Content delivery networks (CDNs) deploy globally distributed systems of caches in a large number autonomous (ASes). It is important for CDN operator to satisfy the performance requirements end users, while minimizing cache deployment cost. In this paper, we study optimization (CaDeOp) problem determining how much server, energy, and bandwidth resources provision each AS, i.e., AS chosen deployment. The CaDeOp objective minimize total cost incurred by CDN, subject meeting end-user...

The trend toward more peering between networks is commonly conflated with the of Internet flattening, i.e., reduction in number intermediary organizations on paths. Indeed, direct interconnections bypass layer-3 transit providers and make flatter. This paper studies an emerging phenomenon that separates two trends: we present first systematic study remote peering, interconnection where peer via a layer-2 provider. Our measurements reveal significant presence at IXPs (Internet eXchange...

Content delivery networks (CDNs) distribute much of the Internet content by caching and serving objects requested users. A major goal a CDN is to maximize hit rates its caches, thereby enabling faster downloads involves two components: an admission algorithm decide whether cache object eviction determine which evict from when it full. In this paper, we focus on propose novel called RL-Cache that uses model-free reinforcement learning (RL) or not admit into CDN's cache. Unlike prior...

Output jitter-the variation in the intercompletion times of successive jobs same task is studied context preemptive uniprocessor scheduling periodic real-time tasks. A formal quantitative model for output jitter proposed. list properties that are desirable any jitter-minimization schedule enumerated. Algorithms presented generating such schedules, and bounds proved maximum schedules generated by these algorithms.

A majority of ISPs (Internet Service Providers) support connectivity to the entire Internet by transiting their traffic via other providers. Although transit prices per Mbps decline steadily, overall costs these remain high or even increase, due growth. The discontent with has yielded notable innovations such as peering, content distribution networks, multicast, and peer-to-peer localization. While above solutions tackle problem reducing traffic, this paper explores a novel approach that...

We revisit the question of how much buffer an IP router should allocate for its droptail FIFO link. For a long time, setting size to bitrate-delay product has been regarded as reasonable. Recent studies interaction between queueing at routers and TCP congestion control offered alternative guidelines. First, we explore reconcile contradictions existing rules. Then, argue that problem link sizing needs new formulation: design algorithm accommodates all Internet applications without engaging in...

A majority of Internet service providers (ISPs) support connectivity to the entire by transiting their traffic via other providers. Although transit prices per megabit second (Mbps) decline steadily, overall costs these ISPs remain high or even increase due growth. The discontent with has yielded notable innovations such as peering, content distribution networks, multicast, and peer-to-peer localization. While above solutions tackle problem reducing traffic, this paper explores a novel...

Video streaming continues to captivate attention of users and service providers, dominate in Internet traffic, form a vibrant research field. Taking pragmatic approach reviewing recent the field, this paper considers most dominant paradigm, main aspects which include transmission two-dimensional videos over best-effort Internet, support from content delivery networks, client-side bitrate adaptation. To make survey more accessible, we incorporate extensive tutorial materials. In contrast with...

The Open Network Laboratory (ONL) is an Internet-accessible virtual laboratory facility that can deliver a high quality experience in advanced networking [3,7]. Our with ONL indicates it has potential to improve student understanding of fundamental concepts and increase enthusiasm for experimentation complex technology. Furthermore, these benefits be delivered less effort from the instructor than using traditional approach socket programming ns-2 simulation exercises. system built around set...

With the Internet offering a single best-effort service, there have been numerous proposals of diversified network services that align better with divergent needs different distributed applications. The failure these innovative architectures to gain wide deployment is primarily due economic and legacy issues, rather than technical shortcomings. We propose new paradigm for service differentiation where design principles account explicitly multiplicity providers users as well their interests...

Economic forces behind the Internet evolution have diversified types of ISP (Internet Service Provider) interconnections. In particular, settlement-free peering and paid proved themselves as effective means for reducing costs. this paper, we propose T4P (Transit Peering), a new type hybrid bilateral relationships that continues trend towards more flexible interconnections at lower With interconnection, one compensates other their by providing with partial-transit service. comparison to...

Content delivery networks (CDNs) distribute much of the Internet content by caching and serving objects requested users. A major goal a CDN is to maximize hit rates its caches, thereby enabling faster downloads involves two components: an admission algorithm decide whether cache object eviction which evict from when it full. In this paper, we focus on propose novel called RL-Cache that uses model-free reinforcement learning (RL) or not admit into CDN's cache. Unlike prior approaches use...

Unrelenting traffic growth, device heterogeneity, and load unevenness create scalability challenges for monitoring. In this paper, we propose Robust Distributed Computation (RoDiC), a new approach that addresses these by shifting portion of the monitoring-task execution from an overloaded network element to another has spare resources. Moving entire task away might be infeasible because on multiple elements is inherent in or requires at least partial participation designated element....

Layered multicast is a common approach for dissemination of audio and video in heterogeneous network environments. schemes can be classified into two categories - feedback-based feedback-free depending on whether or not the scheme delivers feedback to sender session. Advocates claim that necessary match receiver capabilities efficiently. Supporters believe introduces significant complexity moderate amount additional layers balance any benefit provides. Surprisingly, there has been no...

Instantaneous fair sharing (IFS) is a traditional network ideal prescribing to share the capacity among competing applications fairly during any infinitesimal time interval. In this paper, we argue that IFS an inappropriate for application of massive data transfers where primary goal minimize message transfer times. We propose alternative paradigm virtual finish first (ViFi) scheduling dedicates entire one at in order times under IFS. Unlike shortest remaining and other earlier algorithms...

With the Internet offering a single best-effort service, there have been numerous proposals of diversified network services that align better with divergent needs different distributed applications. The failure these innovative architectures to gain wide deployment is primarily due economic and legacy issues, rather than technical shortcomings. We propose new paradigm for service differentiation where design principles account explicitly multiplicity providers users as well their interests...

Quality of experience (QoE) becomes both the holy grail and a free-for-all in adaptive bitrate (ABR) video streaming. On one hand, design, operation, evaluation ABR algorithms increasingly rely on QoE. other QoE frequently receives only cursory attention this supporting role, with many its important aspects treated insufficient care. As complex subjective notion, is directly measurable through tests, which incur evident overhead. While an objective model represents scalable automated means...

Whereas prefix hijacking is usually examined from security perspectives, this paper looks at it a novel economic angle. Our study stems an observation that transit AS (Autonomous System) has financial interest in attracting extra traffic to the links with its customers. Based on real data about actual incident Internet, we conduct simulations AS-level Internet topology synthetic demands for hijacked traffic. Then, measure all inter-AS and compute payments of providers. The analysis our...

The single best-effort service of the Internet struggles to accommodate divergent needs different distributed applications. Numerous alternative network architectures have been proposed offer diversified services. These innovative solutions failed gain wide deployment primarily due economic and legacy issues rather than technical shortcomings. Our paper presents a new simple paradigm for differentiation that accounts explicitly multiplicity providers users as well their interests in...

To disseminate data to a heterogeneous body of receivers efficiently,congestion control protocols for IP multicast compose session from several groups and prescribe guidelines that enable each receiver subscribe an appropriate subset the groups. However, misbehaving can ignore group subscription rules inflate its acquire unfairly high throughput. In this paper, we present first solution problem inflated subscription. Our design guards access with dynamic keys consists two independent...

Virtual laboratories are a potential replacement for standard laboratory facilities. Use of these virtual resources can reduce cost and maintenance overheads teaching institutions while still ensuring that students have access to real equipment. Previous work indicates respond well such environments, but one important operational aspect has been overlooked. In this work, we consider instructor overhead by comparing the amount required teach courses with without use laboratory. particular,...