#GoogleNext18: Automating Large Scale Cloud Migrations to GCP with Velostrata

Velostrata will monitor workloads on VMware and recommend classifications with the appropriate sizing. This expands on Google Clouds cost savings like sustained use discounts and custom workload sizes. You can live test your migrations using a Test Clone. This allows you to snapshot a number of VMs and bring them up in isolation on GCP. You can do that via the console or an API call. You can also due site-level bandwidth throttling. This is important because most enterprises will throttle in but not necessarily throttle outgoing traffic. Velostrata Network settings enable you to set a bandwidth cap on all migration traffic from a specific site.

The solution is built to work within your private enterprise space so it will work behind NAT and proxies. Velostrata runs a hosted service to collect telemetry and log aggregation data for your migrations. Google recommends that you use discovery and assessment tools to define what moves and the appropriate order. It is important to develop a pipeline so that you can keep the migrations moving over at a predetermined rate and tempo for large customers. Google recommends that you run the migration in sprints, migrating within in a week while identifying the next workloads for the week following .

The Velostrata migration component is tightly integrated with VMware vCenter. You simply right click the VM and select migrate. The migration wizard enables you to select the Google VPC and makes recommendations for sizing. The monitoring is of the migration is done within the vCenter console. Any migration alerts or notifications are propagated into the vCenter monitoring system. You can both failover and failback.

Velostrata has full runbook automation capabilities. You create a runbook from within the Velostrata web console. The runbook is exported in csv format which allows you to filter and order the migrations. You can then pass the csv through the rightsizing module to determine what the appropriate Google class for the workload is. As it is developed in spreadsheet format the runbook is self documenting. To effect the migration you pass the csv to the migration job engine which displays the status of the migration of the group of VMs in the runbook.