Recently, Mobile Edge Computing (MEC) has used lightweight container-based microservices to provide resources for artificial intelligence applications, which will be decomposed into multiple dependent components, forming a Directed Acyclic Graph (DAG). In MEC, users will partition the input of the computation-intensive tasks into multiple sub-tasks for parallel execution acceleration. To satisfy concurrency, app vendors must deploy multiple container replicas for a microservice. Due to the limited capacity of edge servers, containers need to be deployed into geographically distributed and heterogeneous edge servers, resulting in significant inter-edge server traffic. To this end, we propose an adaptive scheme to guide microservice deployment for data partition-based applications in the MEC. We model the multi-replica microservice deployment problem as an integer programming problem to minimize operation costs. We propose a Deterministic Local Search-based Microservice Deployment algorithm (DLSMD), that chooses a superior neighborhood solution iteratively to solve it. We also formulate a more general problem considering both computing and communication time to minimize the total completion time and devise a Heuristic Microservice Deployment (HMD) algorithm to solve it. Extensive simulation results show that DLSMD and HMD outperform other benchmarks, achieving up to $4\times$ and $3.12\times$ speedups in terms of the inter-server traffic and total completion time reduction, respectively.

With the continuous enrichment of cloud services, an increasing number of applications are being deployed in data centers. These emerging applications are often communication-intensive and data-parallel, and their performance is closely related to the underlying network. With their distributed nature, the applications consist of tasks that involve a collection of parallel flows. Traditional techniques to optimize flow-level metrics are agnostic to task-level requirements, leading to poor application-level performance. In this paper, we address the heterogeneous task-level requirements of applications and propose task-aware flow scheduling. First, we model tasks’ sensitivity to their completion time by utilities. Second, on the basis of Nash bargaining theory, we establish a flow scheduling model with heterogeneous utility characteristics, and analyze it using Lagrange multiplier method and KKT condition. Third, we propose two utility-aware bandwidth allocation algorithms with different practical constraints. Finally, we present Tasch, a system that enables tasks to maintain high utilities and guarantees the fairness of utilities. To demonstrate the feasibility of our system, we conduct comprehensive evaluations with real-world traffic trace. Communication stages complete up to 1.4 $\mathrm{\times }$ faster on average, task utilities increase up to 2.26 $\mathrm{\times }$, and the fairness of tasks improves up to 8.66 $\mathrm{\times }$ using Tasch in comparison to per-flow mechanisms.