Existing blockchain systems scale poorly because of their distributed consensus protocols. Current attempts at improving scalability are limited to cryptocurrency. Scaling under general workloads (i.e., non-cryptocurrency applications) remains an open question. This work takes a principled approach apply sharding in order improve transaction throughput scale. is challenging, however, due the fundamental difference failure models between databases and blockchain. To achieve our goal, we first...

Smart contracts have enabled blockchain systems to evolve from simple cryptocurrency platforms general transactional systems. A new architecture called execute-order-validate has been proposed in Hyperledger Fabric support parallel transactions. However, this might render many invalid transactions when serializing them. This problem is further exaggerated as the block formation rate inherently limited due other factors beside data processing, such cryptography and consensus. Inspired by...

With 5G on the verge of being adopted as next mobile network, there is a need to analyze its impact landscape computing and data management. In this paper, we both traditional emerging technologies project our view future research challenges opportunities. predicted increase 10-100× in bandwidth 5-10x decrease latency, expected be main enabler for smart cities, IoT efficient healthcare, where machine learning conducted at edge. context, investigate how can help development federated...

This paper presents a systematic and comprehensive survey on blockchain interoperability, where interoperability is defined as the ability of blockchains to flexibly transfer assets, share data, invoke smart contracts across mix public, private, consortium without any changes underlying systems. Analyzing vast landscape both research papers industry projects, we classify existing works into five categories, namely, (1) sidechains, (2) notary schemes, (3) hashed time lock (HTLC), (4) relays,...

Blockchain has come a long way - system that was initially proposed specifically for cryptocurrencies is now being adapted and adopted as general-purpose transactional system. As blockchain evolves into another data management system, the natural question how it compares against distributed database systems. Existing works on this comparison focus high-level properties, such security throughput. They stop short of showing underlying design choices contribute to overall differences. Our work...

With the emergence of hybrid blockchain database systems, we aim to provide an in-depth analysis performance and trade-offs among a few representative systems. To achieve this goal, implement Veritas BlockchainDB from scratch. For Veritas, two flavors target crash fault-tolerant (CFT) Byzantine (BFT) application scenarios. Specifically, with Apache Kafka CFT scenarios, Tendermint BFT We compare these three systems existing open-source implementation BigchainDB. BigchainDB uses Tender-mint...

In the last few years, distributed machine learning has been usually executed over heterogeneous networks such as a local area network within multi-tenant cluster or wide connecting data centers and edge clusters. these networks, link speeds among worker nodes vary significantly, making it challenging for state-of-the-art approaches to perform efficient training. Both centralized decentralized training suffer from low-speed links. this paper, we propose approach, namely NetMax, that enables...

The continuous increase in volume, variety and velocity of Big Data exposes datacenter resource scaling to an energy utilization problem. Traditionally, datacenters employ x86-64 (big) server nodes with power usage tens hundreds Watts. But lately, low-power (small) systems originally developed for mobile devices have seen significant improvements performance. These could lead the adoption such small servers, as announced by major industry players. In this context, we systematically conduct a...

While edge computing is gaining traction, organizations operating in geographically distributed locations are still using cloud to collect and post-process data. In this context, it useful analyze the performance trade-offs of cloud-only, edge-only hybrid edge-cloud processing. To facilitate analysis, we provide an analytic model validated by measurements on representative platforms. Our easy apply even without performing target hardware, as long specifications available. measurement-driven...

Wireless sensor and actuator networks (WS&ANs) are a new technology based on of small radio-enabled embedded devices that being deployed in areas such as environmental monitoring, vehicle tracking, building management, body monitoring other applications. Power sources for network nodes often limited, which imposes restrictions hardware resources their use by the underlying software. We propose wireless architecture is especially designed task home automation. Our system relies low power...

The rapid growth of blockchain systems leads to increasing interest in understanding and comparing performance at scale. In this paper, we focus on analyzing the Hyperledger Fabric v1.1 - one most popular permissioned systems. Prior works have analyzed v0.6 depth, but newer versions system undergo significant changes that warrant new analysis. Existing benchmarking are limited their scope: some consider only small networks, others scalability parts instead whole. We perform a comprehensive...

Motivated by the increasing importance of edge computing and improvements in heterogeneous low-power, wimpy systems, we present a novel time-energy-cost analysis comparison with traditional brawny cloud computing. For this analysis, use Amazon EC2 instance GPU two systems represented Jetson TK1 TX1. As paradigm shift to is due challenges Big Data processing, select six representative MapReduce applications applicability IoT Using our both compute nodes GPUs, several key insights. Firstly,...

Motivated by the increasing interest in posit numeric format, this paper we evaluate accuracy and efficiency of arithmetic contrast to traditional IEEE 754 32-bit floating-point (FP32) arithmetic. We first design implement a Posit Arithmetic Unit (PAU), called POSAR, with flexible bit-sized suitable for applications that can trade savings chip area. Next, analyze POSAR series benchmarks including mathematical computations, ML kernels, NAS Parallel Benchmarks (NPB), Cifar-10 CNN. This...

Traditional datacenter systems advocate the use of high-performance hardware, resulting in increased power consumption and cooling costs. With increasing availability having diverse performance-to-power ratios, we analyze energy efficiency mixing low-power nodes a cluster. Using model-driven analysis, predict heterogeneous mix that is most energy-efficient while maintaining given deadline. Considering service demands workloads on cores, memory I/O devices, derive Pareto-optimal...

Hybrid programming model is becoming increasingly popular for HPC applications as it has the dual-advantage of exploiting inter-node distributed-memory scalability and intra-node shared-memory performance in a cluster system. One key challenges energy efficient execution hybrid programs to determine time hardware configurations among large system configuration space. Given program with an deadline budget, we propose measurement-based analytical modelling approach these configurations. In...

Motivated by the massive energy usage of blockchain, on one hand, and significant performance improvements in low-power, wimpy systems, other we perform an in-depth time-energy analysis blockchain systems low-power nodes comparison to high-performance nodes. We use three represent a wide range performance-power spectrum, while covering both x86/64 ARM architectures. show that low-end are struggling run full-fledged blockchains mainly due their small low-bandwidth memory. On with balanced...

Database systems are paying more attention to data security in recent years. Immutable such as blockchains, verifiable databases, and ledger databases equipped with various verifiability mechanisms protect data. Such often adopt different threat models, techniques, therefore, have performance implications compared traditional database systems. So far, there is no uniform benchmarking tool for evaluating the of these systems, especially at level verification functions. In this paper, we first...

Clouds offer great flexibility for scaling applications due to the wide spectrum of resources with different cost-performance, inherent resource elasticity and pay-peruse charging. However, determining cost-time-efficient cloud configurations execute a given application in large configuration space remains key challenge. The growing importance elastic which accuracy is function consumption introduces new opportunities exploit on clouds. In this paper, we introduce CELIA, measurement-driven...

Motivated by the proliferation of both homogeneous and heterogeneous shared-memory multicore systems with large core counts, energy usage issue faced computing today we extend Amdahl's Gustafson's laws for speedup to estimate savings a system compared using single core. To define savings, introduce two key parameters, (i) active power fraction (APF) representing ratio between core's average idle system, (ii) inter-core (ICS) depicting difference in speed among different types cores...

In this paper, we propose the Blockchain Benchmark Standardized Format (BBSF), a framework for standardized, transparent, and fair benchmarks blockchains. BBSF enables users developers to compare blockchain platforms using metrics derived from realistic workloads. We outline challenges in developing benchmark with degree of breadth flexibility. contrast results our approach prior implementations show why generates that are more verifiable than published benchmarking data. present an...

Energy efficiency is an area of growing importance in datacenters. While the energy proportionality wall poses a challenge to further improve dynamic power range servers, heterogeneous systems offer new opportunity achieve higher by improving match between application workload demands on large system configuration space. This paper proposes model-driven analysis clusters consisting server nodes with different performance-to-power ratio (PPR). Our shows that inter-node heterogeneity has...