On the Analysis of Caches with Pending Interest Tables Mostafa - PowerPoint PPT Presentation

On the Analysis of Caches with Pending Interest Tables Mostafa Dehghan 1 , Bo Jiang 1 Ali Dabirmoghaddam 2 , Don Towsley 1 1 University of Massachusetts Amherst 2 University of California Santa Cruz ICN, October 2, 2015 College of Information and Computer Sciences

Overview Named Data Networking Data names instead of IP addresses Data Structures: Content Store Pending Interest Table Forwarding Information Base Packets: Interest packet Data packet 1 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Interest: /icn/pit.pdf Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 Interest: /icn/pit.pdf 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 Interest: /icn/pit.pdf 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 Interest: /icn/pit.pdf 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0,2 Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 Interest: /icn/pit.pdf 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0,2 Data: /icn/pit.pdf Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ss Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0,2 Data: /icn/pit.pdf Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ... ss /icn/pit.pdf Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0,2 Data: /icn/pit.pdf Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ... ss /icn/pit.pdf Pending Interest Table Name Face Face 1 ss /icn/pit.pdf 0,2 Data: /icn/pit.pdf Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Overview Content Store Name Data Face 0 ... ss /icn/pit.pdf Pending Interest Table Name Face Face 1 ss Forwarding Information Base Face 2 Prefix Face ss /icn/ 1 2 / 25

Pending Interest Table (PIT) Performs request aggregation Keep track of Interests forwarded but not yet satisfied 3 / 25

Pending Interest Table (PIT) Performs request aggregation Keep track of Interests forwarded but not yet satisfied Advantages: lower bandwidth usage communication without knowledge of source and destination better security . . . 3 / 25

Pending Interest Table (PIT) Performs request aggregation Keep track of Interests forwarded but not yet satisfied Advantages: lower bandwidth usage communication without knowledge of source and destination better security . . . Affects cache behavior hit probability response time 3 / 25

Motivation Need analytical models to predict cache behavior Better design choices 4 / 25

Motivation Need analytical models to predict cache behavior Better design choices PIT sizing 4 / 25

Motivation Need analytical models to predict cache behavior Better design choices PIT sizing State of the art: Ignores download delay Does not consider PIT 4 / 25

Motivation Need analytical models to predict cache behavior Better design choices PIT sizing State of the art: Ignores download delay Does not consider PIT Goal: Analytically model cache with PIT 4 / 25

Outline Time-To-Live Caches Model and Analysis Replacement Based Policies Simulation Results 5 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache Reset TTL cache reset timer upon each hit 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache Reset TTL cache reset timer upon each hit T t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache Reset TTL cache reset timer upon each hit T t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache Reset TTL cache reset timer upon each hit T t Cache 6 / 25

Time-To-Live (TTL) Caches Content eviction occurs upon timer expiration T t Cache Reset TTL cache reset timer upon each hit T t Cache 6 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Model Requests: renewal arrivals Reset TTL cache with PIT t CS PIT D T 7 / 25

Metrics Z t D L # misses M # hits N 8 / 25

Metrics Z Cache hit prob E [ N ] h = E [ M ] + E [ N ] t D L # misses M # hits N 8 / 25

Metrics Z Cache hit prob E [ N ] h = E [ M ] + E [ N ] t D L Cache response time # misses M # hits N r = E [ w ( D )] / ( E [ M ]+ E [ N ]) � t w ( t ) = t + ( t − x ) d m ( x ) 0 m ( t ) : renewal function Poisson: m ( t ) = λ t 8 / 25

Metrics Z Cache hit prob E [ N ] h = E [ M ] + E [ N ] t D L Cache response time # misses M # hits N r = E [ w ( D )] / ( E [ M ]+ E [ N ]) Cache occupancy prob � t w ( t ) = t + ( t − x ) d m ( x ) o = E [ L ] / E [ Z ] 0 m ( t ) : renewal function Poisson: m ( t ) = λ t 8 / 25

Metrics Z Cache hit prob E [ N ] h = E [ M ] + E [ N ] t D L Cache response time # misses M # hits N r = E [ w ( D )] / ( E [ M ]+ E [ N ]) Cache occupancy prob � t w ( t ) = t + ( t − x ) d m ( x ) o = E [ L ] / E [ Z ] 0 m ( t ) : renewal function PIT occupancy prob Poisson: m ( t ) = λ t p = E [ D ] / E [ Z ] 8 / 25

Replacement Based Policies Characteristic Time Approximation Introduced by Che et al. for LRU policy w/ Poisson arrivals Extended to other policies, e.g. FIFO, Random, . . . Extended to general request arrival processes Characteristic time T solution of � i o i ( T ) = C o i – cache occupancy probability of file i C – cache size 9 / 25

LRU Cache with Poisson Arrivals Modeled as reset TTL with constant T Cache hit/occupancy probability e λ i T − 1 h i = o i = λ i E D i + e λ i T λ i – arrival rate of requests for file i E D i – expected download delay for file i characteristic time T solution of e λ i T − 1 � λ i E D i + e λ i T = C i 10 / 25