Without an autoscaler, customers frequently overprovision resources in anticipation of spikes in demand, a practice that often leads to increased cloud capacity costs. On the other hand, attempting to save on cloud capacity by provisioning what is needed manually, which is often accompanied by readjustment of assumptions, naturally leads to performance challenges. Manual scaling in response to increasing demand also risks delayed scaling action, leading to deterioration in service availability. Stable operations are either costly or difficult maintain.
With Tanzu Service Mesh Service Autoscaler, developers can enable autoscaling of their microservices in production environments and influence scaling behavior by setting scaling limits, restricting how quickly scaling should occur, determining what metrics and levels result in scaling responses, setting a grace period before microservices scale down, and setting the default number of minimum service instances for when there is insufficient metrics information.
Initially, developers might not know the scaling behavior of their services. In this case, you can use Tanzu Service Mesh Service Autoscaler as a testing tool to determine how many instances of services are needed to maintain various levels of demand on an application and what metrics are most important to consider to ensure smooth operations.