Placing data as close possible to computation is a common practice of intensive systems, commonly referred the locality problem. By analyzing existing production we confirm benefit and find that have different popularity varying correlation accesses. We propose DARE, distributed adaptive replication algorithm aids scheduler achieve better locality. DARE solves two problems, how many replicas allocate for each file where place them, using probabilistic sampling competitive aging independently...

Serverless computing shows good promise for efficiency and ease-of-use. Yet, there are only a few, scattered sometimes conflicting reports on questions such as 'Why do so many companies adopt serverless?', 'When serverless applications well suited?', 'How currently implemented?' To address these questions, we analyze 89 from open-source projects, industrial sources, academic literature, scientific - the most extensive study to date.

Over the last five years, all major cloud platform providers have increased their serverless offerings. Many early adopters report significant benefits for serverless-based over traditional applications, and many companies are considering moving to themselves. However, currently there exist only few, scattered, sometimes even conflicting reports on when applications well suited what best practices implementation are. We address this problem in present study about state of applications....

The address resolution protocol (ARP) is used by computers to map network addresses (IP) physical (MAC). has proved work well under regular circumstances, but it was not designed cope with malicious hosts. By performing ARP cache poisoning or spoofing attacks, an intruder can impersonate another host (man-in-the-middle attack) and gain access sensitive information. Several schemes mitigate, detect prevent these attacks have been proposed, each its limitations. In this paper we analyze of...

The rapid adoption and the diversification of cloud computing technology exacerbate importance a sound experimental methodology for this domain. This work investigates how to measure report performance in cloud, well research community is already doing it. We propose set eight important methodological principles that combine best-practices from nearby fields with concepts applicable only clouds, new ideas about time-accuracy trade-off. show these are using practical use-case experiment. To...

As a key part of the serverless computing paradigm, Function-as-a-Service (FaaS) platforms enable users to run arbitrary functions without being concerned about operational issues. However, there are several performance-related issues surrounding state-of-the-art FaaS that can deter widespread adoption FaaS, including sizeable overheads, unreliable performance, and new forms cost-performance trade-off. In this work we, SPEC RG Cloud Group, identify six challenges arise specifically in model,...

Serverless functions are an emerging cloud computing paradigm that is being rapidly adopted by both industry and academia. In this model, the provider opaquely handles resource management tasks such as provisioning, deployment, auto-scaling. The only task developers still in charge of selecting how much resources allocated to each worker instance. However, optimal size serverless quite challenging, so often neglect it despite its significant cost performance benefits. Existing approaches...

Dispelling the confusion around serverless computing by capturing its essential and conceptual characteristics.

Intrusion detection is an important part of networked-systems security protection. Although commercial products exist, finding intrusions has proven to be a difficult task with limitations under current techniques. Therefore, improved techniques are needed. We argue the need for correlating data among different logs improve intrusion systems accuracy. show how attacks reflected in and that some not evident when single log analyzed. present experimental results using anomaly virus Yaha....

This paper presents Natjam, a system that supports arbitrary job priorities, hard real-time scheduling, and efficient preemption for Mapreduce clusters are resource-constrained. Our contributions include: i) exploration evaluation of smart eviction policies jobs tasks, based on resource usage, task runtime, deadlines; ii) work-conserving mechanism Mapreduce. We incorporated Natjam into the Hadoop YARN scheduler framework (in 0.23). present experiments from deployments test cluster, Emulab...

A huge increase in data storage and processing requirements has lead to Big Data, for which next generation systems are being designed implemented. However, we have a limited understanding of the workloads Data systems. We consider case one common type cluster: cluster dedicated supporting mix MapReduce jobs. analyze 6-month traces from two large Hadoop clusters at Yahoo! characterize file popularity, temporal locality, arrival patterns workloads. identify several interesting properties...

Fast deployment and execution of cloud functions in Function-as-a-Service (FaaS) platforms is critical, for example, microservices architectures. However, that require large packages or libraries are bloated start slowly. An optimization to cache at the worker nodes instead bundling them with functions. existing FaaS schedulers vanilla load balancers, agnostic cached response prior function executions, cannot properly reap benefits package caching. We study case package-aware scheduling...

Microservices, containers, and serverless computing belong to a trend toward applications composed of many small, self-contained, automatically managed components. Core computing, function-as-a-service (FaaS) platforms employ state-of-the-art container technology microservices-based architectures enable users manage complex without the need for system-level expertise. Victim its own success, partially due proprietary technology, currently community has limited overview these platforms. To...

In only a decade, cloud computing has emerged from pursuit for service-driven information and communication technology (ICT), becoming significant fraction of the ICT market. Responding to growth market, many alternative services their underlying systems are currently vying attention users providers. To make informed choices between competing service providers, permit cost-benefit analysis cloud-based systems, enable system DevOps evaluate tune performance these complex ecosystems,...

Data locality is a fundamental problem to data-parallel applications where data-processing tasks consume different amounts of time and resources at locations. The especially prominent under stressed conditions such as hot spots. While replication based on data popularity relieves spots due contention for single file, caused by skewed node popularity, files co-located with each other, are more complex, unpredictable, hence difficult deal with. We propose Pandas, light-weight acceleration...

We consider the problem of scheduling small cloud functions on serverless computing platforms. Fast deployment and execution these is critical, for example, microservices architectures. However, that require large packages or libraries are bloated start slowly. A solution to cache at worker nodes instead bundling them with functions. existing FaaS schedulers vanilla load balancers, agnostic any may have been cached in response prior function executions, cannot reap benefits package caching...

The Address Resolution Protocol (ARP) is used by computers to map network addresses (IP) physical (MAC). protocol has proved work well under regular circumstances, but it was not designed cope with malicious hosts. By performing ARP cache poisoning or spoofing attacks, an intruder can impersonate another host (man-in-the-middle attack) and gain access sensitive information. Several schemes mitigate, detect prevent these attacks have been proposed, each its limitations. In this paper we...

Efficient namespace metadata management is increasingly important as next-generation file systems are designed for peta and exascales. New schemes have been proposed, however, their evaluation has insufficient due to a lack of appropriate traces. Specifically, no Big Data storage system trace publicly available existing ones poor replacement. We studied traces one from Yahoo! note some the differences implications studies. discuss insufficiency approaches present first step towards...

Despite the increase in university courses and curricula on ethics of computing there are few studies about how computer science (CS) programs should account for diverse ways ethical dilemmas approaches to situated cultural, philosophical, governance systems, religions, languages. We draw experiences insights 46 educators practitioners Latin America, South-Asia, Africa, Middle East, Australian First Nations who participated surveys interviews. Our modest study seeks prompt conversation...

Multicasting at the IP layer has not been widely adopted due to a combination of technical and non-technical issues. End-system multicast (also called application-layer multicast) is an attractive alternative for reasons user management (set-up control) attack avoidance. Sessions can be established on demand such that there are no static points failure target in advance. In end-system multicast, overlay network built top available services packets multicasted application layer. The organized...

Cloud computing providers offer a variety of pricing models, complicating the client decision, as no single model is cheapest in all scenarios. In addition, small to medium-sized organizations frequently lack personnel that can navigate intricacies each model, and result, end up opting for sub-optimal strategy, leading overpaying resources or not being able meet performance goals. this paper, we: (1) present results study shows that, Ecuador, considerable percentage companies choose...

Serverless computing services, such as Function-as-a-Service (FaaS), hold the attractive promise of a high level abstraction and performance, combined with minimization operational logic. Several large ecosystems serverless platforms, both open- closed-source, aim to realize this promise. Consequently, lucrative market has emerged. However, performance trade-offs these systems are not well-understood. Moreover, it is exactly opaqueness operational-side that make evaluation studies platforms...

We present the design and implementation of VisFlowConnect, a powerful new tool for visualizing network traffic flow dynamics situational awareness. The visualization capability provided by VisFlowConnect allows an operator to assess state large complex given overall view entire filter/drill-down features with friendly user interface that users request more detailed information interest such as specific protocol flows. value specifically security awareness is any event, only few minor...