Search

This corrigendum corrects and extends our results on the benefit of peer locking in mitigating the propagation of route leaks on the Internet, originally published in [2]. The updated results show even higher benefits of peer locking than originally reported, and an extended analysis covering additional peer locking deployment scenarios shows partial deployments also yield significant reduction in propagation of leaked routes.

Building recommendation systems for enterprise software has many unique challenges that are different from consumer-facing systems. When applied to different organizations, the data used to power those recommendation systems vary substantially in both quality and quantity due to differences in their operational practices, marketing strategies, and targeted audiences. At Salesforce, as a cloud provider of such a system with data across many different organizations, naturally, it makes sense to pool data from different organizations to build a model that combines all values from different brands. However, multiple issues like how do we make sure a model trained with pooled data can still capture customer specific characteristics, how do we design the system to handle those data responsibly and ethically, i.e., respecting contractual agreements with our clients, legal and compliance requirements, and the privacy of all the consumers. In this proposal, We present a framework that not only utilizes enriched user-level data across organizations, but also boosts business-specific characteristics in generating personal recommendations. We will also walk through key privacy considerations when designing such a system.

As more and more cloud providers attracting users to participate in technological innovation activities, we use the extended Hotelling model to analyze the influence of user participation on cloud provider's equilibrium price , profits and competitive strategy from the perspective of cloud market competition. On the basis of these, we apply the relevant knowledge of game theory and consider the costs and performance of user participation in innovation. In this paper, we conclude that user participation in innovation strategy can help cloud providers to make the reasonable strategy and obtain competitive advantage.

The Tier-1 ISPs have been considered the Internet's backbone since the dawn of the modern Internet 30 years ago, as they guarantee global reachability. However, their influence and importance are waning as Internet flattening decreases the demand for transit services and increases the importance of private interconnections. Conversely, major cloud providers -- Amazon, Google, IBM, and Microsoft-- are gaining in importance as more services are hosted on their infrastructures. They ardently support Internet flattening and are rapidly expanding their global footprints, which enables them to bypass the Tier-1 ISPs and other large transit providers to reach many destinations.In this paper we seek to quantify the extent to which the cloud providers' can bypass the Tier-1 ISPs and other large transit providers. We conduct comprehensive measurements to identify the neighbor networks of the major cloud providers and combine them with AS relationship inferences to model the Internet's AS-level topology to calculate a new metric, hierarchy-free reachability, which characterizes the reachability a network can achieve without traversing the networks of the Tier-1 and Tier-2 ISPs. We show that the cloud providers are able to reach over 76% of the Internet without traversing the Tier-1 and Tier-2 ISPs, more than virtually every other network.

Online services all seek to provide their customers with the best Quality of Experience (QoE) possible. Milliseconds of delay can cause users to abandon a cat video or move onto a different shopping site, which translates into lost revenue. Thus, minimizing latency between users and content is crucial. To reduce latency, content and cloud providers have built massive, global networks. However, their networks must interact with customer ISPs via BGP, which has no concept of performance.The shortcomings of BGP are many and well documented, but in this paper we ask the community to take a step back and rethink what we know about BGP. We examine three separate studies of performance using large content and cloud provider networks and find that performance-aware routing schemes rarely achieve lower latency than BGP. We lay out a map for research to further study the idea that beating BGP may be more difficult than previously thought.

The growing demand for an ever-increasing number of cloud services is profoundly transforming the Internet's interconnection or peering ecosystem, and one example is the emergence of "virtual private interconnections (VPIs)". However, due to the underlying technologies, these VPIs are not publicly visible and traffic traversing them remains largely hidden as it bypasses the public Internet. In particular, existing techniques for inferring Internet interconnections are unable to detect these VPIs and are also incapable of mapping them to the physical facility or geographic region where they are established.In this paper, we present a third-party measurement study aimed at revealing all the peerings between Amazon and the rest of the Internet. We describe our technique for inferring these peering links and pay special attention to inferring the VPIs associated with this largest cloud provider. We also present and evaluate a new method for pinning (i.e., geo-locating) each end of the inferred interconnections or peering links. Our study provides a first look at Amazon's peering fabric. In particular, by grouping Amazon's peerings based on their key features, we illustrate the specific role that each group plays in how Amazon peers with other networks.

There is an increasing demand among machine learning researchers for powerful computational resources to train their machine learning models. In order to train these models in a reasonable amount of time, the training is often distributed among multiple machines; yet paying for such machines is costly. DeepMarket attempts to reduce these costs by creating a marketplace that integrates multiple computational resources over a distributed tensorflow framework. Instead of requiring users to rent expensive resources from a third party cloud provider, DeepMarket will allow users to lend their computing resources to each other when they are available. Such a marketplace, however, requires a credit mechanism that ensures users receive resources in proportion to the resources they lend to others. Moreover, DeepMarket must respect users' needs to use their own resources and the resulting limits on when resources can be lent to others. This Demo will introduce the audience to PLUTO: DeepMarket's intuitive graphical user interface. The audience will be able to see how PLUTO in coordination with DeepMarket servers tracks the performance of each user's training jobs, matches jobs to resources made available by other users, and tracks the resulting credits that regulate the exchange of resources.

The network communications between the cloud and the client have become the weak link for global cloud services that aim to provide low latency services to their clients. In this paper, we first characterize WAN latency from the viewpoint of a large cloud provider Azure, whose network edges serve hundreds of billions of TCP connections a day across hundreds of locations worldwide. In particular, we focus on instances of latency degradation and design a tool, BlameIt, that enables cloud operators to localize the cause (i.e., faulty AS) of such degradation. BlameIt uses passive diagnosis, using measurements of existing connections between clients and the cloud locations, to localize the cause to one of cloud, middle, or client segments. Then it invokes selective active probing (within a probing budget) to localize the cause more precisely. We validate BlameIt by comparing its automatic fault localization results with that arrived at by network engineers manually, and observe that BlameIt correctly localized the problem in all the 88 incidents. Further, BlameIt issues 72X fewer active probes than a solution relying on active probing alone, and is deployed in production at Azure.

Cloud computing is one of the most advanced technologies in information technology, a convergence of many achievements in research and development and application of new technologies. Cloud computing has also helped to reduce the cost of small and medium enterprises based on cloud provider services. As cloud computing evolves rapidly, researching optimizations such as task execution time, completion time, responce time, and virtual machine resources (VMs) are tremendous challenges. This article proposes an MMSIA algorithm to improve the Max-Min scheduling algorithm, which improves the completion time of the requests by using the "learned learning" machine learning, by clustering size of requests and clustering utilization percent of VMs. The algorithm then assigns the largest cluster requests to the VM with the least utilization percent, which is repeated when the request list is empty. In particular, the MMSIA algorithm has improved the completion time. The simulation results show that the proposed MMSIA algorithm has improved the completion time compared to the three algorithms: Max-Min, Min-Min and Roud Robin.

The cloud computing is a powerful technology in IT environment to provide cost effective, easy to manage, elastic, provide on-demand scalable services over the Internet and is wish list of many organizations that they are still concerned about the security risks implied by moving their infrastructure into cloud computing, and by integrating their resources within the cloud computing environment.. To take advantage of all the benefits of cloud computing, it is necessary to have security and privacy requirements for the implementation and use of the cloud. A company planning to implement a public cloud solution should be aware of the risks associated with cloud computing. This is due to the rapidly evolving threats to cloud security that are emerging at an increasing rate in recent years, also to multi-tenant and open nature of the public cloud. A risk analysis must be done initially with a cloud provider in order to identify the existing vulnerabilities and to implement the measures to counter those threats. This paper explores the key issues surrounding cloud security, and passes on valuable guidance about not only information security and broadly risk. Also, it will help providers to make informed security decisions about their diverse cloud set-ups and better understand how to reap all the benefits of cloud without compromising their organizations' security. As well, it discusses the security requirements, and various risks related to the public cloud.

We propose VNF chain abstraction to decouple a tenant's view of the VNF chain from the cloud provider's implementation. We motivate the benefits of such an abstraction for the cloud provider as well as the tenants, and outline the challenges a cloud provider needs to address to make the chain abstraction practical. We describe the design requirements and report on our initial prototype.

As the cloud computing market matures, effective pricing design remains a challenging problem. Electricity is a major contributor to the cloud's recurring costs and may increase significantly in the near future. As the cloud market gets more competitive, there will be a need for more cost-conscious dynamic pricing design. Additionally, today's cloud provider's pricing mechanisms act almost oblivious to the fact that the large tenants of the cloud are able to perform automated demand response in the form of delaying and dropping jobs. In this work we introduce differential dynamic pricing which provides different tenants with tailored prices according to their price sensitivity, while accounting for each tenant's energy costs. Furthermore, the cloud providers can engage in demand response to utility prices using on-site batteries. Battery management complicates things further, as it brings in additional costs, e.g., wear-and-tear, which is challenging to attribute to each tenant. So, we use virtual battery management which accounts for individual tenant's usage of batteries. We also, show how virtual battery management can alleviate possible price fluctuations and act in synergy with differential dynamic pricing scheme and result in a less variable and more predictable (stable) pricing framework, called Stable Differential Dynamic Pricing (SD2P). We navigate different real-world IBM tenants' incident power demands with different price elasticities in conjunction with real-world electricity prices, and realistic battery degradation costs, to show the efficacy of our pricing mechanism. We show that under SD2P cloud's profit increases up to 31% while average tenants' utility increases up to 18%. Prices under SD2P exhibit up to 54% less variation compared to other dynamic pricing schemes, as a result of engaging the batteries at times of energy price spikes.

Whilst there are many attributes that need to be considered for cloud service selection, performance remains one of the most crucial aspects. Thus, we argue for a transparent cloud provider comparison framework in this study. We initiate the development of TCloud: a transparent framework for public cloud service comparison. Our framework helps prospective cloud users to decipher public cloud benchmarking data and appraise the performance of public cloud services relative to their performance goals. We carried out experiments on the real public cloud environment to implement our framework and demonstrated how prospective cloud users can use the TCloud framework in understanding how well virtualized public cloud resources meet their application requirements. Unlike previous studies, the TCloud framework presents a more realistic method of appraising the performance of virtualized resources in the public cloud. TCloud is unique in the sense that it collates public cloud benchmarking data and correlates the observed performance metrics to prospective cloud users' actual application workload requirements.

Services offered by cloud providers (e.g., for key-value storage and for redirecting clients to nearby data centers) simplify the development of applications deployed in the cloud. However, since any cloud provider’s service offerings are far from complete, many cloud providers host a service marketplace, wherein third parties can advertise new services.In this paper, we argue that existing cloud service marketplaces need to be redesigned so that third-parties can not only offer new independent services but also augment the capabilities of existing cloud services. Enabling efficient implementation of such third-party add-ons requires a careful examination of the API that the cloud provider should expose and the manner in which add-ons can be interposed on interactions between applications and cloud services. We discuss how recent advances in software-defined networking with P4 can be leveraged to facilitate this redesign of cloud service marketplaces.

Nowadays, Cloud computing became one of the most important fields in computer paradigm researches, because it provides many advantages and benefits for all users whether the organizations or the individuals, by granting them the needed resources and services. Service level agreement (SLA) is an essential topic in terms of cloud computing, it is a contractual agreement signed between cloud user and cloud provider in order to manage and define the level of access for the resources and quality of services provided. Continues changes of technologies and requirements for both user and provider make updating SLA necessary, to ensure that services provided are meeting their needs among multiple QoS (Quality of Service) parameters. Intensive studies have been presented for building a flexible SLA. In our study, we proposed a dynamic model for updating cloud computing SLA overtime, in order to get a flexible, adaptive SLA and avoid costly violations.

Cloud computing is essential now for IT companies and for the users of IT companies. Recently researchers were focusing in peer-peer cloud computing, which also named as user as a cloud provider system. There are a few of researches discussed and implemented trust problem in peer-to-peer models. This paper presents a framework for automatically generate trust ranking for user as a cloud provider. The ranking is generated first by testing method for ranking QoS parameters, the second phase is applied trust factors to generate ranking for user as cloud provider. The generated ranking method is useful for cloud users to get benefit of their wasting resources, and other users will be trusted in renting cloud services while paying less cost.

As the cloud computing market continues to grow, the cloud platform is becoming an attractive target for attackers to disrupt services and steal data, and to compromise resources to launch attacks. In this paper, using three months of NetFlow data in 2013 from a large cloud provider, we present the first large-scale characterization of inbound attacks towards the cloud and outbound attacks from the cloud. We investigate nine types of attacks ranging from network-level attacks such as DDoS to application-level attacks such as SQL injection and spam. Our analysis covers the complexity, intensity, duration, and distribution of these attacks, highlighting the key challenges in defending against attacks in the cloud. By characterizing the diversity of cloud attacks, we aim to motivate the research community towards improving future security solutions for cloud systems.

In the age of cloud computing, companies still have the problem to manage access rights for resources. This is especially true, if companies are combined to virtual organizations and want to share resources that are located at cloud providers. For a consistent authorization model, an up to date knowledge about partner organizations is indispensable.This contribution proposes an approach to request the automatic deployment of resources from a cloud provider. The access rights to the resources are managed and administered by the proprietary company, even if partner organizations are involved. They are not published to the cloud provider, but remain in the owning company. This establishes a separation of resources (i.a. systems) and authorization, which alleviates security risks. Attackers of resources can not access them because the authorization model is not implemented on the same location as the resources. This makes the intrusion much more complex.

Garbled circuits offer a powerful primitive for computation on a user's personal data while keeping that data private. Despite recent improvements, constructing and evaluating circuits of any useful size remains expensive on the limited hardware resources of a smartphone, the primary computational device available to most users around the world. In this work, we develop a new technique for securely outsourcing the generation of garbled circuits to a Cloud provider. By outsourcing the circuit generation, we are able to eliminate the most costly operations from the mobile device, including oblivious transfers. Our proofs of security show that this technique provides the best security guarantees of any existing garbled circuit outsourcing protocol. We also experimentally demonstrate that our new protocol, on average, decreases execution time by 75% and reduces network costs by 60% compared to previous outsourcing protocols. In so doing, we demonstrate that the use of garbled circuits on mobile devices can be made nearly as practical as it is becoming for server-class machines.

Resource metering in distributed environments is notoriously difficult. In clouds, the metering technology has to be distributed by design. This paper shows that the technology used for (financial) micropayments fulfills several objectives of the new metering process once the concept of money is replaced with the concept of a resource. Like in micropayments, resource metering in clouds is often a 3-party relationship like that among cloud provider, service provider and customer. The technology proposed in this paper can be viewed both as a better alternative to traditional micropayments as well as a new way to meter resource usage in clouds.

To enhance the reliability of cloud services, many application providers leverage multiple cloud providers for redundancy. Unfortunately, such techniques fail to recognize that seemingly independent redundant clouds may share third-party infrastructure components, e.g., power sources and Internet routers, which could potentially undermine this redundancy.This paper presents iRec, a cloud independence recommender system. iRec recommends at best-effort independent redundancy services to application providers based on their requirements, minimizing costly and ineffective redundancy deployments. At iRec's heart lies a novel protocol that calculates the weighted number of overlapping infrastructure components among different cloud providers, while preserving the secrecy of each cloud provider's proprietary information. We sketch the iRec design, and discuss challenges and practical issues.

Data redundancy is key to preventing data loss and achieving fault-tolerance in cloud storage. Cloud storage provider usually charges users according to the requested level of redundancy. However, a cloud provider may fail to offer the committed level of data redundancy intentionally or accidentally, but the users may not be able to detect such breach until data and economic losses have occurred. In this paper, we propose a scheme to allow users remotely assess the actually deployed data redundancy in the cloud storage without knowing the file layout information. Our layout-free scheme establishes response-time profiles for each different file placements, and uses these profiles to evaluate the level of deployed data redundancy. Our experimental results show the feasibility of our scheme on both an in-house cloud and a public cloud.

Elasticity, the ability to rapidly scale resources up and down on demand, is an essential feature of public cloud platforms. However, it is difficult to understand the elasticity requirements of a given application and workload, and if the elasticity provided by a cloud provider will meet those requirements. We introduce the elasticity mechanisms of a typical Infrastructure as a Service (IaaS) cloud platform (inspired by Amazon EC2). We have enhanced our Service Oriented Performance Modelling method and tool to model and predict the elasticity characteristics of three realistic applications and workloads on this cloud platform. We compare the pay-as-you-go instance costs and end-user response time service level agreements for different elasticity scenarios. The model is also able to predict the elasticity requirements (in terms of the maximum instance spin-up time) for the three applications. We conclude with an analysis of the results.