The performance of direct TCP streaming using analytical models was investigated. It was found that for both constrained and unconstrained streaming, the performance improves as the value of T/μ, where T/μ represent...
详细信息
The performance of direct TCP streaming using analytical models was investigated. It was found that for both constrained and unconstrained streaming, the performance improves as the value of T/μ, where T/μ represents how much the achievable TCP throughput is higher than the video playback rate, increases. The performance was found to be sensitive to the values of the various parameters in the models. It was observed that for both constrained and unconstrained streaming, the performance is generally good when the achievable TCP throughput is roughly twice the video bitrate, with only a few seconds of startup delay.
We describe Harpoon, a new application-independent tool for generating representative packet traffic at the IP flow level. Harpoon is a configurable tool for creating TCP and UDP packet flows that have the same byte, ...
详细信息
We describe Harpoon, a new application-independent tool for generating representative packet traffic at the IP flow level. Harpoon is a configurable tool for creating TCP and UDP packet flows that have the same byte, packet, temporal, and spatial characteristic as measured at routers in live environments. We validate Harpoon using traces collected from a live router and then demonstrate its capabilities in a series of router performance benchmark tests.
The problem of delivering large files on-demand from a server to potentially large numbers of requesting clients was studied using scalable protocols that employ application-level multicast. The average server bandwid...
详细信息
The problem of delivering large files on-demand from a server to potentially large numbers of requesting clients was studied using scalable protocols that employ application-level multicast. The average server bandwidth required, was minimized to achieve a given average or maximum client delay, or download time. A more efficient multicast protocol was considered, that turns off the multicast when no clients are listening. The multicast can be allowed to terminate after the clients whose requests arrived before commencement of the multicast have received the full file.
The microbenchmarking of network interfaces as a tool to isolate the low-level behavior that impact application-level performance, is discussed. A high-performance server NIC must support data volumes dominated by sen...
详细信息
The microbenchmarking of network interfaces as a tool to isolate the low-level behavior that impact application-level performance, is discussed. A high-performance server NIC must support data volumes dominated by sends of large frames while also efficiently receiving and sending small frames. The suite includes UDP and TCP unidirectional send and receive tests and UDP bidirectional tests, where all tests are performed separately for maximum-sized and minimum-sized frames. The results show that all NICs tested can achieve near wire-speed in sending large frames, but that the performance of the NIC varies greatly when processing bidirectional streams of large frames or unidirectional streams of small frames.
The effectiveness of the chord as a protocol for file sharing using measurements from a pear-to-peer (P2P) applications, was evaluated. The measurements were from a real P2P file-sharing protocol implementation based ...
详细信息
The effectiveness of the chord as a protocol for file sharing using measurements from a pear-to-peer (P2P) applications, was evaluated. The measurements were from a real P2P file-sharing protocol implementation based on a centralized architecture. The users sent queries to a server that maintained a database of the current peers and their shared libraries, and the user explicitly consented to anonymized logging of their actions. It was found that the number of keys stored by each peer is not necessarily decided by the size of its key space, and that the node is responsible for a very popular file key, even if it has a small key space, will have a large set of values to store for that key.
Trace files are widely used in research and academia to study the behavior of programs. They are simple to process and guarantee repeatability. Unfortunately, they tend to be very large. This paper describes vpc3, a f...
详细信息
Trace files are widely used in research and academia to study the behavior of programs. They are simple to process and guarantee repeatability. Unfortunately, they tend to be very large. This paper describes vpc3, a fundamentally new approach to compressing program traces. Vpc3 employs value predictors to bring out and amplify patterns in the traces so that conventional compressors can compress them more effectively. In fact, our approach not only results in much higher compression rates but also provides faster compression and decompression. For example, compared to bzip2, vpc3's geometric mean compression rate on SPECcpu2000 store address traces is 18.4 times higher, compression is ten times faster, and decompression is three times faster.
Networks of mobile devices and embedded systems represent a new computing platform. Typical network nodes range from sensors, cell phones, PDA's, to laptop computers. Wireless ad-hoc networks are used to connect t...
详细信息
ISBN:
(纸本)9781450377997
Networks of mobile devices and embedded systems represent a new computing platform. Typical network nodes range from sensors, cell phones, PDA's, to laptop computers. Wireless ad-hoc networks are used to connect these heterogeneous nodes, each of which has different capabilities and resources and may provide different services. Most applications targeting such networks will exploit the physical locations of the network nodes. SpatialViews is a language for applications executing on volatile networks of mobile devices and embedded systems. SpatialViews provides high- level abstractions for dynamic service discovery, location-awareness, and in-network aggregation. The compiler translates a SpatialViews program into low-level representations that use light-weight execution migration and property based routing. This paper investigates the impact of parallelization and replication on program performance metrics such as response time, energy consumption, and quality of result (QoR). For a simple application program running over networks with a range of different node failure rates, analytical modeling, simulation, and physical measurement revealed different energy/response time/QoR tradeoffs among different parallelization strategies and replication degrees.
The tightest link of a network path is the link where the end-to-end available bandwidth is limited. We propose a new probe technique, called Dual Rate Periodic Streams (DRPS), for finding the location of the tightest...
详细信息
The tightest link of a network path is the link where the end-to-end available bandwidth is limited. We propose a new probe technique, called Dual Rate Periodic Streams (DRPS), for finding the location of the tightest link. A DRPS probe is a periodic stream with two rates. Initially, it goes through the path at a comparatively high rate. When arrived at a particular link, the probe shifts its rate to a lower level and keeps the rate. If proper rates are set to the probe, we can control whether the probe is congested or not by adjusting the shift time. When the point of rate shift is in front of the tightest link, the probe can go through the path without congestion, otherwise congestion occurs. Thus, we can find the location of the tightest link by congestion detection at the receiver.
Most state-of-the-art commercial storage virtualization systems focus only on one particular storage attribute, capacity. This paper describes the design, implementation and evaluation of a multi-dimensional storage v...
详细信息
Most state-of-the-art commercial storage virtualization systems focus only on one particular storage attribute, capacity. This paper describes the design, implementation and evaluation of a multi-dimensional storage virtualization system called Stonehenge, which is able to virtualize a cluster-based physical storage system along multiple dimensions, including bandwidth, capacity, and latency. As a result, Stonehenge is able to multiplex multiple virtual disks, each with a distinct bandwidth, capacity, and latency attribute, on a single physical storage system as if they are separate physical disks. A key enabling technology for Stonehenge is an efficiency-aware real-time disk scheduling algorithm called dual-queue disk scheduling, which maximizes disk utilization efficiency while providing Quality of Service (QoS) guarantees. To optimize disk utilization efficiency, Stonehenge exploits run-time measurements extensively, for admission control, computing latency-derived bandwidth requirement, and predicting disk service time.
We investigate the origins of server-induced latency to understand how to improve latency optimization techniques. Using the Flash Web server, we analyze latency behavior under various loads. Despite latency profiles ...
详细信息
We investigate the origins of server-induced latency to understand how to improve latency optimization techniques. Using the Flash Web server, we analyze latency behavior under various loads. Despite latency profiles that suggest standard queuing delays, we find that most latency actually originates from negative interactions between the application and the locking and blocking mechanisms in the kernel. Modifying the server and kernel to avoid these problems yields both qualitative and quantitative changes in the latency profiles - latency drops by more than an order of magnitude, and the effective service discipline also improves. We find our modifications also mitigate service burstiness in the application, reducing the event queue lengths dramatically and eliminating any benefit from application-level connection scheduling. We identify one remaining source of unfairness, related to competition in the networking stack. We show that adjusting the TCP congestion window size addresses this problem, reducing latency by an additional factor of three.
暂无评论