Efficient Data Center Design with FabricPath (Advantage Webinar)
 

Efficient Data Center Design with FabricPath (Advantage Webinar)

on

  • 5,305 views

Cisco FabricPath is an innovation that simplifies Layer 2 switching by removing spanning tree and replacing it with IS-IS. FabricPath facilitates bandwidth provisioning, use efficient multipathing, ...

Cisco FabricPath is an innovation that simplifies Layer 2 switching by removing spanning tree and replacing it with IS-IS. FabricPath facilitates bandwidth provisioning, use efficient multipathing, provides safe forwarding for unicast and multicast frames, simplifies the design and reduces the costs of network and application deployment and operation.

Watch the WebEx replay: https://cisco.webex.com/ciscosales/lsr.php?AT=pb&SP=EC&rID=62375117&rKey=ac4438d06042f282

Statistics

Views

Total Views
5,305
Views on SlideShare
5,297
Embed Views
8

Actions

Likes
3
Downloads
283
Comments
1

4 Embeds 8

http://www.verious.com 4
http://localhost 2
https://si0.twimg.com 1
https://twitter.com 1

Accessibility

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
  • Very good Silde
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Efficient Data Center Design with FabricPath (Advantage Webinar) Efficient Data Center Design with FabricPath (Advantage Webinar) Presentation Transcript

  • Cisco IOS Advantage Webinars Efficient Data Center Design with FabricPath Babi Seal and Patrick Warichet We’ll get started a few minutes past the top of the hour. Note: you may not hear any audio until we get started.© 2012 Cisco and/or its affiliates. All rights reserved. 1
  • Speakers Panelists Babi Seal Patrick Warichet Francois Tallet Vimala Veerappan Product Manager Technical Marketing Product Manager Technical Marketing sseal@cisco.com Engineer Engineering Engineering pwariche@cisco.com ftallet@cisco.com vveerapan@cisco.com© 2012 Cisco and/or its affiliates. All rights reserved. 2
  • •  Submit questions in Q&A panel and send to “All Panelists” Avoid CHAT window for better access to panelists •  For Webex audio, select COMMUNICATE > Join Audio Broadcast •  For Webex call back, click ALLOW Phone button at the bottom of Participants side panel •  Where can I get the presentation? Or send email to: ask_iosadvantage@cisco.com •  Please complete the post-event Survey •  Join us on September 5 for our next IOS Advantage Webinar: A Closer Look: Comparing Benefits of EIGRP and OSPF www.cisco.com/go/iosadvantage© 2012 Cisco and/or its affiliates. All rights reserved. 3
  • •  Why Layer 2 in the Data Center?•  FabricPath solution overview•  Inside the Fabric•  FabricPath Unicast Details Agenda Back End Why Layer 2 in the Data Center?•  FabricPath Multicast Details FabricPath Solution Overview•  FabricPath designs Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 4
  • © 2012 Cisco and/or its affiliates. All rights reserved. 5
  • Key Decision Factors  Some protocols rely on the functionality Core  Simple, almost plug and play  No addressing Layer 3  Required for implementing subnets Layer 2 Aggregation  Allows easy server provisioning  Allows virtual machine mobility Services So what changed ? Access© 2012 Cisco and/or its affiliates. All rights reserved. 6
  • Change # 1 – Workload Virtualization Flexibility & Provisioning •  Partitioning •  Clustering •  Physical devices partitioned •  Applications distributed across into multiple servers virtual devices App App App App App OS OS OS OS OS OS OSVirtual MachinesPhysical Servers© 2012 Cisco and/or its affiliates. All rights reserved. 7
  • Big Data – Hadoop, NoSQL & HDFS Click Social Application Streams Media Virtualized, Mobility Bare-Metal, Cloud Event Sensor Trends Data Data Logs Fabric Traditional Storage “Big Data” “Big Data” Database NoSQL Real-Time Store And RDBMS SAN/NAS Capture, Read Analyze & Update HDFS - Hadoop Distributed File© 2012 Cisco and/or its affiliates. All rights reserved. System 8
  • Just extend STP to the whole network L3 L2 STP© 2012 Cisco and/or its affiliates. All rights reserved. 9
  • •  Local problems have network-wide impact, troubleshooting is difficult•  Tree topology provides limited bandwidth•  Tree topology introduces sub-optimal paths•  STP convergence is disruptive•  MAC address tables don’t scale•  Host flooding impacts the whole network© 2012 Cisco and/or its affiliates. All rights reserved. 10
  • An NX-OS Innovation Enhancing L2 with L3Layer 2 strengths Layer 3 strengths Fabric icity ility  Simple configuration  Flexible provisioning Path  All links active pl ib  Fast Convergence Sim Flex  Low Cost  Highly Scalable Resilience Simplicity Flexibility Bandwidth Availability Cost "The FabricPath capability within Ciscos NX-OS offers dramatic increases in network scalability and resiliency for our service delivery data center. FabricPath extends the benefits of the Nexus 7000 in our network, allowing us to leverage a common platform, simplify operations, and reduce operational costs.” Mr. Klaus Schmid, Head of DC Network & Operating, T-Systems International GmbH © 2012 Cisco and/or its affiliates. All rights reserved. 11
  • CY2011 1H CY2012 1H CY2013 •  Scale: 128 Switch IDs •  vPC+ on Fabric Extender host •  F1/M1 L2 proxy learning •  FabricPath on F2 I/O modules interfaces •  Multiple topologies •  Introducing native Fabric •  Conversational learning on F2 •  Overload bit Extender support •  PVLAN on F2 (FabricPath and (for least disruptive convergence) Nexus 7000 •  No conversational CE) •  route-map and mesh groups for scale learning •  FabricPath Traceroute (PONG) •  Vlan pruning enhancement (based on CE forwarding ports) •  Support on F2 10GbaseT I/O Modules •  Affinity to the closest rooted tree for multidestination traffic •  Anycast HSRP •  Scale 4K vlans, 512 switch IDs •  Cisco FabricPath SHIPPING NX-OS 5.2 •  Anycast HSRP Nexus 5500 •  16-Way ECMP •  Multi-Topology Support (2) •  Overload bit •  128 Switch IDs (for least disruptive convergence) •  STP Boundary & Termination •  FabricPath Traceroute (PONG) •  vPC+ (FabricPath Boundary) •  L2 proxy learning •  PIM SSM over vPC+© 2012 Cisco and/or its affiliates. All rights reserved. 12
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 13
  • Shipping Since 2010: turn your network into a Fabric FabricPath   Connect a group of switches using an arbitrary topology   With a simple CLI, aggregate them into a Fabric: N7K(config)# interface ethernet 1/1 N7K(config-if)# switchport mode fabricpath   An open protocol based on Layer 3 technology provides Fabric-wide intelligence and ties the elements together© 2012 Cisco and/or its affiliates. All rights reserved. 14
  • MAC IF A e1/1 … … FabricPath B s8, e1/2 e1/1 s3 s8 e1/2 A B•  Single address lookup at the ingress edge identifies the exit port across the fabric•  Traffic is then switched using the shortest path available•  Reliable L2 connectivity any to any (as if it was the same switch, no STP inside)© 2012 Cisco and/or its affiliates. All rights reserved. 15
  • Equal Cost MultiPathing (ECMP) e1/1 s3 s8 e1/2 A B•  Mutipathing (up to 256 links active between any 2 devices)•  Traffic is redistributed across remaining links in case of failure, providing fast convergence© 2012 Cisco and/or its affiliates. All rights reserved. 16
  • Conversational Learning MAC IF MAC IF A e1/1 A s1,e1/1 … … … … FabricPath B s8, e1/2 B e1/2 s3 s5 s8 e1/1 e1/2 A B MAC IF … …•  Per-port mac address table only needs to learn the peers that are reached across the fabricA virtually unlimited number of hosts can be attached to the fabric© 2012 Cisco and/or its affiliates. All rights reserved. 17
  • vPC+ FabricPath s3 s4 s7 s8 A VLAN X B VLAN Y VLAN Z•  Allows extending vlans with no limitation (no risks of loop)•  Devices can be attached active/active to the fabric using IEEE standard port channels and without resorting to STP© 2012 Cisco and/or its affiliates. All rights reserved. 18
  • Hosts Can Leverage Multiple Active L3 Default Gateways L3 FabricPath dg dg s3 A•  Hosts see a single default gateway•  The fabric provide them transparently with multiple simultaneously active default gateways•  Allows extending the multipathing from the inside to the fabric to the L3 domain outside the fabric© 2012 Cisco and/or its affiliates. All rights reserved. 19
  • Large tables (XL Hardware), SVIs Anywhere L3 FabricPath L3•  The fabric provides seamless L3 integration•  An arbitrary number of routed interfaces can be created at the edge or within the fabric•  Attached L3 devices can peer with those interfaces•  The hardware is capable of handling million of routes© 2012 Cisco and/or its affiliates. All rights reserved. 20
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 21
  •   Send/receive FabricPath frame   No STP, no MAC learning, no MAC address table   Using a routing table computed by IS-IS FP Core Ports S10 S20 S30 S40 Spine Switch Clos Fabric FabricPath (FP) S100 S200 S300 Leaf Switch 1/1 1/2 Classical Ethernet (CE) A B CE Edge Ports   Send/receive regular Ethernet frames   Run STP, do MAC address learning using a MAC address table© 2012 Cisco and/or its affiliates. All rights reserved. 22
  • Plug-n-Play L2 IS-IS Manages Forwarding Topology•  IS-IS assigns addresses to all FabricPath switches automatically•  Compute shortest, pair-wise paths•  Support equal-cost paths between any FabricPath switch pairs S10 S20 S30 S40 FabricPath Routing Table Switch IF S10 L1 S20 S30 L2 L3 FabricPath S40 L4 L1 L2 L3 S200 L1, L2, L3, L4 L4 … … S400 L1, L2, L3, L4 S100 S200 S300 S400© 2012 Cisco and/or its affiliates. All rights reserved. 23
  • •  The association MAC address/Switch ID is maintained at the edge S10 S20 S30 S40 Switch ID space: S300: FabricPath Routing decisions Routing Table are made based on A è B S100 è S300 the FabricPath Switch IF routing table FabricPath (FP) … … S100 S200 S300 S100 L1, L2, L3, L4 MAC adress space: 1/1 1/2 S300: CE MAC Switching based on Address Table MAC address tables Classical Ethernet (CE) MAC IF A B B 1/2 … A S100 …•  Traffic is encapsulated across the Fabric© 2012 Cisco and/or its affiliates. All rights reserved. 24
  • 16-Byte MAC-in-MAC Header Classical Ethernet Frame DMAC SMAC 802.1Q Etype Payload CRC 16 bytes Original CE Frame FabricPath Outer DA Outer SA FP Tag DMAC SMAC 802.1Q Etype Payload CRC (new) Frame (48) (48) (32) 6 bits 1 1 2 bits 1 1 12 bits 8 bits 16 bits 16 bits 10 bits 6 bits OOO/DL RSVDEndnode ID Endnode ID Sub Etype U/L I/G Switch ID LID Ftag TTL (5:0) (7:6) Switch ID 0x8903 •  Switch ID – Unique number identifying each FabricPath switch •  Sub-Switch ID – Identifies devices/hosts connected via VPC+ •  LID – Local ID, identifies the destination or source interface •  Ftag (Forwarding tag) – Unique number identifying topology and/or distribution tree •  TTL – Decremented at each switch hop to prevent frames looping infinitely © 2012 Cisco and/or its affiliates. All rights reserved. 25
  • S10 S20 S30 S40 A è B S100 è M S100 FabricPath S200 S300 Lookup B: Hit Learn source A Lookup B: Miss Flood Lookup B: Miss Don’t learnS100: CE MAC 1/1 S200: CE MAC 1/2 S300: CE MACAddress Table Address Table Address Table MAC IF MAC IF MAC IF A B … A … 1/1 … … B 1/2 … … … … … A S100 … Classical Ethernet© 2012 Cisco and/or its affiliates. All rights reserved. 26
  • S10 S20 S30 S40 S300: FabricPath Routing Table B è A S300 è S100 Lookup A: Hit Lookup A: Hit Switch IF Learn source B S100 FabricPath S200 S300 Send to S100 … … S100 L1, L2, L3, L4S100: CE MAC 1/1 S200: CE MAC 1/2 S300: CE MACAddress Table Address Table Address Table MAC IF MAC IF MAC IF A B … A … 1/1 … … B 1/2 … B … S300 … … … A S100 … Conversational Learning Classical Ethernet© 2012 Cisco and/or its affiliates. All rights reserved. 27
  • •  The Nexus 7000 features M and F I/O Modules•  FP Core and CE Edge ports must be on an F module•  New FabricPath/CE locally significant VLAN mode: S100(config)# vlan 10 S100(config-vlan)# mode ? FabricPath F ce Classical Ethernet VLAN mode FabricPath Core Port fabricpath Fabricpath VLAN mode F S100(config-vlan)# mode fabricpath Classical Ethernet Edge S100(config-vlan)# Port•  FabricPath VLANs can only be enabled on F modules•  FabricPath VLANs are also relevant to the Nexus 5500© 2012 Cisco and/or its affiliates. All rights reserved. 28
  • Traffic Forwarding Based on a Routing Table•  IS-IS assigns addresses to all FabricPath switches automatically•  Compute shortest, pair-wise paths•  Support equal-cost paths between any FabricPath switch pairs FabricPath Routing Table S10 S20 S30 S40 Switch IF S10 L1 S20 L2 S30 L3 S40 L4 S200 L1, L2, L3, L4 … … S400 L1, L2, L3, L4 L1 L2 L3 L4 FabricPath S100: CE MAC Address Table MAC IF S100 S200 S300 S400 A 1/1 B 400 A B© 2012 Cisco and/or its affiliates. All rights reserved. 29
  • Load Balancing on a Per-Tree Basis•  IS-IS computes several trees automatically•  Location of the root switches can be configured•  Multicast traffic is pinned to a tree at the edge S10 S20 S30 S40 Ingress switch for FabricPath decides which “tree” to be used and add tree number in the header L1 L2 L3 L4 FabricPath S100 S200 S300 S400 A© 2012 Cisco and/or its affiliates. All rights reserved. 30
  • Automatically Handled by IS-IS FabricPath V10 V20 V30 V30 V10 V20 V10 V30 V10 V20 V30© 2012 Cisco and/or its affiliates. All rights reserved. 31
  • Virtual Port Channel in FabricPath Environment → FabricPath → CE•  Allows non FabricPath capable devices to connect redundantly to the fabric using FabricPath port channels VPC+•  Provides active/active HSRP S1 S2•  Configuration virtually identical to standard VPC CE Device Host© 2012 Cisco and/or its affiliates. All rights reserved. 32
  • S10 S20 S30 S40 •  Mac address flapping on S300 •  Single path to A A è B S200 S100 è S300 S100 S200 FabricPath S300 1/1 1/2 S300: CE MAC Address Table MAC IF B 1/2 A A S100 S100 S200 B Classical Ethernet© 2012 Cisco and/or its affiliates. All rights reserved. 33
  • S10 S20 S30 S40 •  A consistently associated to S1 •  Multipathing to A A è B S1 è S300 S100 S200 FabricPath S300 1/2 S300: CE MAC Address Table MAC IF S1 virtual B 1/2 A A S1 B Classical Ethernet© 2012 Cisco and/or its affiliates. All rights reserved. 34
  • Identifies a VPC Off a Virtual Switch OOO/DL RSVD Endnode ID Endnode ID Sub U/L I/G Switch ID LID (5:0) (7:6) Switch ID FabricPath S300 è S1.22 C è B S1 S100 S200 S300 virtual subswitch ID 99 subswitch ID 22 A B Switch ID Sub Switch ID C© 2012 Cisco and/or its affiliates. All rights reserved. 35
  • Time To Live (TTL) and Reverse Path Forwarding (RPF) Check Root STP S1 S2 Root TTL=2 TTL=1 FabricPath S10 TTL=0 TTL=3•  The control protocol is the only   TTL in FabricPath header mechanism preventing loops   RPF Check for multi-destination traffic•  If STP fails infinite loop   The data plane is protecting against •  No backup mechanism in the data plane loops •  Flooding impacts the whole network© 2012 Cisco and/or its affiliates. All rights reserved. 36
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 37
  • •  Complete separation of control plane and data plane DRAP State Supervisor Database (PSS) Engine•  Fully modular software implementation FabricPath IS-IS of control plane and infrastructure components U2RIB L2FM•  Fully distributed data plane forwarding with hardware-based MAC learning / U2FIB MTM forwarding and hardware SID / ECMP Hardware Drivers lookups Switch Table Other HW MAC Table F1 I/O Module© 2012 Cisco and/or its affiliates. All rights reserved. 38
  • •  Describes shortest (best) paths to each Switch ID based on link metrics•  Equal-cost paths supported between FabricPath switches S10 S20 S30 S40 FabricPath Routing Table on S100 Switch IF One ‘best’ path S10 L1 to S10 (via L1) S20 L2 S30 L3 S40 L4 S200 L1, L2, L3, L4 … … S100 S200 FabricPath S300 Four equal-cost S300 L1, L2, L3, L4 paths to S300© 2012 Cisco and/or its affiliates. All rights reserved. 39
  • show fabricpath isis route S100# sh fabricpath isis route Fabricpath IS-IS domain: default MT-0 Topology 0, Tree 0, Swid routing table Destination Switch ID 10, L1 via port-channel10, metric 20 Routing metric 20, L1 via port-channel20, metric 20 30, L1 via port-channel30, metric 20 40, L1 via port-channel40, metric 20 200, L1 via port-channel30, metric 40 via port-channel40, metric 40 Next-hop interface(s) via port-channel20, metric 40 via port-channel10, metric 40 300, L1 S10 S20 S30 S40 via port-channel30, metric 40 via port-channel40, metric 40 via port-channel20, metric 40 via port-channel10, metric 40 po10 po20 S100# po30 po40 S100 S200 S300 FabricPath A B C© 2012 Cisco and/or its affiliates. All rights reserved. 40
  • show fabricpath route S100# sh fabricpath route Topology (ftag), Switch FabricPath Unicast Route Table ID, Sub-Switch ID a/b/c denotes ftag/switch-id/subswitch-id [x/y] denotes [admin distance/metric] ftag 0 is local ftag subswitch-id 0 is default subswitch-id Administrative distance, routing metric FabricPath Unicast Route Table for Topology-Default 0/100/0, number of next-hops: 0 Route age via ---- , [60/0], 0 day/s 04:43:51, local 1/10/0, number of next-hops: 1 via Po10, [115/20], 0 day/s 02:24:02, isis_fabricpath-default Client protocol 1/20/0, number of next-hops: 1 via Po20, [115/20], 0 day/s 04:43:25, isis_fabricpath-default 1/30/0, number of next-hops: 1 via Po30, [115/20], 0 day/s 04:43:25, isis_fabricpath-default Next-hop interface(s) 1/40/0, number of next-hops: 1 via Po40, [115/20], 0 day/s 04:43:25, isis_fabricpath-default S10 S20 S30 S40 1/200/0, number of next-hops: 4 via Po10, [115/40], 0 day/s 02:24:02, isis_fabricpath-default via Po20, [115/40], 0 day/s 04:43:06, isis_fabricpath-default via Po30, [115/40], 0 day/s 04:43:06, isis_fabricpath-default po10 via Po40, [115/40], 0 day/s 04:43:06, isis_fabricpath-default po20 1/300/0, number of next-hops: 4 po30 via Po10, [115/40], 0 day/s 02:24:02, isis_fabricpath-default po40 via Po20, [115/40], 0 day/s 04:43:25, isis_fabricpath-default S100 S200 S300 via Po30, [115/40], 0 day/s 04:43:25, isis_fabricpath-default via Po40, [115/40], 0 day/s 04:43:25, isis_fabricpath-default FabricPath S100# A B C© 2012 Cisco and/or its affiliates. All rights reserved. 41
  • (1) Broadcast ARP Request Root for Root for Multidestination Tree 1 Tree 2 Trees on Switch 10 S10 S20 S30 S40 4 Tree IFFtag → 1 po100,po200,po300 DA→FF Ftag→1 po300 2 po100 SA→100.0.12 DA→FF po100 po200 DMAC→FF Ftag→1 SMAC→A SA→100.0.12 Multidestination Payload DMAC→FF Trees on Switch 100 SMAC→A po10 po20 po30 po20 po30 po40 3 Tree IF po40 po10 PayloadBroadcast → 1 po10 S100 S200Multidestination S300 2 po10,po20,po30,po40 Trees on Switch 300 5 Tree IF 6 FabricPath MAC Table on S100 DMAC→FF e1/13 Ftag → 1 po10,po20,po30,po40 e2/29 Payload 2 po40 SMAC→A MAC IF/SID SMAC→A DMAC→FF A e1/13 (local) 2 Payload MAC A MAC B 1 FabricPath MAC Table on S200 MAC IF/SID Learn MACs of directly-connected Don’t learn MACs from devices unconditionally flood frames © 2012 Cisco and/or its affiliates. All rights reserved. 42
  • •  S100: S100# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------ MAC A learned as * 10 0000.0000.000a dynamic 0 F F Eth1/13 local entry on e1/13 S100#•  S10 (and S20, S30, S40, S200): S10# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link MAC A not learned VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID on other switches ---------+-----------------+--------+---------+------+----+------------------ S10#•  S300: S300# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------© 2012 Cisco and/or its affiliates. All rights reserved. 43
  • (2) Unicast ARP Reply Root for Root for Multidestination Tree 1 Tree 2 Trees on Switch 10 S10 S20 S30 S40 10 Tree IFFtag → 1 po100,po200,po300 po300 2 po100 DA→MC1 DA→MC1 Ftag→1 Ftag→1 po100 po200 SA→300.0.64 SA→300.0.64 DMAC→A Multidestination DMAC→A Trees on Switch 100 SMAC→B SMAC→B po20 po30 po40 po10 po20 po30 Payload Payload 11 Tree IF po40 po10Ftag → 1 po10 S200Multidestination S300 2 po10,po20,po30,po40 Trees on Switch 300 9 Tree IF 7 FabricPath MAC Table on S100 Payload e1/13 Unknown → 1 po10,po20,po30,po40 e2/29 DMAC→A 2 po40 SMAC→B MAC IF/SID SMAC→B Payload A e1/13 (local) 12 DMAC→A FabricPath MAC A MAC B B 300.0.64 (remote) MAC Table on S300 8 MAC IF/SID A→ MISS B e2/29 (local) If DMAC is known, then learn remote MAC *MC1 = 01:0f:ff:c1:01:c0 © 2012 Cisco and/or its affiliates. All rights reserved. 44
  • •  S100: S100# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------ * 10 0000.0000.000a dynamic 90 F F Eth1/13 S100 learns MAC B as 10 0000.0000.000b dynamic 60 F F 300.0.64 remote entry reached through S300 S100#•  S300: S300# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------ MAC B learned as * 10 0000.0000.000b dynamic 0 F F Eth2/29 local entry on e2/29 S300#© 2012 Cisco and/or its affiliates. All rights reserved. 45
  • (3) Unicast Data FabricPath Routing Table on S30 S10 S20 S30 S40 Switch IF … … S300 → S300 po300 16 po300 DA→300.0.64 DA→300.0.64 FabricPath Routing Ftag→1 Ftag→1 Table on S100 SA→100.0.12 SA→100.0.12 Switch IF DMAC→B DMAC→B S10 po10 SMAC→A po20 po30 po40 SMAC→A S20 po20 po10 po20 po30 Payload Hash po40 po10 Payload S30 po30 S40 po40 S200 FabricPath Routing S300 S100 po10, po20, Table on S300 S200 po30, po40 17 15 Switch IF po10, po20, e1/13 … … e2/29S300 → S300 po30, po40 Payload S300 → S300 Use LID (64) SMAC→A FabricPath DMAC→B FabricPath DMAC→B MAC Table on S100 SMAC→A MAC A MAC B Payload MAC Table on S300 MAC IF/SID A e1/13 (local) 14 13 MAC IF/SID 18 A S100.0.12 (remote)B→ B 300.0.64 (remote) If DMAC is known, then B e2/29 (local) learn remote MAC © 2012 Cisco and/or its affiliates. All rights reserved. 46
  • •  S100: S100# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------ * 10 0000.0000.000a dynamic 90 F F Eth1/13 10 0000.0000.000b dynamic 60 F F 300.0.64 S100#•  S300: S300# sh mac address-table dynamic Legend: * - primary entry, G - Gateway MAC, (R) - Routed MAC, O - Overlay MAC age - seconds since last seen,+ - primary entry using vPC Peer-Link VLAN MAC Address Type age Secure NTFY Ports/SWID.SSID.LID ---------+-----------------+--------+---------+------+----+------------------ S100 learns MAC A as 10 0000.0000.000a dynamic 30 F F 100.0.12 remote entry reached * 10 0000.0000.000b dynamic 90 F F Eth2/29 through S100 S300#© 2012 Cisco and/or its affiliates. All rights reserved. 47
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 48
  • •  Control plane: Build several multidestination trees Run IGMP snooping on FabricPath edge switches Advertise receivers location with dedicated LSPs•  Data plane (hardware): Selects which multidestination tree or each flow based on hash function Forward traffic along selected tree© 2012 Cisco and/or its affiliates. All rights reserved. 49
  • State Supervisor DRAP•  Complete separation of control plane Database (PSS) Engine and data plane FabricPath IS-IS IGMP•  Fully modular software implementation M2RIB L2FM for control plane and infrastructure MFDM components•  Fully distributed data plane forwarding M2FIB MTM with hardware-based MAC learning Hardware Drivers Switch Table Other HW MAC Table F1 I/O Module© 2012 Cisco and/or its affiliates. All rights reserved. 50
  • Multicast Trees Determination Root for Root for Tree 1 Tree 2 S100 S20 S10 S20 S30 S40 S10 S200 S30 Root S300 S40 Logical Tree 1 S100 S10S100 S200 FabricPath S300 S40 S200 S20 Root S300 S30 Logical Tree 2 •  Switch with highest priority value becomes root for primary tree Highest system ID, then highest Switch ID value, in case of a tie •  Primary root designates different secondary root(s) ensuring path variety. © 2012 Cisco and/or its affiliates. All rights reserved. 51
  • Multicast Tree Pruning Root for Tree 1 S10 S20 S30 S40 Switches interested in G1 VLAN 10 S200 S100 GM-LSP GM-LSP GM-LSP GM-LSP S100 S200 FabricPath S300 IGMP IGMP snooping snooping IGMP report IGMP report Rcvr-G1 Rcvr-G1 Src-G1•  IS-IS Group Membership LSPs contain multicast forwarding information© 2012 Cisco and/or its affiliates. All rights reserved. 52
  • Multicast Load Balancing Root for Root for Tree 1 Tree 2 S10 S20 S30 S40 S100 S200 FabricPath S300 Rcvr-G1 Rcvr-G1 Src-G1© 2012 Cisco and/or its affiliates. All rights reserved. 53
  • Multicast Data Plane Step by Step Root for Root for Tree 1 Tree 2 S10 S20 S30 S40 FabricPath MAC Table Ftag VLAN Group IFs po100 (Tree) po200 po300 1 10 G1 po300 FabricPath MAC Table Ftag 2 → 2 10 G1 po100,po200 Ftag (Tree) VLAN Group IFs   Ftag: Forwarding tag – 1 10 G1 po10 Unique 10-bit number 2 10 G1 po40, po50 ← Ftag 2 identifying topology and/or S100 po40 po10 po40 S200 FabricPath po10 po40 S300 distribution tree po50 po50 Data TrafficRcvr-G1 Rcvr-G1 FabricPath Src-G1 FabricPath MAC Table Multicast Trees Hash Ftag VLAN Group IFs VLAN Ftag (Tree) (Tree) 10 1 1 10 G1 po10 Packet data → → 10 2 → Ftag 2 → 2 10 G1 po40 © 2012 Cisco and/or its affiliates. All rights reserved. 54
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 55
  • FabricPath vs. vPC/STP L3 L3 Core •  Simple configuration •  No constraint in the design •  Seamless L3 integration •  No STP, no traditional bridging •  Mac address table scaling •  Virtually unlimited bandwidth •  Can extend easily and without FabricPath POD vPC POD operational impact© 2012 Cisco and/or its affiliates. All rights reserved. 56
  • Efficient POD Interconnect L3 •  FabricPath in the Core L2+L3 •  VLANs can terminate at the FabricPath distribution or extend between Core PODs. •  STP is not extended between PODs, remote PODs or even remote data centers can be aggregated. •  Bandwidth or scale can be vPC+ POD vPC+ POD introduced in a non-disruptive way© 2012 Cisco and/or its affiliates. All rights reserved. 57
  • FabricPath Site 1 •  Requires dark fiber •  Arbitrary interconnect topology (not dependent of port channels) •  Any number of sites FabricPath FabricPath FabricPath Site 4 Site 2 •  High bandwidth, fast convergence FabricPath Dark fiber •  Spanning tree isolation interconnect Site 3 (DWDM) •  Mac address table scaling •  VLANs can be selectively extended/terminated© 2012 Cisco and/or its affiliates. All rights reserved. 58
  • Provider B Provider A IXP Requirements •  Layer 2 Peering FabricPath •  10GE non-blocking Fabric •  Scale to thousands of portsProvider C FabricPath Benefits for IXP •  Layer 2 Fabric •  Non-blocking up to thousands 10GE ports Provider D Provider F •  Simple to manage Provider E •  No design constraint, easy to grow © 2012 Cisco and/or its affiliates. All rights reserved. 59
  • L3 •  FabricPath in the Core L2+L3 L2+L3 FabricPath FabricPath •  FabricPath extended Core down to the leaves vPC+ POD FabricPath POD vPC+ POD© 2012 Cisco and/or its affiliates. All rights reserved. 60
  • “Flattening” L3 •  FabricPath in the Core •  FabricPath extended down to the leaves •  There is enough bandwidth and port density on the core Nexus 7000s for FabricPath aggregating the whole network. •  There is no need for a distribution layer for POD isolation© 2012 Cisco and/or its affiliates. All rights reserved. 61
  • The Network Can Evolve With No Disruption•  Need more edge ports? → Add more leaf switches•  Need more bandwidth? → Add more links and spines L3 L3 FabricPath FabricPath© 2012 Cisco and/or its affiliates. All rights reserved. 62
  • Example: 2,048 x 10GE Server Design•  16X improvement in bandwidth performance•  6 to 1 consolidation (from 74 managed devices to 12 devices)•  2X+ increase in network availability•  Simplified IT operations (fewer devices, vlans anywhere) - Traditional Spanning Tree Based Network FabricPath Based Network Blocked Links Oversubscription 16:1 Fully Non-Blocking 2:1 FabricPath 4 Pods 8:1 64 Access Switches 8 Access Switches 2, 048 Servers 2, 048 Servers© 2012 Cisco and/or its affiliates. All rights reserved. 63
  • Business as Usual…•  Existing designs and best practices apply easily FabricPath FabricPath FabricPath SVIs, active/active vPC+ external Active/Standby© 2012 Cisco and/or its affiliates. All rights reserved. 64
  • Trying to adapt the existing model 100-200 300-400 100-400 100-400 FabricPath FabricPath Split VLANs GLBP Anycast FHRP •  Host ispolarization Some pinned to a •  All active •  single gateway can Inter-VLAN traffic •  Available in the future •  be suboptimal load Less granular for routing balancing •  Might not be available soon for service modules/appliances© 2012 Cisco and/or its affiliates. All rights reserved. 65
  • N-Way VPC+ •  Would allow to connect any device supporting port channeling FabricPath •  Not in the roadmap© 2012 Cisco and/or its affiliates. All rights reserved. 66
  • FabricPath Capable Router/Appliance •  Possible with Nexus platforms as routers FabricPath •  Other appliances not FabricPath capable© 2012 Cisco and/or its affiliates. All rights reserved. 67
  • Insert a FabricPath Switch •  Allows non-FabricPath capable devices to be attached to the fabric in an optimal way •  2 hops to reach the router/ appliance  Nexus 7000 F series I/O FabricPath modules and Nexus 5500 have lower latency anyway •  Thanks to FabricPath, enough bandwidth can be provisioned to the router/ appliance© 2012 Cisco and/or its affiliates. All rights reserved. 68
  • Attach to a Leaf Switch •  Allows non-FabricPath capable devices to be attached to the fabric in an optimal way •  2 hops to reach the router/ appliance  Nexus 7000 F series I/O FabricPath modules and Nexus 5500 FabricPath have lower latency anyway •  Thanks to FabricPath, enough bandwidth can be provisioned to the router/ appliance© 2012 Cisco and/or its affiliates. All rights reserved. 69
  • A 1/10G connectivity to Nexus 7000 F1 I/O Module FabricPath CX-1 cabling provide cost effective solution B 1/10G connectivity to Nexus 7000 F2 I/O Module C 1/10G connectivity to Fabric Extender attached to Nexus 7000 F2 I/O Module D A B C D E 1/10G connectivity to Nexus 5500 E 1/10G connectivity to Fabric Extender attached to Nexus 5500© 2012 Cisco and/or its affiliates. All rights reserved. 70
  • Agenda Back End Why Layer 2 in the Data Center? FabricPath Solution Overview Inside the Fabric FabricPath Unicast Details FabricPath Multicast Details FabricPath Designs© 2012 Cisco and/or its affiliates. All rights reserved. 71
  • •  FabricPath is simple, keeps the attractive aspects of Layer 2 Transparent to L3 protocols No addressing, simple configuration and deployment•  FabricPath is efficient High bi-sectional bandwidth (ECMP) Optimal path between any two nodes•  FabricPath is scalable Can extend a bridged domain without extending the risks generally associated to Layer 2 (frame routing, TTL, RPFC)© 2012 Cisco and/or its affiliates. All rights reserved. 72
  • © 2012 Cisco and/or its affiliates. All rights reserved. 73
  • What is TRILL? http://datatracker.ietf.org/wg/trill/•  TRILL, Transparent Interconnection of Lots of Links•  IETF standard for Layer 2 multipathing (since summer 2011)•  Pushed by Cisco but by other competitors too.Why aren’t we doing TRILL already?•  We delivered FabricPath before the standard, and some critical features are still missing to TRILL•  Our FabricPath capable hardware is also TRILL capable•  We will provide TRILL as a FabricPath mode© 2012 Cisco and/or its affiliates. All rights reserved. 74
  • FabricPath TRILL Frame routing Yes Yes (ECMP, TTL, RPFC etc…) vPC+ Yes No FHRP active/active Yes No Multiple topologies Yes No Conversational learning Yes No Inter-switch links Point-to-point only Point-to-point OR shared•  Cisco will push FabricPath specific enhancements to TRILL© 2012 Cisco and/or its affiliates. All rights reserved. 75
  • •  Thank you! •  Please complete the post-event survey. •  Join us September 5 for our next webinar: A Closer Look: Comparing Benefits of EIGRP and OSPF To register, go to www.cisco.com/go/iosadvantage© 2012 Cisco and/or its affiliates. All rights reserved. 76
  • Thank you.