THE CHALLENGE -- how to cost-effectively provide Load Balancing as a Service (LBaaS) for cloud workloads. Customers expect the cloud to support their elastic workloads no matter how large they grow. On one hand, services, such as LbaaS, must have the resources to handle huge workloads; on the other hand, for the many customers that have very small workloads, the (free tier) LBaaS has to use as little resources as possible.
THE SOLUTION -- Elastic Load Balancing (ELB) An Elastic Load-Balancing Service dynamically aligns the load-balancing resources with the customer's elastic workloads. Use OpenStack Octavia to implement.
THE "DEEP-DIVE" -- Octavia provides an operator-grade reference implementation of LBaaS for OpenStack, utilizing HAProxy. Octavia supports various load balancing configurations to support different levels of performance, scalability, and availability. Octavia's new N+1 (a.k.a active-active, all active) configuration provides the basis for an Elastic LBaaS.
WHAT YOU GET --
1. A short introduction to Octavia and the new N+1, active-active configuration
2. Active-active configuration parameters and features
3. Introducing the Distributor -- the LB for LBaaS itself
3a. HW Distributor for CPU intensive configuration
3b. DNS Distributor for the IP rich
3c. Affinity and statefull load balancing nodes
4. Even better High Availability with HA-Proxy pairing
5. Containers integration
6. Demo: Elastic workloads under Octavia active-active configuration