Rackspace Auto Scale Overview
Rackspace Auto Scale automatically scales resources that are allocated to your application up or down based on user-defined need. Scaling events can be based on scaling rules that you define through your monitoring system, or on a simple schedule that you configure.
How Is Auto Scale Used?
Scaling is triggered either by webhooks that are placed in monitoring services or by a schedule. An Auto Scale group, or scaling group, consists of the following items:
- Scaling group configuration: The group name, cooldown time limit, and minimum and maximum number of needed servers
- Launch configuration: When an Auto Scale event is triggered, the actions taken to create a server by specifying which image to start and which load balancer to attach to the server
- Scaling policy: Defines the action of the policy (for example, scale up by 5 servers)
- Webhook (capability-based URL): Triggers a scaling policy (for example, “When this URL is hit, execute this scaling policy.”)
- eCommerce sale
Customer W has a one-day sale on its site on Monday, that is, beginning at 12:01 a.m. Monday and ending at 12:00 a.m. Tuesday. The customer predicts substantially increased traffic because of this sale. The customer can create a schedule-based policy to add 10 servers to its pool starting at 11:45 p.m. Sunday, which gives enough time for the servers to be created, configured, and added to the load balancer. The customer can create a second schedule to decrease the number of servers by 10 starting at 12:00 a.m. Tuesday morning, after the sale has ended.
- Marketing campaign
Similar to the preceding use case, Customer X is running a week-long marketing campaign and expects 20 percent more traffic to their website. The campaign begins at 7:00 a.m. EST on Monday and ends at 7:00 p.m. EST on Friday. The customer can create a schedule-based policy to increase the number of servers in their pool by 20 percent starting at 6:30 a.m. EST on Monday, which gives enough time for the servers to be created, configured, and added to the load balancer. The customer can create a second schedule-based policy to decrease the number of servers by 20 percent at 7:00 p.m. EST on Friday evening.
- Known traffic fluctuations
Customer Y notices that, traffic to their website decreases by 15 percent every weekend. That is, traffic decreases every Friday night and returns to normal on Monday morning. The customer can create a recurring schedule-based policy to decrease the number of servers by 15 percent starting at 6:00 p.m. every Friday. The customer can also create a second schedule-based policy to increase the number of servers by 15 percent every Monday morning at 6:00 a.m.
- Integration with Monitoring
Customer Z has unpredictable traffic when publishing blog posts. The customer already monitors the environment with Nagios. The customer creates a scaling policy to add 10 percent more servers with a webhook. The customer then creates a Nagios alert to trigger the webhook when traffic exceeds 6000 requests per minute. The customer creates a second policy to return the servers to the baseline when the spike decreases.
© 2011-2013 Rackspace US, Inc.
Except where otherwise noted, content on this site is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License
See license specifics and DISCLAIMER