Lightweight Framework for Reliable Job Scheduling in Heterogeneous Clouds

It is crucial to ensure reliability, security and stability of cloud services without sacrificing too much resources in the area of workload management in clouds. The paper evaluates and compares lightweight decentralized algorithms for scheduling a workload part of which could be unreliable, in the context of heterogeneous cloud data centers. This unreliability could be caused by various types of failures or attacks. The framework for robust workload scheduling efficiently combines classic fault tolerant and security tools, such as packet/job scanning, with workload scheduling, and it does not use any heavy resource consuming tools, e.g., cryptography or non-linear optimization. More specifically, the framework uses a novel objective function to allocate jobs to servers and constantly decides which job to scan based on a formula associated with the objective function. In previous work it was shown how to set up the objective function and the corresponding scanning procedure of the central job scheduler to make the system provably stable, provided a specific capacity condition is satisfied. As a result, it was shown that the framework assures cloud stability even though naive scanning-all and scanning-none strategies are not stable for both centralized and decentralized scheduling in homogeneous data centers. In this work we extend the work to heterogeneous data centers, for which we show that decentralized algorithms based on Join Shortest Queue and Join Shortest Work policies are stable for every workload within the system capacity, while the algorithms based on popular Power of Two Choices, Round Robin and Uniform Random policies are not stable for a substantial amount of workloads even within the system capacity