Faizal discussing and demonstrates Windows Azure Traffic Manager which on CTP. Traffic Manager provides several ways to load balance traffic to multiple hosted services. You can choose from three load balancing methods: Performance, Failover, or Round Robin.
3. Windows Azure Traffic Manager Load balancing solution. Enables the distribution of incoming traffic among different hosted services in your Windows Azure subscription. Regardless of their physical location. Traffic routing occurs as a the result of policies that you define .
4. Windows Azure Traffic Manager Load balancing across multiple Hosted Services Integrated in the Windows Azure Platform portal Three scenarios enabled with the CTP: Performance Round Robin Fault Tolerance Redirect traffic to another deployment based on availability Directs the user to the best / closest deployment Traffic routed to deployments based on fixed ratio
5. Windows Azure Traffic Manager (WATM) Control the distribution of user traffic Performance, Business continuity, Price, Legal or Tax purposes. Compliance
6. Windows Azure Platform Availability Northern Europe North Central USA Eastern Asia Western Europe South Central USA Southeast Asia
7. How Windows Azure Traffic Manager works Assign each policy a DNS name Associate it with multiple hosted services. The load balancer responds to queries for the policy DNS name with the address of one of the associated hosted services that satisfies the criteria for the policy. Traffic Manager constantly monitors hosted services to ensure they are online and will not route traffic to any service that is unavailable.
9. How Windows Azure Traffic Manager Works Traffic Manager Policy Traffic Manager Domain Performance.worldapps.ctp.trafficmgr.com Load Balance Method Monitoring East Asia World Apps West Europe World Apps South Central World Apps Windows Azure Hosted Services
10. Performance Performance Times Table built periodically Incoming request Performance table lookup Best performing service chosen DNS name returned Service call
12. Windows Azure Traffic Manager - Performance Northern Europe North Central USA Eastern Asia Western Europe South Central USA Southeast Asia 4 6 3 1 Closest? 2 5
14. Windows Azure Traffic Manager - Round Robin Northern Europe North Central USA Eastern Asia Western Europe 6 South Central USA Southeast Asia 5 Next ? Last = West Europe 2 Last = Eastern Asia 4 1 3
16. Windows Azure Traffic Manager - Failover Northern Europe North Central USA Based your policy the traffic is sent to the next in order Eastern Asia Western Europe South Central USA 2 Southeast Asia 4 Your Application Offline @ US Central Data Center Online Hosted Service 1 3
19. Best Practices Services should be in a single subscription Production services only Name your hosted services so they can be easily found and clustered into policies All services in a policy should service the same operations and ports All services in a policy must use the same monitoring method
The numbers in the diagram correspond to the numbered explanation below it.GET - The Traffic Manager monitoring system performs a GET on the file you specify in the relative path area. 200 OK - The monitoring system expects a 200 OK message back within 5 seconds. When it receives this response, it assumes that the hosted service is online. 30 Seconds between checks - It will perform this check every 30 seconds. Hosted service offline - The hosted service goes offline. Traffic Manager will not know until the next monitor check.3 Tries to access monitoring file - The monitoring system performs a GET, but does not receive a response in 5 seconds or less. It performs two more tries at 30 second intervals. This means that at most it takes 1.5 minutes for the monitoring system to detect when a service goes down. If one of the tries is successful, then the number of tries is reset. Although not shown in the diagram, if the 200 OK response(s) come back greater than 5 seconds after the GET, the monitoring system will still count this as a failed check.Marked Offline - After the third failure in a row, the monitoring system will mark the hosted service as offline. Traffic to hosted service decreases – Traffic may continue to flow to the offline hosted service. Clients will experience failures because the hosted service is down. Clients and secondary DNS servers have cached the DNS address of the offline hosted service. They continue to resolve the DNS name of the company domain to the IP address of the hosted service. As DNS servers are updated, traffic to the offline hosted service IP address will stop. The monitoring system continues to perform checks at 30 second intervals and in this example the hosted service does not respond. It remains offline.Traffic to hosted service stops – By this time, most DNS servers and clients should be updated and traffic to the offline hosted service will stop. The maximum amount time before traffic completely stops is dependent on the TTL time. The default DNS TTL is 300 seconds (5 minutes). Using this value, clients stop using the service after 5 minutes. The monitoring system continues to perform checks at 30 second intervals and the hosted service does not respond.Hosted service comes back online – The hosted service comes back online, but Traffic Manager does not know until the monitoring system performs a check.Hosted service receives traffic again - Traffic Manager sends a GET and receives a 200 OK in under 5 seconds. It begins to hand out the hosted services DNS name DNS servers as they request updates. As a result, traffic starts to flow to the hosted service once again.