Changes in 45936be: documented the use of latency aliases in Route 53

				wiki/amazon-ec2.md
			
          @@ -300,6 +300,20 @@ With this set-up, please keep in mind that administration of the sailing server 

           You can monitor the central RabbitMQ message queueing system at [http://54.246.250.138:15672/#/exchanges](http://54.246.250.138:15672/#/exchanges). Use `guest/guest` for username and password. You should find the exchange name you configured for you master there and will be able to see the queues bound to the exchange as well as the traffic running through the exchange.

          +### Using Latency-Based DNS across Regions

          +

          +ELBs don't work across regions (such as eu-west-1 and ap-southeast-1) but only across the availability zones within one region. Therefore, if you want location and latency-based routing for your event, you have to set up so-called latency-based routing policies using alias record sets that point to your ELBs, one per region. The alias records are all for the same domain name and are each an alias for one ELB in one region. The DNS service will then find out for a requesting client which of the ELBs for the same name will provide the shortest network latency and return that record. Additionally, the DNS service can be configured to evaluate the ELBs health checking status. If the ELB has bad health, DNS requests will subsequently be answered using other alias records, therefore pointing to ELBs in other regions.

          +

          +<img src="/wiki/images/amazon/Route53_LatencyAliases.png" />

          +

          +The above image shows what you need to do: After you've set up an ELB in each of the regions as described above, for each of them add an alias record set. Enter the domain name and make sure it's the same for all ELBs. Change the "Alias" radio button to "Yes." The alias target can usually be selected from a drop-down, but in case you aren't offered the ELB you want to add, you can as well copy the ELB DNS name shown in the configuration page of that ELB.

          +

          +Change the routing policy to "Latency" and select the region in which your ELB is located. A meaninful ID string may be helpful later to identify the record set from a list of record sets.

          +

          +Select the "Yes" radio button for "Evaluate Target Health" in order to make sure the DNS server checks the ELB's health status and fails over to any of your other latency-based record sets if the ELB has bad health.

          +

          +Should you be switching from an Apache reverse proxy set-up with a record in `/etc/httpd/conf.d/001-events.conf` for your sub-domain, remember to comment or remove this record and to reload the httpd service on the central web server using `service httpd reload`. Its effects may otherwise interfere with the effects of the DNS entries for that same sub-domain. Conversely, before removing the DNS record sets for the sub-domain, if migrating back to a non-ELB, non-DNS scenario, remember to first re-activate the `001-events.conf` reverse proxy entry before removing the DNS record sets.

          +

           ### Access MongoDB database

           To launch a local MongoDB instance on the default port, enter a command such as `mkdir /home/sailing/mongodb; /opt/mongodb-linux-x86_64-1.8.1/bin/mongod --dbpath /home/sailing/mongodb`. To make sure the process survives log-off, you may want to launch it in a tmux session. See above for how to create a tmux session. In an existing tmux session, a new window can be created using `Ctrl-a c`.

				wiki/images/amazon/Route53_LatencyAliases.png
			
          Binary files /dev/null and b/wiki/images/amazon/Route53_LatencyAliases.png differ

...	...	@@ -300,6 +300,20 @@ With this set-up, please keep in mind that administration of the sailing server
300	300
301	301	You can monitor the central RabbitMQ message queueing system at [http://54.246.250.138:15672/#/exchanges](http://54.246.250.138:15672/#/exchanges). Use `guest/guest` for username and password. You should find the exchange name you configured for you master there and will be able to see the queues bound to the exchange as well as the traffic running through the exchange.
302	302
	303	+### Using Latency-Based DNS across Regions
	304	+
	305	+ELBs don't work across regions (such as eu-west-1 and ap-southeast-1) but only across the availability zones within one region. Therefore, if you want location and latency-based routing for your event, you have to set up so-called latency-based routing policies using alias record sets that point to your ELBs, one per region. The alias records are all for the same domain name and are each an alias for one ELB in one region. The DNS service will then find out for a requesting client which of the ELBs for the same name will provide the shortest network latency and return that record. Additionally, the DNS service can be configured to evaluate the ELBs health checking status. If the ELB has bad health, DNS requests will subsequently be answered using other alias records, therefore pointing to ELBs in other regions.
	306	+
	307	+<img src="/wiki/images/amazon/Route53_LatencyAliases.png" />
	308	+
	309	+The above image shows what you need to do: After you've set up an ELB in each of the regions as described above, for each of them add an alias record set. Enter the domain name and make sure it's the same for all ELBs. Change the "Alias" radio button to "Yes." The alias target can usually be selected from a drop-down, but in case you aren't offered the ELB you want to add, you can as well copy the ELB DNS name shown in the configuration page of that ELB.
	310	+
	311	+Change the routing policy to "Latency" and select the region in which your ELB is located. A meaninful ID string may be helpful later to identify the record set from a list of record sets.
	312	+
	313	+Select the "Yes" radio button for "Evaluate Target Health" in order to make sure the DNS server checks the ELB's health status and fails over to any of your other latency-based record sets if the ELB has bad health.
	314	+
	315	+Should you be switching from an Apache reverse proxy set-up with a record in `/etc/httpd/conf.d/001-events.conf` for your sub-domain, remember to comment or remove this record and to reload the httpd service on the central web server using `service httpd reload`. Its effects may otherwise interfere with the effects of the DNS entries for that same sub-domain. Conversely, before removing the DNS record sets for the sub-domain, if migrating back to a non-ELB, non-DNS scenario, remember to first re-activate the `001-events.conf` reverse proxy entry before removing the DNS record sets.
	316	+
303	317	### Access MongoDB database
304	318
305	319	To launch a local MongoDB instance on the default port, enter a command such as `mkdir /home/sailing/mongodb; /opt/mongodb-linux-x86_64-1.8.1/bin/mongod --dbpath /home/sailing/mongodb`. To make sure the process survives log-off, you may want to launch it in a tmux session. See above for how to create a tmux session. In an existing tmux session, a new window can be created using `Ctrl-a c`.