Opportunistic Off-Path Content Discovery in Information-Centric Networks Notes Onur Ascigil, Vasilis Sourlas, Ioannis Psaras, and George Pavlou Department of Electronic and Electrical Engineering, University College London, UK

Outline Content discovery in Information-centric Networks (ICN) Opportunistic Content Discovery Evaluation Future Work & Conclusions

Content Discovery in ICN

Content Discovery in ICN: ICN Features Information-centric Networking (ICN) features Naming of content e.g., /ucl/ee/onur/lanman_presentation2016.ppt Network Storage Ubiquitous caching Stateful Forwarding Request (i.e, Interest) leaves breadcrumbs, which the data follows Routing on names: Mostly focused on locating the content on-path, at origin or a designated cache Disclaimer: ignoring scalability issues in this talk Goal: integrate content delivery as a native network feature Is it achieved? Content can be transparently delivered from any cache enabled node or router. Routing and forwarding are based on content names.

Content Discovery in ICN Find a nearby (ideally nearest!) copy of the content Difficult to achieve without significant ``overhead’’ in practice Why? Volatile nature of content in the caches: Contents of caches may change at very short time-scales Request-to-cache routing Mostly focused on routing requests to the content origin Search content on-path (i.e., along the shortest or default path) Existing Solutions for Content Discovery: Opportunistic on-path Coordinated off-path

Content Discovery in ICN: Request Routing Opportunistic on-path: limited gain Caching: Content is cached on-path as it travels to the user from the content origin Request routing: Route requests to content origin & retrieve content opportunistically from on-path caches E.g., Barebones NDN with default forwarding strategy Coordinated off-path: coordination and communication overhead Caching: content is assigned to off-path caches according to predefined rules Request routing: adheres to the same rules in order to retrieve contents from caches in a coordinated manner E.g., Hash routing: a hash function determines both the placement of content and routing of requests by mapping content identifiers to cache nodes E.g., Coordinate content placement and routing with the control plane: advertise cache contents in the control plane and direct requests to caches

Opportunistic Content Discovery

Opportunistic Content Discovery Stateful forwarding of data packets: data packets leave breadcrumbs Downstream FIB table (D-FIB) FIB Prefix Next-hop /facebook T D-FIB D-FIB R Name Next-hop /…./x.mpg S Name Next-hop /…./x.mpg T Data: 10010101… H2 T U Request: /facebook/user/x.mpg Request: /facebook/user/x.mpg Request: /facebook/user/x.mpg Upstream & downstream. Store state to keep track of the direction (i.e., next hop) in which the Data packets were sent in the past. Multiple copies of the data … Request: /facebook/user/x.mpg Request: /facebook/user/x.mpg Request: /facebook/user/x.mpg FIB S Prefix Next-hop /facebook U FIB Prefix Next-hop /facebook T H1

Opportunistic Content Discovery: Downstream FIB Table Caches trails of data packets towards users LRU policy An exact match table

Opportunistic Content Discovery: Routing using D-FIB & FIB Goal: improve cache hits Reduce latency in obtaining content limit overhead and reduce the number of requests reaching the content origin Each request is associated with a Total Forwarding Counter (TFC) value spend it on sending a copy of a request downstream spend it on following the FIB table towards the content origin (upstream) spend it on both (multicast) TFC is initially set by the access router New Forwarding Strategies based on D-FIB Determines how TFC quota is spent at each router

Opportunistic Content Discovery: Downstream FIB Table Multicast requests using FIB and D-FIB Z Q L Name Next-hop /x/y/z Q, Y FIB Prefix Next-hop Distance /x S 4 Request: /x/y/z Off-path T R S Request: /x/y/z Quota = 4 Request: /x/y/z Off-path Once a request is forwarded downstream, it only follows the D-FIB tables. They don’t travel back towards the content origin, following the FIB table N Request: /x/y/z Quota = 4 + 3 Request: /x/y/z Quota = 3 Request: /x/y/z K Y M

Opportunistic Content Discovery: Forwarding Strategies Check Content Store, if no matching content, then: Lookup FIB and D-FIB If D-FIB returns no entries, follow FIB (forward upstream) If D-FIB returns one or more entries, then the forwarding strategy decides what action to perform Two simple strategies: ALL strategy: Send a copy of the request to all the next-hops in the D-FIB entry the cache is closer (number of hops) than the content origin ONE strategy: Send a copy of the request to only one next-hop in the D-FIB entry Freshest entry which is closer than the content origin

Performance Evaluation

Evaluation Implemented our approach in ndnSIM — an ns-3 based simulator Performance metrics: Cache hit ratio: percentage of the interests that have been satisfied Off-path/on-path The minimum hop distance: number of hops traveled by the (first) data arriving at the user from a responding router or the content origin for each successful request The mean traffic overhead: the mean number of hops that the initiated Data packets travel in the network Variables: Cache size at each node D-FIB size w.r.t. content population size Initial Quota

Evaluation: Scenario Using a RocketFuel topology: AS 4755 VSNL (India) 191 nodes: 148 edge, 39 gateway, and 4 backbone routers 242 bi-directional links Request rate: 100 requests/sec Randomly select an edge router Content Population: 10,000 One chunk per item One content server attached to a randomly chosen edge router our results comparing performance of on-path/off-path is best-case scenario Popularity of the items determined by a Zipf law of exponents Zipf parameter z: 0.7 Quota: Shortest path length + 3 Duration: 1 hour (following an hour of warm-up phase)

Evaluation: Impact of Router’s Cache Size Impact of D-FIB size w.r.t. content population on the performance

Evaluation: Impact of Router’s Cache Size Hop distance is slightly better with ALL strategy

Evaluation: Impact of Router’s Cache Size Overhead difference is negligible between ALL and ONE

Evaluation: Impact of D-FIB size

Evaluation: Impact of Initial Quota

Evaluation: Impact of Initial Quota What percentage of the first requests manage to fetch the content?

Future Work Augment the strategies with additional information Thresholds for freshness of state in the D-FIB On-path (i.e., upstream) hints to steer requests along a particular (fresh) downstream trail along the path to the content origin. L T S Z Data: /x/y/z 10010101… Name Next-hop Age Distance x/y/z K T sec. 2 N Y K M

Conclusions Opportunistic scheme to discover content using two simple yet effective strategies Reduces load on the content origin Improves cache hits Adds little overhead Basic framework with opportunities to augment with more sophisticated forwarding strategies

Thank you for listening! Questions?

Backup slides

Zipf Distribution Requests are generated in the network with rate r = {r1,...,rM}, where rm denotes the aggregate incoming request rate (in requests per second) for content item m 2 M. The request rate for each item is determined by its popularity. Here we approximate the popularity of the items by a Zipf law of exponents z [20]. In that way the aggregate incoming request rate (in requests per second) for an information item m 2 M is given by: rm=z· 1/kz ,