Search

Cloud deployments now increasingly provision FPGA accelerators as part of virtual instances. While commercial clouds still essentially expose single-tenant FPGAs to the users, the growing demand for hardware acceleration raises the need for architectures supporting FPGA multi-tenancy. In this work, we explore the trade-off between hardware consolidation and performance. Experiments show that FPGA multi-tenancy increases hardware utilization and decreases IO performance in the order of microseconds compared to the single-tenant model. The experiments also demonstrate that implementing on-chip communication between the hardware workloads of a cloud user significantly reduces the overall communication overhead.

Cloud computing has recently emerged as a compelling method for deploying and delivering services over the Internet. In this paper, we aim to shed new light on the learning of cloud user interest. Our study for the first time shows the existence of cloud users in such real-world content distribution systems as BitTorrent. Based on this observation, we further explore the similarity of content preferences between cloud and non-cloud users. Surprisingly, our statistical model analysis indicates that the users in the cloud AS have significantly different interests from all the observed non-cloud ASes. More dedicated researches are therefore required to better manage this elevating yet unique cloud traffic in the future.

Over the last decade, cloud computing has changed the IT world, mainly in the ways of storing, accessing and securing the users' data. However, there is no single cloud management platform in the market that can be used to manage different clouds (ex: AWS and AZURE), together with container communication between the clouds, to cater different cloud user requirements. We have proposed a platform for cloud service providers that automates the multi-cloud management related services with following modules: (1) communication, (2) resource allocation and (3) log view. Our results show that, using our proposed modules, the multi-cloud management related services can be automated efficiently and reliably.

Many scientific research communities and institutions are adopting the cloud as a primary platform to support their computing needs. Rapid adoption of the cloud for scientific computing is a result of the simplicity with which an individual researcher can obtain large amounts of customized compute and storage resources. At the same time, most public cloud providers have rolled out many advanced networking services. Many of these services, like their compute services, are simple to access by any cloud user (e.g. routing between regions and private networking spaces within a cloud). However, it is not possible for most researchers to access expensive low-level, externally facing cloud network services without complicated support by campus IT staff, as well as national and regional network providers. This paper describes how to use Chameleon, ExoGENI, and Internet2’s Cloud Connect service to deploy research experiments that use AWS Direct Connect without requiring a privately owned Direct Connect endpoint or support from local campus IT staff.

The IaaS cloud computing model is the foundation over which all other cloud computing services can be built. Thus, the migration from the traditional datacenter model to the cloud-computing model is highly dependent on the guarantees that the IaaS Cloud Provider can offer Cloud User to sustain service quality. These guarantees are expressed using Service Level Agreement (SLA) whose need to be aligned and tightly coupled with the cloud service lifecycle so as to reflect the financial and technical dynamicity of the SLA. However, we argue that the SLAs used at present are deficient in expressing solid guarantees on the service quality and lack the auto-negotiation feature required by cloud services. In that respect, SLAs used in practice do not fully abide by the "Pay as you Go" model for cloud computing. In this paper, we propose the "Get as you Pay" model that extends the "Pay as You Go" model by augmenting it with a descriptive SLA that supports the auto-scaling of resources. For validating the "Get as you Pay" model, we walk through a scenario to show how applying the model provides an expressive and negotiable SLA.

Cloud computing is developing rapidly and the open environment in cloud computing is much more complex and unpredictable than other applications, so just providing the identity certification cannot ensure the security of cloud user and cloud service provider, in order to find a effective way to ensure cloud security, we have to solve the problem of trustiness of cloud user's behavior first, in this paper, we propose a dynamic trust evaluate method to deal with cloud user's bahavior, through using Entropy Method to reflect the essential regular patterm of user's behavior evidence, making the evaluate way become a dynamic model, weaken the subjectivity of simply using AHP, moreover, still need AHP to make the result fit people's subjective experience, so we also put forward a integrate algorithm that combine Entropy Method and AHP, in this way, the final evaluate value will keep the balance between objective and subjective and provide quantitative analysis foundation for security control, The simulation shows that the dynamic trust evaluate method can effectively distinguish user's abnormal behavior.

The trustiness of cloud platform is the key issues relating to its success or failure, and the authentication of the trustiness of user behavior is an important part of ensuring the credibility of cloud platform. In order to solve the problem of trustiness of cloud users' behaviors, a cloud user behavior authentication model based on multi-partite graphs is proposed, including the layer of user behavior evidence, the layer of building behavior multi-partite graphs and the layer of behavior authentication; Identity re-certification and Risk Game are introduced to enhance security and accuracy of the model. Simulation experiments show that, the model is accurate and effective in measuring the normal behavior of cloud users and in distinguishing malicious user with the risk user, and has a higher detection ratio and a lower false positive ratio.

In cloud computing, computational resources are provided to remote users in the form of leases. For a cloud user, he/she can request multiple cloud services simultaneously. In this case, parallel processing in the cloud system can improve the performance. When applying parallel processing in cloud computing, it is necessary to implement a mechanism to allocate resource and schedule the tasks execution order. Furthermore, a resource allocation mechanism with preemptable task execution can increase the utilization of clouds. In this paper, we propose an adaptive resource allocation algorithm for the cloud system with preemptable tasks. Our algorithms adjust the resource allocation adaptively based on the updated of the actual task executions. And the experimental results show that our algorithms works significantly in the situation where resource contention is fierce.