Multi-Cluster Container Replication

Abstract

An improved scalable object storage system includes methods and systems allowing multiple clusters to work together. In one embodiment, there is a multi-cluster synchronization system between two or more clusters. Each cluster has a cluster-internal network, with object storage services and container services. The container services track and replicate metadata associated with the object storage service. An intercluster network connects the two clusters and performs a one-way synchronization of the objects and metadata associated with a particular container. This can be done either through the direct association of the container and object storage services, such as through a trust and federation relationship, or it can be opaque, so that the cross-cluster replication treats the remote repository as a black box and uses the external API to call and manipulate the files. In a further embodiment, multiple synchronization relationships can be set up, either in a cycle (with two or more participants), in a line, or in a tree. For example, the multi-cluster replication could be used to transparently synchronize objects in a CDN network.

1 . (canceled) 
     
     
         2 . A multi-cluster synchronization system, comprising:
 a first cluster including a first cluster-internal network, the first cluster further including a first structured information repository and a first object storage, wherein the first structured information repository contains metadata corresponding to stored information objects in the first object storage, and wherein the first structured information repository and the first object storage are coupled via the first cluster-internal network;   a second cluster including a second cluster-internal network, the second cluster further including a second structured information repository and a second object storage, wherein the second structured information repository contains metadata corresponding to stored information objects in the second object storage, and wherein the second structured information repository and the second object storage are coupled via the second cluster-internal network;   an intercluster network coupling the first cluster and the second cluster; and   an intercluster repository synchronizer coupled to the first and second structured information repositories; the intercluster repository synchronizer further including:
 a target module identifying the first cluster as a replication source and the second cluster as a replication target; 
 a consistency evaluation module adapted to evaluate the differences between the first structured information repository and the second structured information repository; and 
 an intercluster replication module adapted to selectively replicate stored information objects in the first object storage in the first cluster to the second object storage in the second cluster based on metadata in the first structured information repository. 
   
     
     
         3 . The system of  claim 2 , wherein the second cluster is also identified as a replication source, and wherein the first cluster is also identified as a replication target, and wherein stored information objects in the second object storage in the second cluster are selectively replicated to the first object storage in the first cluster. 
     
     
         4 . The system of  claim 3 , wherein the second cluster is identified as a replication source and the first cluster is identified as a replication target by a second target module associated with a second intercluster repository synchronizer. 
     
     
         5 . The system of  claim 1 , further including a third cluster including a third cluster-internal network, the third cluster further including a third structured information repository and a third object storage, wherein the third structured information repository contains metadata corresponding to stored information objects in the third object storage, and wherein the third structured information repository and the third object storage are coupled via the third cluster-internal network;
 wherein the second cluster is also identified as a replication source and the third cluster is identified as a replication target, and wherein stored information objects in the second object storage in the second cluster are selectively replicated to the first object storage in the third cluster.   
     
     
         6 . The system of  claim 5 , wherein the second cluster is identified as a replication source and the third cluster is identified as a replication target by a second target module associated with a second intercluster repository synchronizer. 
     
     
         7 . The system of  claim 2 , further including a third cluster including a third cluster-internal network, the third cluster further including a third structured information repository and a third object storage, wherein the third structured information repository contains metadata corresponding to stored information objects in the third object storage, and wherein the third structured information repository and the third object storage are coupled via the third cluster-internal network;
 wherein the third cluster is also identified as a replication target, and wherein stored information objects in the first object storage in the first cluster are selectively replicated to the third object storage in the third cluster.   
     
     
         8 . The system of  claim 2 , wherein the consistency evaluation module evaluates the differences between the first structured information repository and the second structured information repository by evaluating differences between a first stored information object in the first object storage and a second stored information object in the second object storage. 
     
     
         9 . The system of  claim 8 , wherein the first stored information object corresponds to the second stored information object. 
     
     
         10 . The system of  claim 9 , wherein the first and second stored information objects have the same name. 
     
     
         11 . The system of  claim 9 , wherein the first and second stored information objects are related replicas. 
     
     
         12 . The system of  claim 11 , wherein the consistency evaluation module evaluates the differences between the first stored information object and the second stored information object by applying an order independent check sum to the first and second stored information objects. 
     
     
         13 . The system of  claim 11 , wherein the consistency evaluation module evaluates the differences between the first stored information object and the second stored information object by applying content hashes on the first and second stored information objects. 
     
     
         14 . The system of  claim 11 , wherein the consistency evaluation module evaluates the differences between the first stored information object and the second stored information object by applying cryptographic hashes on the first and second stored information objects. 
     
     
         15 . The system of  claim 8 , wherein the metadata includes a version of the first stored information object and a version of the second stored information object. 
     
     
         16 . The system of  claim 8 , wherein the metadata includes a timestamp of an update to the first stored information object and a timestamp of an update to the second stored information object. 
     
     
         17 . The system of  claim 8 , wherein the consistency evaluation module aggregates objects stored in the first object storage based on a heuristic associated with low total differences. 
     
     
         18 . The system of  claim 17 , wherein the consistency evaluation module identifies a set of objects in the first object storage having the same name and size since a last replication pass and bundles the set of objects into a single virtual file. 
     
     
         19 . The system of  claim 17 , wherein the intercluster replication module sends the set of objects in the single virtual file to the second object storage in the second cluster.