Jia Tong

An Analysis on Availability Commitment and Penalty in Cloud SLA

Availability is the most essential attribute of qualities of cloud services. Most popular public cloud services claim availability commitments with corresponding penalties in their SLAs. However, lack of clarity in availability commitment and penalty make it hard for consumers to understand the SLAs well and furthermore, to compare different cloud providers under different contexts, which would even become an obstacle for enterprise consumers to embrace public clouds. We present a cloud SLA availability commitment framework including availability calculation and penalty calculation services, and compare SLAs of well-known public IaaS cloud providers with investigation of their merits and defects. We also present a business model for cloud providers to find the optimal penalty degree for their SLAs, which will help in defining availability based SLA of cloud services.

A Competitive Penalty Model for Availability Based Cloud SLA

Availability is one of the most essential attributes of qualities of cloud services. Most popular public cloud services claim availability commitments with corresponding penalties in their SLAs. To gain the maximal profits, cloud providers should choose an optimal penalty strategy in the competitive cloud market. In this paper, we firstly survey the penalty calculation methods of cloud providers. Based on the survey, we propose a competitive penalty model and a corresponding penalty based profit maximization algorithm for cloud providers. According to the model, each cloud provider would choose the best fit penalty strategy to gain the maximal expected profit during the game procedure. The proposed model is evaluated with real data of popular cloud providers with sensitive analysis, and is valuable for cloud providers to define their penalty strategy.Availability is one of the most essential attributes of qualities of cloud services. Most popular public cloud services claim availability commitments with corresponding penalties in their SLAs. To gain the maximal profits, cloud providers should choose an optimal penalty strategy in the competitive cloud market. In this paper, we firstly survey availability and penalty calculation methods of cloud providers. Based on the survey, we propose a competitive penalty model and a corresponding penalty based profit maximization algorithm for cloud providers. According to the model, each cloud provider would choose the best fit penalty strategy to gain the maximal expected profit during the game procedure. The proposed model is evaluated with real data of popular cloud providers with sensitive analysis, and is valuable for cloud providers to define their penalty strategy.

Characterizing and Predicting Bug Assignment in OpenStack

Open source software is becoming increasingly important in cloud computing. However, many cloud computing systems suffer from software bugs that cause significant dependability issues. Bug assignment and fixing are crucial parts of software maintenance to improve dependability. In this paper, we conduct an empirical study of 42,880 bug reports from OpenStack bug repository. We study the characteristics (e.g., distribution of bugs, distribution of assignees) of bug assignments in OpenStack and find the bug assignment pattern which we call as long tail. The findings can support the follow-up research on improving efficiency of bug assignment, that is, we propose a prediction method based on long tail model, and experimentally evaluate this method by applying it to OpenStack bug assignment.

Time Series Based Killer Task Online Recognition Service: A Google Cluster Case Study

To better understand task failures in cloud computing systems, we analyze failure frequency of tasks based on Google cluster dataset, and find what we call as killer tasks that suffer from long-term failures and repeated rescheduling. Killer task can be a big concern of cloud systems as it causes unnecessary resource wasting and significant increase of scheduling workloads. Hence there is a need to provide a service for cloud system operators to recognize killer tasks in time. In this paper, we propose an online killer task recognition service based on the resource usage time series which can recognize killer tasks at the very early stage of their occurrence so that they can be handled appropriately instead of being rescheduled. The experiment results show that the proposed service performs a 93.6% accuracy in recognizing killer tasks with an 87% timing advance and 86.6% resource saving for the cloud system averagely.

An Approach to Pinpointing Bug-Induced Failure in Logs of Open Cloud Platforms

Software bugs have been one of the dominantcauses of system failures, especially in cloud systems basedon open source platforms. One big challenge fortroubleshooting these cloud systems is to pinpoint thesoftware bug-induced failure in large and complex log fileswhich is a nightmare for administrators. So far, there hasbeen little study on how to identity bug-induced failuresbased on log analysis. In this paper, we analyze and describefeatures of bug-induced failure logs from bug repository and Q&A websites, and then propose a general automaticapproach to pinpoint logs of bug-induced failure from logfiles of open cloud platform. In the approach, two algorithmscalled MPIN and SPIN are presented for log classification. We evaluate our approach by applying logs collected frombug repositories of OpenStack and Hadoop, and five Q&A websites. The experimental result shows that the proposedapproach can identify logs of bug-induced failure inOpenStack logs with 83.9% precision, and for Hadoop logswith 82.52% precision.

An Approach for Cross-Community Content Recommendation: A Case Study on Docker

With the boom of open source software, open source communities are formed and involved in software development, deployment and application with unprecedented level. However, the rapid expansion of open source communities results in a lot of redundant contents within the community, and most importantly, among communities since they overlap each other with shared issues. On the one hand, redundant contents that are expressed in informal free texts highly increase the size of contents, which makes people suffering from finding what they exactly need from communities; on the other hand, these communities are mutually complementary that the knowledge sharing across communities can be very beneficial to users. It is crucial to recommend content for users’ need through retrieving knowledge across communities. Current studies mainly focus on acquiring knowledge from one specific community to treat communities as isolated islands, and few of them have tackle the problem of content recommendation across multiple communities. In this paper, we firstly analyze five popular open source communities, and then propose an approach of cross-community content recommendation based on LDA topic model, integrating and distilling information from multiple communities to make knowledge acquisition easier and more efficient. Taking Docker as the case study, extensive experiments show that after performing a cross-community recommendation, more than 34 % overall unanswered questions find matched answers when similarity threshold β is set to 0.85. When setting β to 0.6, almost 90 % unanswered question can be answered with existing community content. It effectively leverages various communities to recommend valuable content to users.