Data Center Automation
About Arista Networks 10/40/100GbE Networks for the Virtualized Cloud & Data Center Founded in 2004 Shipping Since Mid-2008 ANET, IPO (NYSE) in June 2014 1000+ Employees More than 3 million ports 3000+ Customers 8 out of 10 of the largest cloud operators Many large financial banks Build the best Open and Automation Network Operating System
2014 Gartner MQ Data Center Networking Arista Visionary Leader Arista best combination of Vision & Execution Arista only company to move up/right 2013-2014 Open and Programmable SDN platform for Datacenter Automation
Expectations within current data centers Always On expectation costs of downtime has skyrocketed and business SLAs has had to rise to keep pace Rise in data center Complexity with virtualization, multitier applications, and heterogeneous platform environments i.e. IT managers are struggling to get arms around the problem! Do more with Less tight resource constraints that hold IT budgets and headcount flat Business Agility quick time to market is key factor for maintaining competitive advantage
What is data center Automation?
Simply put Data Center Automation: is: automating changes which occur predictably and frequently, which would otherwise have been done manually!
Daily Jobs in Datacenter 10 Years Ago Systems Admin Network Admin
How about Systems Admin nowadays?
How about Network Admin nowadays?
Can Network Admin have a better life?
What is a Network device? L2 switching, VLAN, Trunk, STP, LACP L3 routing, IP, OSPF, BGP, PIM multicast OK other than that It s a Linux server with many NICs too Can we apply the beauties from servers to networks?
Automatic Network Topology Monitor and Report Topology and Config Template Various proactive types of alerts can be generated here based on network device connectivity: ZTP server Topology validation once ZTP process finish by LLDP checking against the defined diagram Proactive Alerting If there is ever a change in any or a key Route (e.g. OSPF, BGP, RIP, ISIS) / ARP / Mac-address, then the switch can send out a proactive alert to a centralized Syslog server notifying of the event LLDP Route changes / ARP changes / Mac address Change Management Once a network change has been complete, the switch automatically sends a Syslog server all the changes to (route / arp / mac) with the details of the engineer making the change Nightly report At the end of each day, a scheduled job can be run per network device listing all the (route / arp / mac) changes that occurred in the day
Automate the frequent network change due to workload mobility from Server virtualization DHCP vsphere vcenter Config Server interface Et1-$ vmtracer vmware-esx vmtracer session ESX1 url http://vsphere.foo.com/sdk username topsecret password letme1n range 100-199 Arista# show vmtracer interface Ethernet48 Ethernet48: esx1.arista.com/ndstest/dvuplink1 VM Name Network Adapter VLAN Status State ----------------------------------------------- Exchange Network adapter 4 7 up/up Apache Network adapter 3 6 up/up vmotion MySQL Network adapter 1 5 up/up FT-A
Talk to your network using IM ichat
One to Many: Grouping Changes, Config Rollback & Roll forward, Troubleshooting XMPP based Cloud Vision Server Management Server Devices are categorized in groups depending on topology Group A Group B Using a single CLI, the change control configuration can be rolled out from 0s to 000s of devices in the infrastructure Keep the last 50 configurations locally on the flash with a date, timestamp. If the user wants to be able to rollback to a previous configuration file, it is simple to do a diff on the current configuration and then roll the config back to a previous version if need be In the event of a network incident, instead of ssh or telnet into each device individually, using Arista s cloud vision server you can log into a GROUP of device and issue a single command eg to find an ip address and get the response back quickly
All boss like reports in GUI or spreadsheet instead of CLI SLA Reporting Syslog Server Monitoring Stations CVS Orchestration Layer 7150S-64 7150S-64 Spine-1 Spine-2 Spine-16 Vmware OpenStack Scripting Server Proxy / LB / Other DNS/ DHCP Customized SLA Report Switch SLA Report 99.999% Switch Event Value Threshold SLA Met Device Uptime 796 days 365 days Interface Uptime 100% 99% Latency <10ms <20ms Redundancy Uptime (MLAG) 800 days 1000 days CPU Historical <50% 75% % of Utilized ports <50% 75% # of VMs YES YES Downlink connectivity YES YES Immediate local view An SLA report card can be assembled per device to view the SLA status of that device by the local engineer to see if there are any issues on there quickly. Nightly Report On a scheduled basis, a report can be sent to a centralized server describing the health of the network Physical servers Virtualized Servers IP Storage Monitoring network Arista 7150S-64 Tap Aggregators DMZ Network 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 7150S-64 Pod-1 Pod-2 Pod-3 Pod-4 Pod-575
Switch Scorecard Switch Scorecard POOR STATUS Switch Event Value Threshold Alert Switch Scorecard POOR STATUS Switch Event Value Threshold Alert CPU 65% 75% Load average 1 min 0.3 1.0 Load average 5 mins 0.2 1.0 Load average 15 mins 0.1 10 Memory (free) 3GB 2.5GB * Flash (free) 3GB 3.5GB CPU Temp Sensor 24.518C 95C Linecard temp sensor 100C 95C * Powersupply 2 FAILED --- * FAN TRAY 1 OK OK FAN TRAY 3 FAILED OK * TCAM USAGE: IPV4 10 60 TCAM USAGE: IPV6 70 50 * LANZ Et1 200 100 * MLAG PORTS 4 3 1 MLAG port is down Port-channel 5 4 1 bundle is down Err-disabled 10 0 * Switch port security max mac 5 3 * Spanning-tree root change YES NO * Switch rebooted in last 24 hours YES NO * Tacacs server reachability NO YES * NTP server reachability NO YES * SNMP server reachability NO YES * CRC errors 2% 1% * QoS queue drops NO YES * GOLDEN Config Template NO YES * Run Config saved NO YES * BGP Neighbors down YES NO * OSPF DR/BDR down YES NO * Multicast RP reachabiliy NO YES * PIM Neighbors down NO YES * MSDP Peer down NO YES * Vcenter connectivity up NO YES * VM status up NONE YES * VXLAN Tunnels UP NO YES * EOS Code version 4.11.1 4.9.5 EOS standard does not match Immediate local view - A report card can be assembled per device to view the status of that device by the local engineer to see if there are any issues on there quickly. Nightly Report On a scheduled basis, a report can be sent to a centralized server describing the health of the network
Arista NEWSPAPER - Nightly Reporting Device in the last 24 HOURS Switch Event Value Threshold Alert # of Syslogs 5000 1000 * # of route changes 50 100 # of ARP changes 100 500 # of mac address chanes 0 500 Device Scorecard PASS PASS SLA Report NO YES * At the end of each night, the switch could be scheduled to send out a NIGHTLY HEALTH STATUS report. These could then be sent to a central arista server which could then be used by an overnight NOC team or be ready for first am status report before the start of the next production day # of users logged into system 5 10 # of configuration changes made 20 10 * Device Rebooted NO NO
Automation by DevOps Tools One for all Infastracture Network configuration and state inventory Standardized routing policy Extending 3 rd party command line tools into EOS CLI Configuration version control Deploying EOS extensions Standardized interface templates Security policy orchestration Software image updates Development QA Deployment Configuration Methodology
Reduce Opex & Capex Reduce MTTR and incident management Reduce Errors and Downtime Improve Productivity Simplify Operations Accelerate deployment of cloud Summary: Advantages by Automating the Data Center $
Interested to know more? Watch how our customers automate their data center network Contact us at sales@arista.com https://www.youtube.com/watch?v=2_1m3e57iio https://github.com/aristanetworks