Abstract-Analyzing behavioral patterns of workloads is critical to understanding Cloud computing environments. However, until now only a limited number of real-world Cloud datacenter tracelogs have been available for analysis. This has led to a lack of methodologies to capture the diversity of patterns that exist in such datasets. This paper presents the first large-scale analysis of real-world Cloud data, using a recently released dataset that features traces from over 12,000 servers over the period of a month. Based on this analysis, we develop a novel approach for characterizing workloads that for the first time considers Cloud workload in the context of both user and task in order to derive a model to capture resource estimation and utilization patterns. The derived model assists in understanding the relationship between users and tasks within workload, and enables further work such as resource optimization, energy-efficiency improvements, and failure correlation. Additionally, it provides a mechanism to create patterns that randomly fluctuate based on realistic parameters. This is critical to emulating dynamic environments instead of statically replaying records in the tracelog. Our approach is evaluated by contrasting the logged data against simulation experiments, and our results show that the derived model parameters correctly describe the operational environment within a 5% of error margin, confirming the great variability of patterns that exist in Cloud computing.

No abstract

Abstract-Cloud computing research is in great need of statistical parameters derived from the analysis of real-world systems. One aspect of this is the failure characteristics of Cloud environments composed of workloads and servers; currently, few metrics are available that quantify failure and repair times of workloads and servers at a large-scale. Workload metrics in particular are critical for characterizing and modeling accurate workload behavior, enabling more realistic workload simulation and failure scenarios of systems. This paper presents the analysis of failure data of a large-scale production Cloud environment (consisting of over 12,500 servers), and includes a study of failure and repair times and characteristics for both Cloud workloads and servers. Our results show that failure characteristics for workload and servers are highly variable and that production Cloud workloads can be accurately modeled by a Gamma distribution. Repair times range between 30 seconds to 4 days, and 25 minutes to 8 days, for workloads and servers respectively.

Abstract-As Cloud Computing becomes the trend of information technology computational model, the Cloud security is becoming a major issue in adopting the Cloud where security is considered one of the most critical concerns for the large customers of Cloud (i.e. governments and enterprises). Such valid concern is mainly driven by the Multi-Tenancy situation which refers to resource sharing in Cloud Computing and its associated risks where confidentiality and/or integrity could be violated. As a result, security concerns may harness the advancement of Cloud Computing in the market. So, in order to propose effective security solutions and strategies a good knowledge of the current Cloud implementations and practices; especially the public Clouds; must be understood by professionals. Such understanding is needed in order to recognize attack vectors and attack surfaces. In this paper we will propose an attack model based on a threat model designed to take advantage of Multi-Tenancy situation only. Before that, a clear understanding of Multi-Tenancy, its origin and its benefits will be demonstrated. Also, a novel way on how to approach Multi-Tenancy will be illustrated. Finally, we will try to sense any suspicious behavior that may indicate to a possible attack where we will try to recognize the proposed attack model empirically from Google trace logs. Google trace logs are a 29-day worth of data released by Google. The data set was utilized in reliability and power consumption studies, but not been utilized in any security study to the extent of our knowledge.

No abstract