* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Lect 5 - ICN
Survey
Document related concepts
Cracking of wireless networks wikipedia , lookup
Distributed firewall wikipedia , lookup
Deep packet inspection wikipedia , lookup
Computer network wikipedia , lookup
Network tap wikipedia , lookup
Net neutrality law wikipedia , lookup
Video on demand wikipedia , lookup
Recursive InterNetwork Architecture (RINA) wikipedia , lookup
TV Everywhere wikipedia , lookup
Airborne Networking wikipedia , lookup
List of wireless community networks by region wikipedia , lookup
Transcript
Internet of Things Amr El Mougy Alaa Gohar Information-Centric Networks Mobile Data Traffic Video Data Traffic Dealing with the Growth in Internet Traffic • Global IP traffic is expected to exceed 1.4 Zettabytes by the end of 2017 • Over 80% of this traffic will be video • Solution: shift the load away from the traditional client-to-server connection pattern • Proposed solutions: Caching (web caching and content delivery networks) and peer-to-peer networks • The idea is to exploit storage and processing capabilities readily existing in the Internet to speed up data transfer and reduce latency • Both ideas work at the application layer, where latency is already high Content Distribution over the Internet DOES NOT SCALE Tier 1 Networks ISPs IRTF Open Meeting @ IETF-81 Why Not? • URLs and IP addresses are overloaded with locator and identifier functionality – Moving information = changing it‘s name => 404 file not found • No consistent way to keep track of identical copies – No consistent representation of information (copy-independent) • Information dissemination is inefficient – Can‘t benefit from existing copies (e.g. local copy on client) – No “anycast”: e.g., get “nearest” copy – Problems like Flash-Crowd effect, Denial of Service, … Why Not? • Can’t trust a copy received from an untrusted node – Security is host-centric – Mainly based on securing channels (encryption) and trusting servers (authentication) • Application and content provider independence – CDNs focus on web content distributions for major players – What about other applications and other players? Information-Centric Networks (ICN) Today’s Internet Focus on nodes In today’s Internet, accessing information is the dominating use case! Information Centric Network Focus on information objects Evolution Web CDN • Considering important requirements – Accessing named resources – not hosts – Scalable distribution through replication and caching – Good control of resolution/routing and access P2P • With ubiquitous caching – But for all applications – And for all users and content/service providers – Use off path caching and on path (in-network) caching Information-Centric Networks (ICN) • Apart from routing protocols that use a direct identifier of nodes, networking can take place based directly on content. • Content can be collected from the network, processed in the network, and stored in the network • Goal is to provide a network infrastructure capable of providing services better suited to today’s application requirements: – content distribution & mobility – more resilience to disruption and failures • We look next at such content-based networking and data aggregation mechanisms. Networking Evolution • Traditional networking – Host-centric communications addressing end-points • Information-centric networking – Data-centric communications addressing information (e.g., data in context). – Decoupling in space – neither sender nor receiver need to know their partner. – Decoupling in time – “answer” not necessarily directly triggered by “question”, asynchronous communication. Example: Content Distribution Example: Content Distribution Example: Content Distribution Example: Content Distribution Example • Content goes only where there’s interest. • It takes at most one trip across any link. • Average latency is minimized. • Total bandwidth is minimized. • There’s no routing or control traffic associated with the replicas. How is it Done? • Approach – Named Data Objects (NDOs) – in-network caching – multi-party communication through replication – decoupled senders from receivers • Architectural questions – How do we address information? – How do we obtain information? – How do we route information? ICN Communication Model Dissemination networking • Data is requested by name, using any and all means available (IP, VPN tunnels, multicast, proxies, etc). • Anything that hears the request and has a valid copy of the data can respond. • The returned data is signed, and optionally secured, so its integrity & association with name can be validated (data centric security) Digital Signature Content-Based Security • Name-content mapping verification via per-data packet signature – Data packet is authenticated with digital signature ICN trust establishment by associating content namespaces w/ public keys ICN Stack (1) Van Jacobson, et al, Networking Named Content, CoNEXT 2009 • • • • • Change of network abstraction from “named host” to “named content” Security built-in: secures content and not the hosts Mobility is present by design Can handle static as well as dynamic content Use of 2 messages: Interest and Data Objects Universal? • Any architecture that runs over anything is an overlay (IP is an overlay). • IP started as a phone system overlay; today much of the phone system is an IP overlay. System theorists would say ‘IP is universal’. • ICN has the same character: it can run over anything, including IP, and anything can run over ICN, including IP. Naming • Solution 1: Name the data – Flat, not human readable identifiers • 1DB76EB8DFD6B0b92A293AADC8421830BDE73CB6 – Hierarchical, meaningful structured names • /nytimes/sport/baseball/mets/game022414/ • Solution 2: Describe the data – With a set of tags • baseball, new york, mets – With a schema that defines attributes, values and relations among attributes Using Names in CCN • The hierarchical structure is used to do ‘longest match’ lookups (similar to IP prefix lookups) which helps guarantee log(n) state scaling for globally accessible data. • Although CCN names are longer than IP identifiers, their explicit structure allows lookups as efficient as IP’s. Basic ICN forwarding • Consumer ‘broadcasts’ an ‘interest’ over any & all available communications media: get ‘/rutgers/ECE544/Lecture06-14.pdf’ • Interest identifies a collection of data - all data items whose name has the interest as a prefix. • Anything that hears the interest and has an element of the collection can respond with that data: HereIs ‘/rutgers/ECE544/presentation.pdf/p1’ <data> Basic ICN transport • Data that matches an interest ‘consumes’ it. • Interest must be re-expressed to get new data. (Controlling the re-expression allows for traffic management and environmental adaptation.) • Multiple (distinct) interests in same collection may be expressed (similar to TCP window). Publish/Subscribe Communications • One of the most important communication patterns in ICN • Publishers are decoupled from subscribers in time and space • Publishers inform a special network node (for example a content broker) that they will generate a particular type of content • Subscribers inform the network of their interest in this content (once, periodic, event-based, etc.) • The network forwards content either from publisher or the nearest cache On-Path Caching **A. Ioannou and S. Weber, “A Survey of Caching Policies and Forwarding Mechanisms in Information-Centric Networking,” IEEE Communications Surveys and Tutorials, Vol. 18, No. 4, 2016. Caching • Storage for caching NDOs is an integral part of the ICN service. • All nodes potentially have caches; requests for NDOs can be satisfied by any node holding a copy in the cache. • ICN combines caching at the network edge as in P2P and other overlay networks with in-network caching (e.g., transparent web caches) Challenges • Caching decisions cannot be determined apriori as in CDNs, it is based on content may add delay • Limited cache capacity. Larger caches may introduce more delay due to extended lookup time • Deciding on a policy for prioritizing which content to be cached • Large diversity in content • How to implement cooperative caching to optimize storage • Deciding on the popularity of content • Correlation between content (people with geographical proximity order similar content) • Exploiting cached content. How to determine best forwarding paths from the closest cache to the subscriber Taxonomy of Caching Mechanisms Probabilistic Caching • Caching decisions are based on a probability p, which could be fixed or dynamic • Fixed probabilistic policies involves no cooperation and thus does not cause additional overhead • On the other hand, it does not consider network topology, correlation, or popularity of content • RND defines a fixed p for all nodes • Unicache assigns uniform p = (1/length of path) • Leave Copy Everywhere (LCE) assigns p = 1 at all nodes Probabilistic Caching • Dynamic policies take into account traffic characteristics, type of content, and network topology to determine p • ProbCache tries to store content at nodes that are farther away from the publisher by calculating Ttw = 2 seconds Nx = Nn = 10 seconds • Where x is the length of the path and y is the number of hops from the source to a node, Nn is the cache size of n, Nx is the average cache size of a path, and Ttw is the time for which content should be cached on a path Graph-Based Caching • Considers the network topology to decide where to cache • Edge caching leaves a copy at the last node on a path, where it is closest to the consumer • Leave copy down (LCD) leaves a copy one more hop towards the consumer each time the same content is received, thus capturing content popularity • However these policies are static and do not consider the position of different nodes on the path relative to consumers Graph-Based Caching • Centrality policies use graph metrics to determine the most possible node for caching 6 Graph centrality 1/ max 𝑑 𝑚, 𝑛 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑚 ≠ 𝑛 Closeness 𝐶𝑙𝑜𝑠𝐶 = 𝑖 Centrality 9 Degree Centrality deg(𝑖) 𝐷𝑒𝑔𝐶𝑖 = 𝑛−1 1 𝑖≠𝑗 𝑑𝑖𝑠 𝑖, 𝑗 4 7 3 5 10 2 Betweenness Centrality 𝐵𝑒𝑡𝑤𝐶𝑖 = 8 1 𝑠𝑝𝑗,𝑘 𝑖 𝑗≠𝑖 𝑠𝑝𝑗,𝑘 Label-Based Caching • The objective here is to increase content diversity in the network by allowing each node to cache only a range of content • The network topology and the content ranges must be defined apriori • Must be restricted in scale of an ISP. Otherwise content ranges are difficult to define Popularity-Based Caching • Tries to minimize the cache pollution problem, where one-timer objects are stored • One-timer object requests are estimated to lie between 45% and 75% of all requested content. Thus, the cache pollution problem may be significant • However, popularity of content also depends on location. Thus we have local-based policies and path-based policies • Popularity is decided based on a content counter • In path-based policies, the counter is initialized to “new” • It does not address the cache pollution problem effectively Popularity-Based Caching • Unlike path-based policies, local-based policies do not require a minimum of one node to cache any content • Local-based policies work by observing the popularity of content, and can be static or dynamic • Static local-based caching defines a threshold for the content counter to start caching any content • Calculating the threshold is challenging • This policy does not adapt the threshold if conditions change Popularity-Based Caching • Dynamic local-based popularity caching observes the popularity within a time interval ΔT • Implicit dynamic popularity defines • Explicit dynamic popularity arranges content in an ordered list of popularity as the fraction of one the popularity (counter) of one request to decreasing popularity and stores the first x objects the total number of requests ICN Approaches **G. Xylomenos, C. Ververidis, V. Siris, N. Fotiou, C. Tsilopoulos, X. Vasilakos, K. Katsaros, and G. Polyzos, ”A Survey of Information-Centric Networking Research,” IEEE Communications Surveys and Tutorials, Vol. 16, No. 2, 2014. Named Data Networking (NDN) – Naming: Hierarchical naming, single address – Security: Signed content – Routing: Longest prefix matching – Caching: Local or network based – Content existence knowledge: Not part of the CCN core – Producer-consumer meeting: Propagation of interests Scalable and Adaptive Internet Solutions (SAIL) – Naming: Flat-ish naming – Security: Confidentiality, integrity, authentication, authorization – Routing: Name resolution and data matching can be either coupled or decoupled – Caching: Local and network based – Content existence knowledge: Through name resolution service – Producer-consumer meeting: Name resolution service provide locations Publish/Subscribe Internet Technology (PURSUIT) – Naming: Multi-level identifiers – Security: Signed content – Routing: (1) Name resolution (2) Information transfer – Caching: Network based – Content existence knowledge: Registrations in Rendezvous system – Producer-consumer meeting: Rendezvous system provides location Data Oriented Network Architecture (DONA) – Naming: Flat naming – Security: Signed content – Routing: Queries are resolved to locations – Caching: Network based – Content existence knowledge: Through resolution infrastructure – Producer-consumer meeting: Resolution infrastructure provides location