A graph more lovely than a tree [Add Presentation Title: Insert tab > Header & Footer > Notes and Handouts] 2/24/2019 A graph more lovely than a tree Joerg Ammon (jammon@brocade.com) Systems Engineer Service Provider © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

[Add Presentation Title: Insert tab > Header & Footer > Notes and Handouts] 2/24/2019 Legal Disclaimer All or some of the products detailed in this presentation may still be under development and certain specifications, including but not limited to, release dates, prices, and product features, may change. The products may not function as intended and a production version of the products may never be released. Even if a production version is released, it may be materially different from the pre-release version discussed in this presentation. NOTHING IN THIS PRESENTATION SHALL BE DEEMED TO CREATE A WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, STATUTORY OR OTHERWISE, INCLUDING BUT NOT LIMITED TO, ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NONINFRINGEMENT OF THIRD-PARTY RIGHTS WITH RESPECT TO ANY PRODUCTS AND SERVICES REFERENCED HEREIN. Brocade, the B-wing symbol, BigIron, DCX, Fabric OS, FastIron, IronView, NetIron, SAN Health, ServerIron, and TurboIron are registered trademarks, and Brocade Assurance, DCFM, Extraordinary Networks, and Brocade NET Health are trademarks of Brocade Communications Systems, Inc., in the United States and/or in other countries. Other brands, products, or service names mentioned are or may be trademarks or service marks of their respective owners. © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. Company Proprietary Information

Algorhyme V2, by Ray Perlner Taken from draft-ietf-trill-rbridge-protocol-nn.txt I hope that we shall one day see A graph more lovely than a tree. A graph to boost efficiency While still configuration-free. A network where RBridges can Route packets to their target LAN. The paths they find, to our elation, Are least cost paths to destination! With packet hop counts we now see, The network need not be loop-free! RBridges work transparently, Without a common spanning tree. © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019

TRILL – Technology Overview [Add Presentation Title: Insert tab > Header & Footer > Notes and Handouts] 2/24/2019 TRILL – Technology Overview © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

LANs Get L2 Multipathing Today: STP Single Path Core Aggregation Next: L2 Multi Path Servers Access Core Aggregation Access Servers Servers

RBridges Overview RB Implement TRILL protocol Use link state routing Perform L2 forwarding Provide point-to-point forwarding with zero configuration Can auto configure themselves RBridges forwarding tables scale with the number of RBridges RBridges know what options other RBridges support Support multipathing for unicast and multicast traffic Compatible with classic bridges and can be deployed in bridged LANs Ingress RBridge adds TRILL & outer MAC headers to frames Outer MAC header is modified hop-by- hop as with routing Egress RBridge decapsulates frames and learns the association of the “Inner MAC SA” with the Source RBridge nickname

How RBridges Work? Designated RBridge - DRB DRB AF VLAN2 RBridges discover each other by exchanging TRILL IS-IS (or FSPF) Hello frames TRILL Hellos are sent to the All-IS-IS-RBridges multicast address Using link state protocol (IS-IS or FSPF), a single Designated RBridge (DRB) is elected from among all RBridges on the LAN The DRB specifies the Appointed Forwarder (AF) for each VLAN The DRB also specifies the Designated VLAN for inter-RBridge communication RB 1 RB 2 RB 3 RB 4 RB 5 DRB AF VLAN2

How RBridges Work? Appointed Forwarder - AF The DRB specifies the Appointed Forwarder (AF) for each VLAN DRB can also be the AF Only ONE AF can be appointed per VLAN; One VLAN - One AF The AF is in charge of handling all native frames in the VLAN Ingress RBridge function: Encapsulates TRILL data frame Egress RBridge function: Decapsulates TRILL data frames

TRILL Encapsulation Unicast data path Known unicast destinations are forwarded RBridge hop by RBridge hop toward the egress RBridge 2. RB1 adds TRILL header and outer MAC header RB3 removes TRILL header and associates MAC-H with RB1 1. Host sends frames 4. Target receives frames

Utilizing ECMP Paths & Reordering TRILL supports up to 64 ECMP paths Packet (frames) ordering maintained within flows RBridges are required to maintain frame ordering internally When multi-pathing is used, all frames for an order-dependent flow must be sent on the same path if unicast or the same distribution tree if multi- destination Re-ordering can occur when A destination address transitions between being known and unknown A topology change occurs © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019

CORE TECHNOLOGY Brocade Virtual Cluster Switching (VCS) ETHERNET FABRIC DISTRIBUTED INTELLIGENCE LOGICAL CHASSIS DYNAMIC SERVICE INSERTION VM VM First data center Ethernet fabric No Spanning Tree Protocol Multi-path, deterministic Auto-healing, non-disruptive Lossless, low latency Built for convergence There are four primary areas in VCS. First, the technology allows construction of the industry’s first true, converged data center Ethernet fabric. That means that you can build a flat, multi-path, deterministic mesh network for the data center. It is inherently multipathing – there is no spanning tree, and there are no links that are shut off or on standby. You can group links between switches to increase capacity. The transport for the network is 10 Gbit Data Center Bridging (DCB), so it has low latency, it is lossless, and it is capable of running any type of traffic – normal data, iSCSI, CIFS, NFS, and FCoE. It is convergence-ready. NAS iSCSI FCoE © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

CORE TECHNOLOGY Brocade Virtual Cluster Switching (VCS) ETHERNET FABRIC DISTRIBUTED INTELLIGENCE LOGICAL CHASSIS DYNAMIC SERVICE INSERTION VM Fully distributed control plane Arbitrary topology, self-forming Network-wide knowledge of all members, devices, VMs Automatic Migration of Port Profiles (AMPP) conf conf Second, VCS supports a fully distributed control plane. That means every switch is aware of the entire network topology. You can configure it in a ring, a mesh, a tree. You can configure enough links to make it entirely non-blocking or configure the topology to be over subscribed at whatever level you choose. Every switch knows about every attached device and virtual machine. When a VM is attached to the network, the network configuration is automatically bound to the VM and then distributed throughout the VCS cloud. If the VM moves, the destination port already knows what to do and the VM is automatically reconnected to all of its old resources. The network configuration automatically migrates with the VM. The products that implement this technology will initially support a mobility sphere of 1000 10 Gigabit ports and 10,000 VMs. The means that you can move a VM within a 10,000 VM sphere and everything happens automatically. NAS iSCSI FCoE © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

CORE TECHNOLOGY Brocade Virtual Cluster Switching (VCS) ETHERNET FABRIC DISTRIBUTED INTELLIGENCE LOGICAL CHASSIS DYNAMIC SERVICE INSERTION Managed as a single switch Logically collapses network layers Auto-configuration for new devices Centralized or distributed management Radically reduces managed elements VCS dramatically reduces management complexity by treating the entire cluster as a single logical switch. As you add switches to an existing cluster, they configure themselves automatically and are simply added to the pool of available ports. Since all of the cluster information is distributed, there is no single master -- you can management the entire cluster from any switch. © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

CORE TECHNOLOGY Brocade Virtual Cluster Switching (VCS) ETHERNET FABRIC DISTRIBUTED INTELLIGENCE LOGICAL CHASSIS DYNAMIC SERVICE INSERTION VM VM Reconfigure network via software Hardware-based flow redirection Incorporation of partner services Non-stop service insertion Minimizes cost and physical moves Network Services Encryption Extension Security Lastly, VCS allows you to insert network services dynamically into existing flows of information. Instead of physically moving services like firewalls or encryption inline by rewiring them, you can simply redirect the flow of information to the attached services. This allows for the dynamic insertion of services and software-based reconfiguration of the network. The services can be systems that you build – like our encryption or extension products – or products from our partners like McAfee or Symantec or Bluecoat. © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

Shortest Path Forwarding Technology Summary TRILL SPB Target Application Data Center Data Center Metro/L2VPN Data Plane TRILL Encapsulation L2 Bridging (Q-in-Q or MAC-in-MAC) Control Plane IS-IS (or FSPF) IS-IS Loop Mitigation Hop Count RPF Check L2 ECMP  Native Multicast DCB Integration Widely Supported/ Market Trend  Standard Expected By End of 2010 Middle to Late 2011

Why another L2-solution? [Add Presentation Title: Insert tab > Header & Footer > Notes and Handouts] 2/24/2019 Why another L2-solution? © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

Trends in the Data Center Capitalize—Stay Ahead Data Center Virtualization ( Cloud) Consolidation Convergence ready Environmentals Gigabit ramp Key Takeaways: 5 key trends in the data center Virtualization Consolidation Convergence Environmental efficiency Gigabit ramp Enabling our customers to capitalize on or stay ahead of these industry trends is a principle driver behind our development strategy. We see 5 principal trends in the date center, beginning with virtualization. The cost savings virtualization offers for servers, networks & storage is well understood. Operational benefits include accelerating application deployment, improving service levels, and performing server maintenance without interruption. And by virtualizing physical resources you have a foundation for cloud computing. Brocade network virtualization capabilities & integrated management for VMware and HyperV environments are examples of advanced virtualization solutions from Brocade today.   Consolidation of server, storage, networks and data centers to reduce capital and operational costs remains a principle trend. The industry-leading performance and density of Brocade LAN and SAN routers and switches enable greater levels of consolidation and provide our customers with competitive advantage. Converging IP and storage traffic over a common network offers savings in networking capital and operations expenses. We are at the forefront of this technology and its capabilities will continue to expand. Today, Brocade offers the industry’s only unified top of rack solution for server I/O convergence. Or for future needs, having a network infrastructure that is convergence-ready provides investment protection. (Note: – We now refer to “Data Center Bridging” (DCB) instead of Converged Enhanced Ethernet or CEE, as customers and press now commonly use the technical term when referring to the enhancements being made to Ethernet to support storage traffic. Interest in server I/O convergence will vary greatly among customers and audience – IP network admins, SAN admins, network architects, C-level. While, many are testing, few are using in production at this time – this will be a long term evolution over many years, and for many may never be utilized with certain applications. What we often find is that customers want to have the infrastructure in place to enable convergence when and where it makes sense. We see Fibre Channel continuing to grow and DCB as a potential driver - in a converged environment every server essentially has access to the SAN. Del Oro sees Fibre Channel growing well into 2015 ) Facing limited power and cooling resources, reducing energy consumption and becoming more environmentally friendly is a global data center requirement. The energy efficiency of Brocade switches is unmatched and a fundamental design requirement. Finally, server virtualization, the rise in video use, and data growth is fueling exponentially greater traffic demand at the network edge. This in turn drives higher bandwidth at the network aggregation and core layers to avoid traffic congestion. Brocade’s LAN and SAN portfolio provides the industry’s highest 10GbE and 8 Gbps FC wire speed density, with platform architectures that will deliver 40 and 100GbE, as well as 16 Gbps FC SAN connectivity. (Note: Gartner sees appx 50 percent of servers to be virtualized by the end of 2012; in a traditional server environment with one application running, link utilization may avg 10-15%, allowing for hardware (switch) over subscription as well as uplink over- subscription. In a virtualized environments, multiple traffic flows drives link utilization rates upwards of 70-80+ percent, necessitating the need for wire-speed performance in the router) Enabling our customers to capitalize on or stay ahead of these industry trends is the principle driver behind our development strategy. © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019 17

MARKET TRENDS: DATA CENTER Data Center Construction App COMPUTE NETWORK STORAGE Traditional Construction Manual Replication Rigid Inflexible Past 15% 70% Utilization Operating Cost © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

MARKET TRENDS: DATA CENTER Data Center Construction App Early VMs COMPUTE NETWORK STORAGE COMPUTE NETWORK STORAGE Past Present 20% Escalating Complexity Virtual Apps (2010)1 1 Source: IDC Virtual Server Forecast 2010–2014, reported April 2010 © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

MARKET TRENDS: DATA CENTER Data Center Construction App Early VMs VMs COMPUTE NETWORK STORAGE COMPUTE NETWORK STORAGE Past Present Future Simplify Automate Scale © 2010 Brocade Communications Systems, Inc. Company Proprietary Information 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only

Data Center Evolution TODAY FUTURE HISTORICAL DYNAMIC NETWORKS SAN DYNAMIC NETWORKS VM Public Cloud LAN App Load FUTURE PRIVATE CLOUD VM Data Center Network ETH DCB FC Dynamic App Provisioning & Balancing Remote Sites Public Cloud HISTORICAL SAN LAN ENTERPRISE VM Key Takeaways: Historical: silos, separate LAN and SAN, and moderate VM usage Today: increasing use of server virtualization and # of VMs per host, mobility often limited to maintenance needs, application loads often balanced between physical servers in the data center; early testing of LAN and SAN I/O consolidation at the top of the rack, and select use of externally hosted services (cloud; e.g. Salesforce.com, Flightdeck) Future: highly virtualized compute/network/storage infrastructure; server connectivity mapped to application needs but viewed/managed as a unified network (key differentiator compared to Cisco positioning); computing services delivered from internal or external clouds; applications dynamically deployed and balanced across servers in primary data center, secondary sites, or in the public cloud (service provider) Historically, data centers scaled via physical replication; i.e, by replicating hardware for each application deployed. Server virtualization was limited to test and development environments and deployment was essentially a manual process. From a network perspective, LANs and SANs were completely separate and managed accordingly. Today, there is broad and increasing use of server virtualization and VM provisioning is more automated. VM mobility is utilized – but often limited to server maintenance. Companies are testing network convergence, but few are deploying in production. Load balancing between physical servers within the data center is commonly utilized for high demand applications. Companies are utilizing external hosting services (Public Cloud) for select application needs.   In the future we see a fully virtualized data center where resources - compute, network and storage - are dynamically allocated. Application mobility is ubiquitous and over distance. Unlike our competitors, we believe customers want the flexibility to map application needs with the most optimum type of network connectivity, be it FC, Ethernet or increasingly Data Center Bridging (DCB) - but view and manage their environment as a simply a data center network. With this highly virtualized infrastructure in place, applications can be provisioned and load balanced across physical servers residing in the Private Cloud - either locally (primary data center) or at remote sites (secondary data center). Or, as scalability/efficiency needs warrant, also servers residing at a service provider in the Public Cloud. © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019 21 21

The Network is Central to the Cloud Virtualization Brings New Demands Challenges Network performance/scalability constraints Application resiliency and performance under load VM mobility limits Infrastructure complexity Management silos Mobility APP OS VM APP OS VM APP OS VM APP OS VM Storage Network Compute Key Takeaways: At the center of the cloud is the network, connecting applications, storage, and users Virtualization brings cost and deployment advantages and areas of simplification, it also brings new demands and challenges that must be overcome While virtualization brings significant cost and deployment advantages, it also brings new demands and network challenges. Industry discussions about data center virtualization and moving to a Cloud model often focus only on the compute and storage infrastructure, assuming that the network – the center of the Cloud connecting apps, storage, and users - remains unchanged.   Yet there are network performance and scalability constraints that must be overcome. Applications must be resilient to failures in the shared resource pools including network failures, as well as meet fluctuating performance demands. VM mobility must not be restricted by the limits of Layer 2 Ethernet (a VMware requirement for VM mobility.) Infrastructure simplicity – not more complexity - is needed. And management must be more holistic versus silo’d while maintaining control of respective storage, network, and physical/virtual server administration domains. © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019

[Add Presentation Title: Insert tab > Header & Footer > Notes and Handouts] 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only 2/24/2019 © 2010 Brocade Communications Systems, Inc. CONFIDENTIAL—For Internal Use Only