Popular in Course
verified elite notetaker
Popular in Computer Information Technology
verified elite notetaker
This 7 page Class Notes was uploaded by Kathleen Cartwright on Monday September 28, 2015. The Class Notes belongs to CIS700 at University of Pennsylvania taught by Staff in Fall. Since its upload, it has received 22 views. For similar materials see /class/215375/cis700-university-of-pennsylvania in Computer Information Technology at University of Pennsylvania.
Reviews for CIS
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 09/28/15
i C18 700 005 Networking Meets Databases Boon Thau Loo Spring 2007 Lecture 1 1 Visemtllantaxedanslldsiiam abe mmis Sean hea minimum i Motivations pomt communication abstraction 2 Send packet p 39om host A to host B s This abstraction allows Internet to be highly scala le and efficien ut not appropriate for applications that require other communications primitives 2 Multicast 2 Anycast 2 Mobility Todays Internet is built around a unicast pointto l way 2 Pointtopoint communication implicitly assumes there is one sender and one receiver and that they are placed at fixed and well known locations 3 Eg a host identified bythe IP address 12832xxxxxx is located in Berkeley i lP Solutions Extend IP to support new communication primitives e 2 Mobile IP 2 IPmulticast 2 IP anycast o Disadvantages 2 Dil cult to implement while maintaining Intemet s scalability eg multicast 2 Require community wide consensus hard to ad1ieve in practice Application Level Solutions 9 Implement the required functionality at the application level eg 2 Application level multicast e g Narada Overcast Scattercast Application level mobility s Disadvantages 2 El ciency hard to ad1ieve 2 Redundancy each application implements the same functionality over and over again 2 No synergy each applica ion implements usually only one service services hard to combine i Key Observation vi Virtually all previous proposals use indirection eg 2 Physical indirection point mobile IP 3 Logical indirection point gt IP multicast Any problem in computer science can be solved by adding a layer of indirection l 3 s Solution Build an efficient indirection layer on top of IP Use an overlay network to implement this layer 2 Incrementally deployable don t need to change IP Application Internet lndirection lnfrastmcmre 3 a Each packet is associated an identifierid a To receive a packet with identifier id receiver R maintains a trigger id R into the overlay network f 4 D trigger ma J l Service Model API 2 sendPacketp 2 insertTriggert 1 removeTriggert optional Besteffort service model like IP Triggers periodically refreshed by endhosts ID length 256 bits l Mobility Hostjust needs to update its trigger as it moves from one subnet to anothe Receiver R2 l Multicast e Receivers insert triggers with same identifier 1 Can dynamically switch between multicastand unicast Sender Receiver R2 Auycast c Use longest pre x matching instead of exact matching 2 Pre x p anycast group identi er 2 Suf x s encode application semantics eg location ReceiverR3 Service Composition Sender Initiated Use a stack of IDs to encode sequence of operations to be performed on data path Advantages 2 Don t need to con gure path 2 Load balancing and robustness easy to achieve quotTranseode39F39m Service Composition Receiver Initiated e Receiver can also specify the operations to be performed on data E Sender Sender Outline Quick implementation Overview gt Implementation 1 Examples 9 Security 1 Applications n is is implemented on top of Chord 1 But can easily use CAN Pastry Tapestry etc Each trigger t id R is stored on the node responsible for id 2 Use Chord recursive routing to find best matching trigger for packetp id data Routing Example eds to WOW onlyone l3 node to use vvs nude 3 R knuvvs nude 35 s gV xxi Esewl ml 820 a R inserts tnggerr37 R 8 sends packetp37 1m A 3 2m n e cnurd ElrElE send1Rdmz l Optimization 1 Path Length enderreceiver caches i3 node mapping a speci c ID ubsequent packets are sent via one i3 no e as as Optimization 2 T angular Routing Use wellknown trigger for initial rendezvous 3 Exchange a pair of private triggers welllocated a Use private triggers to send data traf c Sender8 l Outline Implementation gt Examples i Heterogeneous multicast 1 Scalable Multicast 3 Load balancing 3 Proximity m Security Applications Example 1 Heterogeneous Multicast Sender not aware of transformations i y Receivei m i JPEG Receivei R2 Mlth l Example 2 Scalable Multicast i3 doesn t provide direct support for scalable multicast Triggers With same ldel itl el al e mapped ehto the same is hode Solution have endhosts build an hierarchy oftrigger of bounded degree l Example 2 Scalable Multicast Unlike IP multicast i3 1 quot I I remain simple robust and scalable 2 Gives endh ts control on routing enable endhosts to 3 Achieve scalability ah timize tree echstmctiOh to match thelrneed e g delay Width 3 l Example 3 Load Balancing 7 Servers insert triggers with IDs that have random suf xes A Clients send packets with IDs that have random sul xes Example 4 Proximity Suffixes of trigger and packet IDs encode the server and client locations 53E I Outline 5i Implementation Examples gt Security Applications Some Attacks Eavesdropping Loop Alta ckel DeadEn d Attackec mm an i More attacks 1 Malicious linking 1 Attacker can sign an endhost R to a high bandwidth traffic stream sent to id by inserting a trigger idR n Impersonation 1 Same as eavesdropping i Defences Encryption 2 A send to B 2 Encrypt private trigger ida using public key of B vice versa r Challenges 2 Nancebased challenges to prevent unjusti ed inser ion of triggers by hird parties 2 Send random hence to endhost Remove if endhost does not respon r Pushback 2 When there is no more matching trigger for packet ID i3 sends a pushback message to previous node v Timeto live TTL i Constrained Triggers h h wellknown oneway hash functions gt Use ho ho to constrain trigger x y must match nght constrained k D key 64 128 64 Left constrained X key him A gtlt key hy key endrhost address l Summary Defenses Egg Pushback hallenges Attacks Constraints Em esdmppmg lmpersonanou Loops Con uence t Dead ends Nod Cannueuee Long Chants Malicious Lmime Malicious TngerRemnvc l Outline r Implementation 3 Examples 2 Security 4 Architecture Optimizations gt Applications 3 Routing as a service 3 Service composition platform 3 Support of legacy applications over overlays l Routing as a Service Goal develop network architectures that 3 Allow endhosts to pick their own routes 3 Allow thirdparties to easily add new routing protocols 2 Ideal model 3 Oracles that have complete knowledge about network 3 Hosts query paths from oracles 1 Path query can replace today s DNS query l Routing as a Service cont d Cliem A H Network measurem ems 3 Queryreply routing info Setup routes Routing i K E Routing l service 2 3 Hosts forward packets along these paths cum 3 Client D l Routing as a Service l Design Principles 1 Give hosts control on routing u Atrigger is like an entry in a routing table Host Infrastructure 3 Flexibility customization u Endhosts can Source route Setup acyclic communication graphs Route packets through desired service points Stop flows in infrastructure a 71 l 2 Implement data forwarding in infrastructure 1 Ef ciency scalability Internet amp Infrastructure overlays p2p amp Endhost overlays l Outhe s Implementation u Examples 1 Security u Architecture Optimizations Applications 3 Routing as a service gt Service composition platform 1 Support of legacy applications over overlays Service Composition Platform a Goal allow thirdparties and endhosts to easily insert new functionality on data path 1 Eg firewalls NATs caching transcoding spam filtering intrusion detection etc 2 Why i3 1 Make middleboxes part ofthe architecture 1 Allow endhoststhirdparties to explicitly route through middleboxes l Example Use Bro system to provide intrusion detection for endhosts that desire so a Spam filtering etc Bro dlerbox i v ldM ldAE data idiot i3 l Outhe Implementation Examples m Security 1 Architecture Optimizations gt Applications 1 Routing as a service 3 Service composition platform i Support of legacy applications over overlays a See httpocalacsberkeley edu l Conclusions lndirection keytechnique to implement basic communication abstractions 2 Multicast Anycast Mobility 3 httpi3cs berkeleyedu Reminder Project proposal was due yesterday Volunteers for second presentations