CONFIGURING A P+V NETWORK FOR NEUTRON Big Switch Networks April, 2014 MODERN NETWORKING: YOU ARE NEEDED AGENDA FOR TODAY • • • • • • Context: Modeling application connectivity in Nova vs Neutron Big Switch P+V Fabric (lab release) moving parts Hands-on 1: Configuring a P+V fabric from the CLI Hands-on 2: Exploring the OpenStack integration Advanced Topics Next Steps [Demo accounts available after the show at http://bsnlabs.bigswitch.com] ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 3 EVOLUTION OF NETWORK PROVISIONING 1993 2013 Terminal Protocol: Telnet ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL Terminal Protocol: SSH 4 A MINDSET TRANSITION IN NETWORKING More change in the competitive frontier in last 2 years than previous 20 years Speeds and feeds Agility and automation Ethernet, Fast Ethernet, Gigabit Ethernet, 10GE SDN Controllers, Fabrics, Linux Shells, Neutron ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 5 Do you ? ANDROMEDA ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 7 A BRIEF HISTORY OF SDN Just entering the third inning 2) SDN for vSwitch overlays 1) SDN for OpenFlow programming ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL We are here 3) SDN for P+V Fabrics 8 http://ow.ly/wPu3T (Help at: https://plus.google.com/hangouts/_/bigswitch.com/erik) CONTEXT / DESIGN DECISIONS NOVA NETWORKING AND NEUTRON IN CONTEXT Application connectivity graph Mosh Pit WWW App WWW WWW WWW Tidy WWW DB App “Enterprise” DB WWW WWW WWW WWW DB App DB App App Threat isolation, fault isolation, troubleshoot-ability, compliance, implicit or explicit contract ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL DB DB 11 NOVA NETWORKING AND NEUTRON IN CONTEXT Enforcing an application connectivity graph Tools Frequently Available: Constraints Frequently Found: • Stateful firewalls • Organization demarc points • Subnets / VLANs / Routes / ACLs • Security Groups • Host IP tables • Surrounding L2/L3 design • Provisioning automateability “Enterprise” WWW WWW WWW App App • Existing appliances • … DB DB Intended as a sample of common cases, not an exhaustive list ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 12 NOVA NETWORKING AND NEUTRON IN CONTEXT Common case Nova design, easily migrated to Neutron Design considerations: Public-facing VMs have 2nd IP address, a floating IP to public V/S • Implementable with either Nova VLAN Mgr or Neutron • L3 isolation of threats and faults • Susceptible to classes of L2-based threats and faults (e.g. arp spoofing / snooping / storms) • SGs are impractical to map to end points other than pairs of OpenStack VMs (LB, BM DB, Storage) • Low effort to get going, but beware “enterprise” issues over time WWW WWW WWW Each tier gets an SG w/ rules App App Each tier gets an SG w/ rules DB DB Each tier gets an SG w/ rules Each project gets one V/S (non-routable) SG: Nova or Neutron Security Group, V/S: VLAN and Subnet ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 13 NOVA UNDERLYING INFRASTRUCTURE VIEW Common case Nova design, moving parts Nova common case design: • Provision a route-able north/south VLAN (subnet used for floating IPs) • Provision non-routed tenant VLANs – ‘all VLANs everywhere’ design • Each project gets a VLAN and, if public connectivity needed, floating IPs • Beware VLAN capacity limits Leaf Switch/Router (static, all VLANs everywhere) vSwitch / Host IP Tables Nova Server (VLAN Mgr) Spine Switch/Router (static, all VLANs everywhere) ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 14 NOVA NETWORKING AND NEUTRON IN CONTEXT Advanced “enterprise” case Neutron design Design considerations: WWW WWW WWW Each tier gets a V/S R App • Neutron only (in practice) • L2 and L3 isolation of threats and faults • Simple to insert L3 services post deployment (next-hops) • Maps to any kind of end point (OpenStack VMs, bare metal, LBs, FWs, etc.) • Higher effort to get going, but maps to known “enterprise” practices over time App Each tier gets a V/S Each project gets a logical router with routes and ACLs DB DB Each tier gets a V/S SG: Nova or Neutron Security Group, V/S: VLAN and Subnet Note: not mutually exclusive with SGs Note: often use floating IPs instead of routes for public connectivity ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 15 NOVA NETWORKING AND NEUTRON IN CONTEXT Where and what is the logical router for each project? • ML2-style systems: specified compute node(s) running OpenStack L3 agent R ? • Overlay/underlay systems (commercial) : distributed L3 agent enforced in the vSwitch and overlay gateways (VTEPs) • Unified P+V fabrics (commercial): distributed L3 agent enforced in the vSwitch and physical fabric SG: Nova or Neutron Security Group, V/S: VLAN and Subnet Note: not mutually exclusive with SGs Note: often use floating IPs instead of routes for public connectivity ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 16 ML2 UNDERLYING INFRASTRUCTURE VIEW Common case ML2 design, moving parts L3 Agent Leaf Switch/Router (dynamic VLAN provisioning and pruning) (w HA Proxy / PaceMaker) vSwitch Neutron Server w ML2 plug-in and vendor DriverMechanism Spine Switch/Router (dynamic VLAN provisioning and pruning) ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 17 O/U UNDERLYING INFRASTRUCTURE VIEW Common case O/U design, moving parts Leaf Switch/Router (static, all VLANs everywhere) L3 Agent vSwitch Overlay Gateway (aka VXLAN VTEP function) Overlay Controller (VMs or physical appliance pair) Neutron Server Spine Switch/Router (static, all VLANs everywhere) ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 18 P+V UNDERLYING INFRASTRUCTURE VIEW Common case P+V design, moving parts Leaf Switch/Router (dynamic) L3 Agent vSwitch Overlay Gateway P+V Controller (VMs or physical appliance pair) Neutron Server Spine Switch/Router (dynamic) ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 19 BSN P+V CLOUD FABRIC INTRODUCTION CLOUD FABRIC HARDWARE AND SOFTWARE Bare Metal SDN Bare Metal and Open Switch Hardware • • Bare Metal Switch Hardware: The same hardware used by hyper scale data centers, purchased directly from OEM manufacturers Open Switch Hardware: Provided by Dell, a branded switch that ships with multiple OS options ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL Switch Light and SDN Controllers • Switch Light SDN Switch OS, Switch Light vSwitch and SDN Controllers • Different controllers for different uses (cloud fabric controller, monitoring fabric controller…) • All configuration, software upgrades and the vast majority of troubleshooting happens on the controller to simplify everything 21 BSN P+V CLOUD FABRIC Moving parts in an OpenStack deployment Switch Light OS on leaf (10G/40G bare metal switches) Switch Light vSwitch or OVS Big Switch Controller (VMs or physical appliance pair) Neutron Server Switch Light OS on spine (2-6 40G bare metal switches) ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 22 BSN CLOUD FABRIC • Good for Nova: CLOS bandwidth scaling properties, provision every VLAN to every edge port with no performance penalty, centralized troubleshooting • Good for Neutron: CLOS bandwidth scaling properties, physical leaf/spine and vSwitch provisioned through a unified controller, centralized troubleshooting, no o/u gateways, UI enhancements to simplify tenant provisioning and troubleshooting ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 23 HANDS ON, PART I BASIC INTRODUCTION START YOUR DEMO Go ahead and start the demo. This button will launch some scripts and load the dashboard’s topology view. Once the topology appears take a moment to observe the different components of the topology. Note: This demo features the manual configuration of BVS using the CLI. BVS can also be configured automatically with OpenStack. To get a glimpse of BVS automation with OpenStack click here. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 25 CONNECTING TO YOUR CONTROLLER Click on the topology’s BVS icon to launch the BVS CLI. This will launch a popup shell. Go ahead and login with the credentials in your row on the sign up sheet. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 26 GET COMFORTABLE Drop into configuration mode and type “help” to bring up a list of all the commands the their descriptions. The BVS CLI is very intuitive. If you’re ever in doubt try using tab-complete or hitting “?” for available commands. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 27 EXPLORING THE BASIC SHOW COMMANDS Show switch Show host Show link ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 28 SHOW THE CURRENT CONFIG Note that the controller-node identifier is a dynamically generated value which can change between demos. Note that there are three tenants that are created by automatically and can't be deleted - default, system and external. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 29 EXPERIMENTING WITH THE REST API Try turning on the REST API debugging feature so you can see the exact REST calls over the wire. Like the CLI, you too can make calls to the REST API. This gives you the powerful opportunity to write scripts against the API. Turning off this feature is as easy as typing “no debug rest”. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 30 LINUX SHELL Other software can be loaded on to this basic linux image, and notice that linux monitoring / troubleshooting commands are available here. Use the exit command or Ctrl-D to exit from the linux shell and return to the CLI. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 31 BASIC LINUX SHELL TOOLS WITHIN THE BVS CLI Note that you can use some of the basic linux shell tools within the BVS CLI, most critically pipe, grep and concatenate. From within the CLI, access to the filesystem is limited (various security/permissions reasons) to a sandbox that is accessed using the "config://<filename>" convention. Output from there can be copied out to the controller filesystem as shown in the snippet below: ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 32 TESTING CONNECTIVITY Hop back to the topology view and check out the VMs hanging off vSwitch11 and vSwitch12. Click on the server icon to hop into each VM’s cli or check their network device info. You will notice that VM1/2 sit in 10.0.1.0/24 while VM3 sits in 10.0.2.0/24, VM4/5 sit in 10.0.3.0/24 and VM6 sits in 10.0.4.0/24. Drop into VM1’s CLI and test connectivity to other VMs. VM1 should not be able to connect to any other VMs because it is in a different subnet. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 33 LOOKING AT FLOWS ON THE CONTROLLER Instead hop into VM2 and initiate an ongoing ping between VM2 and VM3. Then hop onto the BVS CLI and try typing “show switch all flow” for a view of all the flows on the network. You can get more granular by specifying a specific switchalias as well. You can also ask for more details about the flows. Apparently enough details that it won’t fit in one terminal line! ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 34 DISTRIBUTED INTRA-TENANT ROUTING Inter VNS communication within the tenant via the Tenant Router Provide secure segmentation / Restriction via Router ACLs Tenant Virtual Router VM3 VM2 red-app red-web VM1 Red Tenant ©2013 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 35 EXPLORING BVS - CARVING UP L2 DOMAINS Virtual Network Segments CREATING YOUR FIRST TENANT: Set up a tenant called “red” with a “web” BVS (VM1, VM2) and an “app” BVS (VM3): ip-10-164-58-91(config)# tenant red ip-10-164-58-91(config-tenant)# bvs-definition red-www ip-10-164-58-91(config-tenant-def-bvs)# interface-rule red-www-if1 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# match ip 10.0.1.1/32 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# exit ip-10-164-58-91(config-tenant-def-bvs)# interface-rule red-www-if2 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# match ip 10.0.1.2/32 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# exit ip-10-164-58-91(config-tenant-def-bvs)# exit ip-10-164-58-91(config-tenant)# bvs-definition red-app ip-10-164-58-91(config-tenant-def-bvs)# interface-rule red-app-if1 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# match ip 10.0.2.3/32 ip-10-164-58-91(config-tenant-def-bvs-if-rule)# exit ip-10-164-58-91(config-tenant-def-bvs)# ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 36 DISTRIBUTED VIRTUAL ROUTING SERVICE Connecting VNSs Each tenant is given its own virtual router, along with the special "system," "external" tenants and their virtual routers. The term "router" may be a bit of an aggrandizement, as the tenant's virtual router configuration is only used for packet forwarding purposes and is not intended to provide the services of a full fledged router (e.g. dhcp). However, it is intended to transparently act on the packet as either a router (gateway ARP replies and subsequent TTL decrement and MAC address swaps) or as an ethernet bridge. The gist is to allow a single way to configure connectivity between groups of hosts regardless of how those hosts' had their subnet boundaries originally configured. Create a route to connect red-www and red-app: First, let's ensure that red-www (VM1 and VM2) are currently not connected to red-app (VM3) ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 37 DISTRIBUTED VIRTUAL ROUTING SERVICE Connecting VNSs Now configure the red tenant's router to connect red-www, red-app and the system tenant router (to be used later). Since the VMs are configured on the same L2 domain, no extra gateways are going to be needed so we can use the virtual router "shorthand" from within the red tenant submode. ip-10-152-143-155(config)# tenant red ip-10-152-143-155(config-tenant)# router vrred ip-10-152-143-155(config-tenant-router)# interface vrred-www-if bvs red-www ip-10-152-143-155(config-tenant-router-intf)# exit ip-10-152-143-155(config-tenant-router)# interface vrred-app-if bvs red-app ip-10-152-143-155(config-tenant-router-intf)# exit ip-10-152-143-155(config-tenant-router)# interface vrred-vrsystem tenant system vrsystem ip-10-152-143-155(config-tenant-router-intf)# exit ip-10-152-143-155(config-tenant-router)# route from bvs red-www to bvs red-app permit ip-10-152-143-155(config-tenant-router)# route from bvs red-app to bvs red-www permit ip-10-152-143-155(config-tenant-router)# exit ip-10-152-143-155(config-tenant)# ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM A virtual router routing rule can be specified by: • source: source can be one of the following: a specified host, a ip subnet, a defined VNS, a defined tenant • destination: destination can also be one of the following: a specified host, a ip subnet, a defined VNS, a defined tenant. • next hop ip address: This is optional. Specifying a numerical next hop on a directly connected interface prevents the router from performing ARP on each destination address. • outgoing interface: This is optional too. Specifying a directly connected outgoing interface forces the flow going through your planned route. • action: permit or deny. 38 ADDING IP INTERFACES TO TENANT ROUTER We can fix this by creating IP interfaces on the tenant router that it can route traffic. ip-10-152-143-155(config)# tenant red ip-10-152-143-155(config-tenant)# router vrred ip-10-152-143-155(config-tenant-router)# interface vrred-www-if bvs red-www ip-10-152-143-155(config-tenant-router-intf)# ip 10.0.1.254/24 ip-10-152-143-155(config-tenant-router-intf)# exit ip-10-152-143-155(config-tenant-router)# interface vrred-app-if bvs red-app ip-10-152-143-155(config-tenant-router-intf)# ip 10.0.2.254/24 ip-10-152-143-155(config-tenant-router-intf)# exit ip-10-152-143-155(config-tenant-router)# Now try pinging VM3 from VM1 and VM2 again… Success! ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 39 TEST PACKET IN Let’s do a test packet in between VM1 and VM3 to understand what is going on here. Type “test packet-in src-host VM1 ether-type ip dst-host VM3” Notice that there is now some information under “Virtual Routing Processing iterations”. You can find the source and destination BVS as well as the specific virtual router that handled the traffic between the two BVSs. Pretty cool. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 40 HANDS ON, PART II EXPLORING OPENSTACK INTEGRATION INITIAL LOGIN TO THE DEMO • Navigate to the demo hostname sent to you in your email. Append “bigdashboard” to the URL to access Big Switch’s plugin enabled OpenStack dashboard e.g. http://hostname/bigdashboard • When prompted, login using the demo credentials sent to you in your email ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 42 CREATE A TENANT USER AND PROJECT FOR THE USER • Navigate to Admin> User • Click the top right hand button “Create User” with following information: User Name: TenantA Password: bsn123 Email: TenantA@bigswitch.com • Create a new Primary Project (Tenant) Project Name: ProjectA ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 43 LOGIN TO BIG DASHBOARD AS A TENANT USER Logout and Log back in to the same URL with user created in previous step: http://hostname/bigdashb oard You will only see the tenant level (Project A) resources with this TenantA login ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 44 START PROVISIONING NETWORKS In this step, we will create three networks for “web”, “app”, and “db” tiers and assign them subnets 10.0.0.0/24 , 10.0.1.0/24, and 10.0.2.0/24 respectively. Select Networks under Project > Manage Networks On the top right, Click “Create Network” ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 45 START PROVISIONING NETWORKS Create the following 3 networks: Network Name: “web” Subnet Name: “web” Network Address: “10.0.0.0/24 “ Network Name: “App” Subnet Name: “App” Network Address: “10.0.1.0/24 “ Network Name: “DB” Subnet Name: “DB” Network Address: “10.0.2.0/24 “ Gateway IP: “10.0.2.1” ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 46 PROVISIONING NETWORKS Your output should look like this after you are done with provisioning all the three networks. Next Step is to create a tenant router for Intersubnet routing. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 47 PROVISIONING A ROUTER Click on the “Routers” menu item under “Manage Networks”. Click on the right top to create a new router Router Name “TenantA-Router”. Click on the router and you’ll find yourself in the router details page. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 48 PROVISIONING ROUTER INTERFACES As you can see there are no interfaces for this router. Click on the “Add Interfaces” button to begin adding interfaces. You will add three interfaces which are corresponding to the three Tenant subnets ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 49 PROVISIONING ROUTER INTERFACES Select all the three interfaces one by one and add to the tenant router ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 50 PROVISIONING ROUTER INTERFACES Once you are done adding all the three interfaces, verify the output under Routers > Interfaces tab. Now, you should be able to communicate between the three subnets via the tenant router. In next step, we will create virtual machine instances in each tier and test the routing functionality ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 51 DISTRIBUTED INTRA-TENANT ROUTING Big Switch Distributed Routing functionality provides distributing routing at each hypervisor level. If two virtual machines communicating with each other whether they belong to same tenant or different tenant, routing happens local to the hypervisor host ©2013 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM TenantA Tenant Virtual Router Web-001 Web App App DB-001 App-001 52 LOOKING AT CONTROLLER CONFIG In this step, we will log in to the BSN controller to find out what configuration has been pushed by OpenStack Plugin to configure Tenant Networks, Ports and Router. SSH to the Big Switch controller using the credentials provided to you. (ssh admin@BVS-ControllerIP). Once logged in, run the following command Show run tenant “Tenant ID” -- Use the tab key to see all the available tenants To find out the Tenant ID on OpenStack, Navigate to Manage Networks > Networks > web Controller CLI ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM OpenStack View 53 LOOKING AT BSN CONTROLLER Interface ID on the OpenStack maps to BVS definition Above steps involving BVS controller login was to show you the mapping between Openstack networking components to BVS controller ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 54 ISOLATING THE TIERS The Big Switch OpenStack integration provides an easy way to enable network policy to restrict the communication among tenant network or communication of tenant networks to external networks. Navigate to the “Tenant A” router’s detailed page and click on the “Router Rules Grid” tab. Note: This is a feature only available under “bigdashboard”. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 55 ISOLATING THE TIERS Disable communication between the database and the web tiers so that traffic must travel between the app tier to reach each tier. With that done, go navigate back to the web virtual machines console and try to ping each virtual machine again. ©2012 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM 56 NEXT STEPS ADVANCED TOPICS Available in whitepapers or 1:1s with our field teams • Designs for Nova and/or ML2 • Configuring HA controller pairs and HA Neutron service pairs • Configuring ACLs in/out of critical infrastructure or bare metal hosts • Inserting an L3 service by using ‘next-hop’ on routing rules (e.g. stateful firewall in between two tiers) • Configuring the ‘external’ tenant (demarc routers) – single and ecmp groups • Configuring dual homed hosts • Demonstrating multi-path across the fabric • Demonstrating resiliency against link failure, leaf failure, spine failure, controller failure, control network failure • Demonstrating scale ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 58 NEXT STEPS Want to learn more? • Demo accounts available at http://bsnlabs.bigswitch.com • Write to us at info@bigswitch.com • Write to me at kyle.forster@bigswitch.com • Or keep an eye out for announcements from us this summer ©2014 BIG SWITCH NETWORKS, INC. WWW.BIGSWITCH.COM PROPRIETARY AND CONFIDENTIAL 59 THANK YOU