1 / 1

Goal: develop and explore scalability techniques for distributed service evaluation

from. routing matrix. to. route lookup. Network Switch. 100 M b /s. 1 G b /s. observe. packets/second. 10Mbit flows. cumulative percentage of transfers. KB/second.

Download Presentation

Goal: develop and explore scalability techniques for distributed service evaluation

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. from routing matrix to route lookup Network Switch 100 Mb/s 1 Gb/s observe packets/second 10Mbit flows cumulative percentage of transfers KB/second ModelNet: Scalability and Accuracy in a Large-Scale Network EmulatorKen Yocum, Kevin Walsh, Amin Vahdat, Priya Mahadevan, Dejan Kostic, Jeff Chase, and David BeckerDuke University, http://issg.cs.duke.edu Goal: develop and explore scalability techniques for distributed service evaluation trade accuracy for scalability ModelNet: Wide-area Emulation adaptive mesh refinement 2.) Distill target topology • simulation • slow, must modify applications • live deployment • limited in size, waiting for failures, not reproducible • Emulation • static, small scale, hard to configure • ModelNet: • accurate: error<1ms @ 95% cpu utilization • multiple techniques trade scalability for accuracy • Run unmodified applications on unmodified OS’s • reproducible results (recreate flash crowds, network partitions, etc) • Supports overlays, wireless networking, content distribution networks, peer-to-peer systems . . . clients stubs preserve first hopwalk_in = 1 1.) Create preserve core walk_out = 1 transits balance load & minimize hardware footprint bandwidth latency loss faults cross traffic ModelNet Performance topological cuts 3.) Assign • Assign links to cores. • balance load • reduce inter-core traffic • replicate pipes • observe/adjust • Graph partitioning problem • Spectral Analysis • Kernighan & Lin core A stubs clients Install ModelNet DataBase configuration core B 5.) Run core C • FreeBSD 4.5, Architecture X • Single Core Capacity (10vn’s/edgenode) • Per-Hop cost = 2.8 µs • Per-packet cost = 8 µs • Distillation matters • packet sizes (1500 bytes), full mesh transits Map virtual to physical 4.) Bind edge nodes emulation hardware client core ModelNet Validation Chord CFS interception DLL w.x.y.z -> 10.0.0.0/8 cross traffic emulation  delay loss app3 app2 app4 app1 kernel emulated link ... Unmodified Kernel Linux/FreeBSD/Solaris route 10.1.0.0/16 -> core ip emulated link modelnet • Emulated RON • reproduce end-to-end effects • uses single edge node ipfw src:10.0.0.0/8 dst:10.0.0.0/8 • multiple cores • payload caching • ship description of packet, not data network interface network interface core x core y

More Related