Investigating the Recursive InterNetwork Architecture as the next generation GÉANT and NREN network architecture

Transcription

1 Investigating the Recursive InterNetwork Architecture as the next generation GÉANT and NREN network architecture Sander Vrijders, Dimitri Staessens, Didier Colle Ghent University - iminds GN3plus JRA1 and TERENA 3rd Network Architects Workshop 1

2 Challenges faced by network engineers Explosion in the complexity of the overall system (hundreds of protocols and thousands of standards documents) Weak security Scalability issues with the routing system (IPv6/BGP multihoming) Mobile end-users No QoS support High OPEX 2

3 The Internet is a live environment Ever growing customer base Ever growing number of devices New and more demanding services RAD of services Fast deployment whac-a-mole approach to solving problems 3

4 A brief introduction to the Recursive InterNetwork Architecture RINA 4

5 Application Application Specific Tasks Renderi ng Task schedul ing Game engine Memory mgt Mgt. Tasks RIB Components Application specific tasks Management tasks Mechanism Static, invariant parts Policy Dynamic, variant parts Occurs in pairs Sender Receiver 5

6 Distributed Application X X X X X Y X Y ing system: hardware and software capable of executing programs as Application es that can coordinate via shared memory ( test and set ) Computing system: a collection of processing systems under the same management domain with no restrictions on connectivity 6

7 Application Application Specific Tasks Renderi ng Task schedul ing Game engine Memory mgt Multiplexing SDU Protection Mgt. Tasks RIB Mgt. Tasks Resource Manager DIF Allocator Components Application specific tasks Management tasks Management tasks DIF Allocator: Finds remote application processes IRM: manages DA requests Multiplexing: SDUs from different tasks SDU protection: Integrity and security 7

8 Distributed Applications host Edge router Internal AS router Edge router host X Common Distributed Application Protocol Y 8

9 Common Distributed Application Protocol (CDAP) Perform operations on RIB objects Create/Delete Read/Write Start/Stop But what about different applications? The objects they manipulate Control and sequencing of operations 10

10 Distributed Applications Provide services host Edge router Internal AS router Edge router host X Y F3 F1 F2 F4 C2 C1 D2 D1 D3 E1 E2 A1 A2 B1 B2 11

11 Effectively extending the model 12

12 API APs communicate using a port, identified by a portid 6 operations: int _registerapp(appname, List<difName>) portid _allocateflow(destappname, List<QoSParams>) int _write(portid, sdu) sdu _read(portid) int _deallocate(portid) int _unregisterapp(appname, List<difName>) QoSParams are defined in a technology-agnostic way Bandwidth-related, delay, jitter, in-order-delivery, loss rates, 13

13 The process Application Specific Tasks Other Mgt. Tasks Mgt. Tasks Multipl exing SDU Protec tion Resource Mgt. DIF Allocator Appl. Authentication of all processes RIB Daemon manages state objects EFCP protocol performs SDU transport API Data Transfer Data Transfer Control Layer Management SDU Delimiting Data Data Transfer Transfer Data Transfer Relaying and Multiplexing SDU Protection State Vector State Vector State Vector Transmission Transmission Control Control Control Retransmission Retransmission Control Control Control Flow Control Flow Control Flow Control RIB Daemon RIB CACEP Authentication CDAP Parser/Generator Enrollment Flow Allocation Resource Allocation Forwarding Table Generator 14

14 Error and Flow Control Protocol DTP Fragmentation Reassembly Sequencing Concatenation Separation DTCP Retransmission control Flow control Transmission control Loosely coupled by a state vector Based on Delta-t 15

15 Delta-t (Watson, 1981) Developed at L.Livermore labs, unique approach. Assumes all connections exist all the time. keep caches of state on ones with recent activity Watson proves that the conditions for distributed synchronization are met if and only if 3 timers are bounded: Maximum Packet Lifetime: MPL Maximum number of Retries: R Maximum time before Ack: A That no explicit state synchronization, i.e. hard state, is necessary. SYNs, FINs are unnecessary 1981:Watson shows that TCP has all three timers and more. 16

16 RMT 17

17 Shims Wrap a technology with the API Physical medium Legacy technology Ethernet IP Hypervisors Not required to add functionality So it s an incomplete DIF 18

18 Basic concept of RINA Theory Applications TCP/UDP (L4) IP (L3) Ethernet (L2) Physical Media (L1) Everyday practice Applications UDP (L4) IP (L3) VXLAN(L2) UDP (L4) IP (L3) IP (L3) IEEE (L2) MPLS (L2.5) IEEE 802.1Q (L2) IEEE 802.1ah (L2) 10GBASE-ER (L1) RINA Applications DIF DIF DIF DIF Physical Media 19

19 Bootstrapping a RINA network host Edge router Internal AS router Edge router host X Y F3 F1 F2 F4 C2 C1 D2 D1 D3 E1 E2 A1 A2 B1 B2 20

20 Architectural Model Application Specific Tasks Other Mgt. Tasks Appl. System (Host) System (Router) Mgmt Agemt Appl. System (Host) Mgt. Tasks Multipl exing SDU Protec tion Resource Mgt. DIF Allocator Mgmt Agemt Shim Shim DIF over TCP/UDP Shim Shim DIF Shim DIF over Ethernet Shim Mgmt Agemt API Data Transfer Data Transfer Control Layer Management SDU Delimiting Data Data Transfer Transfer Data Transfer Relaying and Multiplexing SDU Protection State Vector State Vector State Vector Transmission Transmission Control Control Control Retransmission Retransmission Control Control Control Flow Control Flow Control Flow Control RIB Daemon RIB CACEP Authentication CDAP Parser/Generator Enrollment Flow Allocation Resource Allocation Forwarding Table Generator Increasing timescale (functions performed less often) and complexity 21

21 IRATI PROTOTYPE 22

22 IRATI OS/Linux implementation Source: S. Vrijders, F. Salvestrini, E.Grasa, M. Tarzan, L. Bergesio, D. Staessens, D. Colle Prototyping [RINA], the IRATI project approach, IEEE Network, March

23 Prototype performances 25

24 GEANT3+ IRINA 27

25 IRINA - Intro Investigating RINA as the next generation GEANT and NREN network architecture (IRINA) GEANT3+ project Started Oct 2013, ends March 2015 (18 months) 4 Partners: [Research] iminds VZW(Belgium) [Research] Fundació Privada i2cat (Spain) [Research] Waterford Institute of Technology Telecommunications Software & Systems Group (Ireland) [SME] Nextworks s.r.l. (Italy) 28

26 IRINA Overview/Objectives 29

27 Internal NREN network design Multi DIFs (e.g. Public Internet DIF, application-specific DIF, etc) GEANT DIF Top-Level DIF Client DIF P2P DIF P2P DIF Aggregation DIF P2P DIF GEANT Border Router NREN Border Router P2P DIF P2P DIF NREN Interior Router Backbone DIF P2P DIF NREN Border Router Client Border Router P2P DIF P2P DIF NREN Border Router NREN Interior Router NREN Border Router GEANT DIF DIF 1 DIF 2 Top Level DIFs DIF 4 DIF 5 DIF 3 DIF 6 Client DIF DIF 1 DIF 2 DIF 3 Aggregation DIFs DIF 4 Backbone DIF 30

28 Lab trials of RINA rina-echo-time application of IRATI basic ping functionality rudimentary bandwidth testing capabilities Traffic generation tool is needed Netperf Distributed Internet Traffic Generator Ostinato 31

29 Traffic modelling IEEE Interrupted Poisson es (IPP) Interrupted Renewal es (IRP) Interrupted Discrete es (IDP) 4IPP IDP, 2IDP, 4IDP 2IRP 32

30 IRATI DEMO 33

31 Demo setup Host A Host B Application Application Normal DIF A Shim Shim DIF Shim 34

32 Where does the IRINA project fit in the big picture? RINA TIMELINE 35

33 An optimistic timeline National and Individual projects (US/EU) IRATI 01/ /2014 PRISTINE 01/ /2016 Future research projects Small lab prototypes Linux kernel prototype Mature Linux kernel prototype Niche Commercial products COTS Commercial products Initial specification (PSOC) Inter-university RINA / IPSec tunnels IRINA 10/13-03/14 Standardisation (ISO/SC6) NREN lab prototypes RINA DIFs supported by NRENs DIFs being adopted by Carriers ALL-RINA networks

34 Upcoming workshops Globecom Workshop Alternatives to TCP/IP 8-12 December, Austin TX US RINA workshop January 2015, Ghent Belgium TERENA TNC 2015 June, Porto, Portugal Summer school 2015 (?) 37

35 Thank you!