为安全的可移植工作负载构建无缝多云平台.pdf

1、Building a Seamless Multi-Cloud Platform for Secure Portable WorkloadsJames Burns&Scott ReynoldsJune 10th 2025A Cloud Outage?!IntroHow many of you have workloads in GCP?How many of you have workloads in AWS?How many of you have workloads in Azure?How many of you have workloads in more than one cloud

2、 provider?How many of you run the same workload in more than one cloud provider?Makes meeting compliance standards easy,lawyers happy and engineers productiveWhat Data Users WantFind out what data is availableGet Access to the DataTest ChangesSpeedWhat Users are Not Asking forUnderstand and transfor

3、m informationWhy is that hard?Databricks Workspace provides the tools users wantUnity Catalog provides the governance layer users needCloud Provider provides storage and computeThe middle is where we connect it all togetherRecognizing Scaling Cost21How can weMake Scaling FreeDatabricks Platform23Mak

4、ing Scaling Free With Databricks at DatabricksKey Data Registry Concepts25Workload Definition26Data Generation or TransformationWorkload Definition27Data Generation or TransformationHow We Make Scaling FreeBaseline WorkspacesHow We Make Scaling FreeUse Case InterfaceUse Case30Use Case31How We Make S

5、caling FreeCloud Account PhaseCloud Account PhaseService AccountsIAM RolesIAM Assume Role Policies(Optionally)Pre-create all metastore storage33Cloud Resources for the Databricks AccountHow We Make Scaling FreeDatabricks Account PhaseDatabricks Account PhaseUnified Login Setup(AWS)SSO Setup(Others)U

6、ser Provisioning SetupAccount SPsAccount AdminMetastore AdminAccount Grant Groups35Databricks Account Baseline ConfigurationDatabricks Account PhaseAccount Groups SetupSCIM Push GroupsAccount Level DefaultsNetwork PoliciesCompliance Security ProfilesIP ACLs(Optionally)Pre-create all metastores36Data