Server Assisted TRILL Edge Linda Dunbar

Slides:



Advertisements
Similar presentations
LAN Segmentation Virtual LAN (VLAN).
Advertisements

© 2006 Cisco Systems, Inc. All rights reserved. ICND v2.3—2-1 Extending Switched Networks with Virtual LANs Introducing VLAN Operations.
Network Virtualization Overlay Control Protocol Requirements draft-kreeger-nvo3-overlay-cp-00 Lawrence Kreeger, Dinesh Dutt, Thomas Narten, David Black,
Mobility Jennifer Rexford COS 461: Computer Networks Lectures: MW 10-10:50am in Architecture N101
CMPE 150- Introduction to Computer Networks 1 CMPE 150 Fall 2005 Lecture 19 Introduction to Computer Networks.
PortLand: A Scalable Fault-Tolerant Layer 2 Data Center Network Fabric. Presented by: Vinuthna Nalluri Shiva Srivastava.
Directory Assisted TRILL Encapsulation by non-TRILL nodes (Directory Reliant Smart End Node) Linda Dunbar Donald Eastlake Radia Perlman Igor Gashinsky.
Switching Topic 4 Inter-VLAN routing. Agenda Routing process Routing VLANs – Traditional model – Router-on-a-stick – Multilayer switches EtherChannel.
VLANs Virtual LANs CIS 278.
Ethernet and switches selected topics 1. Agenda Scaling ethernet infrastructure VLANs 2.
Network Overlay Framework Draft-lasserre-nvo3-framework-01.
OpenFlow Deployment Anecdotes and Solutions David Erickson Stanford University October 17 th, 2011.
1 Interconnecting LAN segments Repeaters Hubs Bridges Switches.
COS 461: Computer Networks
1 K. Salah Module 4.3: Repeaters, Bridges, & Switches Repeater Hub NIC Bridges Switches VLANs GbE.
A Scalable, Commodity Data Center Network Architecture.
NVO3 NVA Gap Analysis Linda Dunbar Donald Eastlake.
(part 3).  Switches, also known as switching hubs, have become an increasingly important part of our networking today, because when working with hubs,
Virtual LANs. VLAN introduction VLANs logically segment switched networks based on the functions, project teams, or applications of the organization regardless.
Delivery, Forwarding and
Network Redundancy Multiple paths may exist between systems. Redundancy is not a requirement of a packet switching network. Redundancy was part of the.
Chapter 4: Managing LAN Traffic
G64INC Introduction to Network Communications Ho Sooi Hock Internet Protocol.
Connectivity Capability Features TOSCA. Aspects of Connectivity GenericIP Connectivity Resolvability: ARP: IP/MAC DNS: Name/IP MDNS: Netconf Routing/bridging/tunneling:
Directory Assisted RBridge Edge draft-dunbar-trill-directory-assisted-edge-05 Update Linda Dunbar: Donald
Semester 3, v Chapter 3: Virtual LANs
Network Aware Resource Allocation in Distributed Clouds.
VLAN V irtual L ocal A rea N etwork VLAN Network performance is a key factor in the productivity of an organization. One of the technologies used to.
Chapter 8: Virtual LAN (VLAN)
© 1999, Cisco Systems, Inc. Module 9: Understanding Virtual LANs.
Floodless in SEATTLE : A Scalable Ethernet ArchiTecTure for Large Enterprises. Changhoon Kim, Matthew Caesar and Jenifer Rexford. Princeton University.
Cisco 3 - LAN Perrine. J Page 110/20/2015 Chapter 8 VLAN VLAN: is a logical grouping grouped by: function department application VLAN configuration is.
© 2002, Cisco Systems, Inc. All rights reserved..
ARMD – Next Steps Next Steps. Why a WG There is a problem People want to work to solve the problem Scope of problem is defined Work items are defined.
BCP for ARP/ND Scaling for Large Data Centers
1 © 2003, Cisco Systems, Inc. All rights reserved. CCNA 3 v3.0 Module 9 Virtual Trunking Protocol.
Cisco S3C3 Virtual LANS. Why VLANs? You can define groupings of workstations even if separated by switches and on different LAN segments –They are one.
Routing and Routing Protocols
1 University of California, Irvine Done By : Ala Khalifeh (Note : Not Presented)
Chapter 3 - VLANs. VLANs Logical grouping of devices or users Configuration done at switch via software Not standardized – proprietary software from vendor.
Switching Topic 2 VLANs.
Chapter 4 Version 1 Virtual LANs. Introduction By default, switches forward broadcasts, this means that all segments connected to a switch are in one.
W&L Page 1 CCNA CCNA Training 2.5 Describe how VLANs create logically separate networks and the need for routing between them Jose Luis.
NVO3 Framework and Data Plane Requirement Addition Lucy Yong Linda Dunbar March 2013 Orlando FL draft-yong-nvo3-frwk-dpreq-addition-00.
1 VLANs Relates to Lab 6. Short module on basics of VLAN switching.
1 Large-scale (Campus) Lan design (Part II)  VLANs  Hierarchical LAN design.
Cisco 3 - Switch Perrine. J Page 12/4/2016 Chapter 9 Which protocol is Cisco proprietary and designed to carry traffic from multiple VLANs? A Q.
Ethernet Virtual LANs Hubs versus Switches –Hubs broadcast bits out all ports –Switches usually send a frame out a one port More fundamentally –In unicasting,
1 Chapter 3: Packet Switching (Switched LANs) Dr. Rocky K. C. Chang 23 February 2004.
VLAN Cisco (Router/Switch)
15.1 Chapter 15 Connecting LANs, Backbone Networks, and Virtual LANs Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or.
Ethernet Packet Filtering - Part1 Øyvind Holmeide Jean-Frédéric Gauvin 05/06/2014 by.
Scaling the Address Resolution Protocol for Large Data Centers (SARP) draft-nachum-sarp-04 Youval NachumMarvell Linda DunbarHuawei Ilan YerushalmiMarvell.
Address Resolution Issues Induced by VPN-oriented Cloud Service
VLAN Cisco (Router/Switch)
20. Switched Local Area Networks
Large-scale (Campus) Lan design (Part II)
Virtual Local Area Networks or VLANs
Routing Jennifer Rexford.
Virtual Local Area Networks (VLANs) Part I
3. Internetworking (part 2: switched LANs)
Chapter 4 Data Link Layer Switching
Chapter 5: Inter-VLAN Routing
Virtual LANs.
(How the routers’ tables are filled in)
2018/11/16 Scalable Address Resolution for Data Center and Cloud Computing Problem Statements Linda Dunbar Sue Hares
Connecting LANs, Backbone Networks,
Dr. Rocky K. C. Chang 23 February 2004
Virtual LAN VLAN Trunking Protocol and Inter-VLAN Routing
Presentation transcript:

Server Assisted TRILL Edge Linda Dunbar

Characteristics of Internet/Cloud Data Center Large number of hosts – Especially in data centers with virtualized servers (VMs). Resource (energy, bandwidth, etc) efficient algorithms emphasizes on minimizing the number of physical servers and switches for all the VMs, – This criteria may lead to hosts belonging to same subnet to be placed under different locations (racks or rows). Page 2

When one subnet spans across multiple locations and multiple subnets on one shelf: Broadcast messages of each subnet will traverse links and switches in the backbone (which involves much more links and switches). The Access switch need to learn all the MACs among all VLANs enabled on the ports. – E.g. One data centers has 80 rows, each row has 8 racks, each rack has 40 servers =>27200 servers. – Suppose each server is virtualized to 20 VMs. Then each TOR switch will have 20*40 = 800 VMs. If they belong to 80 different VLANs and each VLAN has 200 hosts, then the TOR switch has handle 80*200=16000 MAC&VLAN entries. Even though switch is only supposed to learn the entries which communicate with hosts underneath, the frequent ARP/ND from all hosts within each subnet will always refresh the switch’s FDB. Page 3 ABCD TRILL domain TOR switches performing TRILL encapsulation

Impact to TRILL by massive amount of hosts If TRILL boundary starts closer to servers, the number of MAC&VLAN entries to be learned by the TRILL edge is limited by the number of hosts belonging to all VLANs enabled on the TRILL edge port. – if the TRILL encapsulation starts at the Ingress port of Access switch (i.e. TOR switch), there are 5 different VLANs enabled on the port (i.e. VMs in one server belongs to 5 different VLANs), and each VLAN has 200 hosts, then TRILL Edge port has to learn 5*200=1000 MAC&VLAN entries. – But, there will be nodes in TRILL routing domain in the data center example. If TRILL boundary starts at End of Row switches, the number of nodes in TRILL routing domain is limited by the number of ports of all End of Row switches. – In the DC example, 40 servers in a Rack and 8 Racks in a row, Assuming 20 VMs per Server, there are 40*8*20 = 6400 hosts to each End of Row switch. If all those hosts belong to 640 VLANs and each VLAN has 200 hosts  total number of MAC&VLAN entries to be learned by the TRILL edge = 640*200= Page 4

Special properties of Data Center network VMs to server assignment is done by Server Manager, not random. Topology is based on racks, rows. There are massive number of hosts, but very small number of switches. Page 5

Directory Server Assisted Edge Directory Server, being connected with VM-manager, is aware of the location of the target VMs. Directory Server can assist TRILL edge by various ways, e.g. – Server sends MAC&VLAN TRILL edge mapping to TRILL edge node, – TRILL ingress node re-directs all ARP/ND to Directory Server and get the MAC&VLAN TRILL edge mapping as reply, or – Others Benefits: – Utilize the unique characteristics of Data Center networks – Reduce flooding within TRILL domain – Reduce the amount of MAC&VLAN TRILL edge mapping learnt by TRILL edge. Page 6