic amounts of duplicates that do not evaluate deduplication systems accurately. Moreover, deduplication systems are now targeting heterogeneous storage environments, with specific duplication ratios, that benchmarks must also simulate. We address these issues with DEDISbench, a novel micro-benchmark for evaluating disk I/O performance of block based deduplication systems. As the main contribution, DEDISbench generates content by following realistic duplicate content distributions extracted from real datasets. Then, as a second contribution, we analyze and extract the duplicates found on three real storage systems, proving that DEDISbench can easily simulate several workloads. The usefulness of DEDISbench is shown by comparing it with Bonnie++ and IOzone open-source disk I/O micro-benchmarks on assessing two open-source deduplication systems, Opendedup and Lessfs, using Ext4 as a baseline. Our results lead to novel insight on the performance of these file systems.