Amazon Technologies, Inc.

Sumeet Talwar

Nagaraju Shiramshetti

Vivek Chawda

Ekanth Sethuramalingam

Keshav Sethi Attrey

Raviprasad Venkatesha Murthy Mummidi

Sophia Yeemei Tsang

Yupeng Zhang

Sagar Kumar

Vivek Menon

Sukwon Kim

Levente Csaba Tamasi

Swetha Sundar

Tianyi Luo

Nicholas King Bendler

Anbuselvan Ramanathan

Julian Ozen

Techniques for securely migrating servers from customer networks into service provider systems are described. A backup proxy can be deployed in a customer's network and associated with one or more servers in the customer's network and with a server migration service of a service provider system. A customer can identify a server in the customer's network to migrate and the server migration service coordinates the migration with the backup proxy. The backup proxy can be instructed to obtain replication data for the server, obtain an encryption key associated with the customer from a key management service (KMS), encrypt the replication data, and upload the encrypted replication data to the service provider system. The service provider system can obtain the same encryption key used to encrypt the replication data from the KMS and decrypt the uploaded encrypted replication data to generate migrated server resources at the service provider system.

Secure migration of servers from customer networks to service provider systems

A migration system or service may implement event-driven replication for migrating a computing resource from a source network to a destination network. A replication event may be detected to obtain a state of a data volume for a computing resource hosted in a client network. A request to a replication agent hosted in the source network may instruct the replication agent to copy at least some of the data of the data volume and send the data to the destination network. The migration service may generate a state of the data volume from the received data in a format that is operable to host the computing resource in the destination network and store the generated state of the data volume.

Event-driven replication for migrating computing resources

Techniques for self-organizing server migration into service provider systems are described. Connectors can be deployed in a customer network and associated with one or more virtual machine (VM) management servers and also with a server migration service. The customer may indicate what VMs are to be migrated, and the server migration service can dynamically select one or more of the connectors to perform migration operations based on operational or configuration characteristics of the connectors and/or the type of migration job to be performed. The connectors can be auto-scaled to increase or decrease operational capacity, and graceful failovers can be performed to eliminate migration failures resulting from connector failures.

Self-organizing server migration to service provider systems

Techniques for graph-based generation of dependency-adherent execution plans for data center migration are described. One or more agents and/or connectors can be deployed to a first network to collect data that can be used, by a server migration service, to identify resources operating in the first network and dependencies between the resources. A graph of the resources can be constructed and processed to remove any bi-dependencies and cycles from the graph that may exist. The resulting graph can be topologically sorted and used to automatically generate a migration plan for the resources that preserves dependencies between resources so that upon each resource being migrated, its inter-resource dependencies are satisfied.

Graph-based generation of dependency-adherent execution plans for data center migrations

Growing community of inventors

Santa Clara, CA, United States of America

Sumeet Talwar