1 / 10

Evaluation of Fairness in ICN

Evaluation of Fairness in ICN. X. De Foy, JC. Zuniga, B. Balazinski InterDigital (first.last@interdigital.com). Motivation.

dex
Download Presentation

Evaluation of Fairness in ICN

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Evaluation of Fairness in ICN X. De Foy, JC. Zuniga, B. Balazinski InterDigital (first.last@interdigital.com)

  2. Motivation As part of the ongoing discussion on the Evaluation Methodology draft, should we measure how “fair” is the usage of network resources in ICN? Is this a real issue? If so, what should we measure?

  3. Potential Fairness Issues in ICN • Since ICNs share resources… • Network resources: bandwidth, in-network storage, NRS, Topology Managers, etc. • Individual nodes resources: CPU, RAM, NIC, Storage, etc. • … a network entity can trick the system to get a larger amount of resources • Receivers, publishers, even routers • … an ICN system design may have a bias • E.g. a too large a bias towards popular content may starve unpopular content objects

  4. Fairness Measures in ICN • Fairness between end users • Throughputs between several competing flows on bottleneck links are compared for interest shaping evaluation in [4] • Fairness between classes of content (e.g. popular vs. unpopular) • Number of replicas depending on popularity in [2] • In these scenarios it is assumed that all users will get similar service for a same class of content: • Response time depending on popularity in [1] • Ratio of throughput gain with/without cache depending on popularity in [3] • … any other examples?

  5. Fairness in a Generic Context • Strategies/Criteria: between fairness and efficiency • Basic strategy is max-min fairness – the goal is to maximize the minimum allocation among all agents • Proportional fairness – compromise between efficiency and a minimal level of fairness between users • Other efficiency-fairness tradeoffs (e.g. α-fairness used in [9] ranging from utilitarian α=0, proportional α=1, max-min α→∞) • Examples of Qualities used to Evaluate Fairness • Share Guarantee: every arriving agent gets at least an equal split of the resources if it wants it • Strategy Proof-ness: no agent may misreport its demand and be better off regardless of demands from other agents • Envy Freeness: an agent would never prefer the allocation of another agent • Pareto Efficiency: changing allocation to make one agent better off would result in making another agent worse

  6. Fairness in Relevant Domains • Sharing of resources in data centers between tenants • Run jobs with resource usage vector (CPU, Memory, etc.), measure number of jobs per tenant • Example: Dominant Resource Fairness [5] • Middle box resource allocation • Multiple resources within a single node, more dynamic allocation (queue scheduling) [6] • Sharing of Resources in P2P Systems • Some peers can use strategies to optimize their gain and minimize their own resource usage [7] • Scheduling in Wireless Ad Hoc networks • Single resource (bandwidth/time slots) allocated at each node in a distributed manner [8]

  7. Evaluation of Fairness in ICN 1/2 • ICN should be fair to end users/receivers. Should it be fair to other entities like publishers and individual content objects as well? • Some resources may be “traded” for others, especially Bandwidth vs. Storage • Multiple content sources/replicas can lead to a distribution of bottlenecks (e.g. network links, routers storage, routers CPU, etc.)

  8. Evaluation of Fairness in ICN 2/2 • Is there an acceptable set of reference setups… • nr receivers, np publishers, no content objects • A network topology with caches/routers/links and placement of receivers/publishers • Some of the topologies used in evaluations are bus, ring, caching tree, topologies with a single congested link, mesh with certain probability of inter-connections. • A usage pattern • … and associated fairness criteria? • General criteria, qualities • Specific agents (receivers, etc.)

  9. Questions to the RG • Is there a common, agreeable definition of fairness for ICN? • Is there a content-oriented aspect of fairness? • Do ICN particularities actually make a difference on how to evaluate fairness (multiple sources, caching vs. bandwidth equivalence)? • Should we include a section on this aspect in the Evaluation Methodology draft?

  10. References [1] Carofiglio et al. “Bandwidth and Storage Sharing Performance in Information Centric Networking”, 2010, Link [2] Tortelli et al. “A Fairness Analysis of Content Centric Networks”, 2012, Link [3] Saucez et al. “Congestion control and in-network caching” (ICNRG IETF84 presentation), 2012, Link [4] Wang et al. “An Improved Hop-by-hop Interest Shaper for Congestion Control in Named Data Networking”, 2013, Link [5] Ghodsi et al. “Dominant Resource Fairness: Fair Allocation of Heterogeneous Resources in Datacenters”, 2010, Link [6] Ghodsi et al. “Multi-resource fair queueing for packet processing”, 2012, Link [7] Piatek et al. “Do incentives build robustness in BitTorrent?”, 2007, Link [8] Tassiulas & Sarkar“Maxmin Fair Scheduling in Wireless Ad Hoc Networks”, 2005, Link [10] Bertsimas et al., “A Characterization of the Efficiency-Fairness Tradeoff”, 2010, Link

More Related