No abstract

No abstract

No abstract

No abstract

Summary Network virtualization (NV) technologies have attracted a lot of attention as an essential solution for future networking infrastructure. The NV enables multiple tenants to share the same physical infrastructure and to create independent virtual networks (VNs) by decoupling the physical network in terms of topology, address, and control functions. One feasible way to realize full NV involves considering solutions based on the software‐defined networking (SDN) paradigm using its programmability. The SDN contributes many benefits to both network operations and management including programmability, agility, elasticity, and flexibility. There are several SDN‐based NV solutions; however, they suffered from a lack of scalability, high availability. Also, they have high latency between control and data plane because of proxy‐based architecture. In this thesis, we introduce a new NV platform, named Open Network Hypervisor (ONVisor). The design objectives include, among the features, (1) multitenancy, (2) scalability, (3) flexibility, (4) isolated VNs, and (5) VN federation. ONVisor was designed and implemented by extending Open Network Operating System, an open‐source SDN controller. The main features of ONVisor are (1) isolated control and data plane per VN, (2) support of distributed operations, (3) extensible translators, (4) on‐platform VN application development and execution, and (5) support of heterogenous SDN data‐plane implementations. Several experiments are conducted on various test scenarios in different test environments in terms of control and data plane performance compared to nonvirtualized SDN network. The results show that ONVisor can provide VNs a little bit lower control plane performance and similar data plane performance.

We propose a novel congestion-aware routing protocol, which leverages local and remote congestion information for routing decisions. Path congestion is difficult to measure, therefore congestion aware protocols consider only local status. This protocol uses an innovative representation of congestion for router-router links. These congestion statistics are then distributed via an aggregation protocol to other routers in the network. Using this efficient and scalable approach, we are able to make routing decisions based on both local and eventual next-hop status.We compare our results with Shortest Path only routing (SP), and the Hash-Threshold variant of ECMP (HTE). We show that our algorithm can achieve up to N times the throughput of SP, where N is the number of distinct paths from source to destination. The protocol itself is media independent, but for test purposes we have employed Ethernet.

The DAQ/HLT system of the ATLAS experiment at CERN, Switzerland, is being commissioned for first collisions in 2009. Presently, the system is composed of an already very large farm of computers that accounts for about one-third of its final event processing capacity. Event selection is conducted in two steps after the hardware-based Level-1 Trigger: a Level-2 Trigger processes detector data based on regions of interest (RoI) and an Event Filter operates on the full event data assembled by the Event Building system. The detector read out is fully commissioned and can be operated at its full design capacity. This places the responsibility on the High-Level Triggers system to select only events of highest physics interest that will finally reach the offline reconstruction farms. This paper brings an overview of the current ATLAS DAQ/HLT implementation and performance based on studies originated from its operation with simulated, cosmic particles and first-beam data. Its built-in event processing parallelism is presented and discussed.

Flow graphs provide an explicit description of the parallelization of an application by mapping vertices onto serial computations and edges onto message transfers. We present the design and implementation of a debugger for the flow graph based Dynamic Parallel Schedules (DPS) parallelization framework. We use the flow graph to provide both a high level and detailed picture of the current state of the application execution. We describe how reordering incoming messages enables testing for the presence of message races while debugging a parallel application. The knowledge about causal dependencies between messages enables tracking messages and computations along individual branches of the flow graph. In addition to common features such as restricting the analysis to a subset of threads or processes and attaching sequential debuggers to running processes, the proposed debugger also provides support for message alterations and for message content dependent breakpoints.