Client-based data replication
A client computing device having a processor and a memory receives, in response to a request, a replication layout from a replication controller. The client device initiates a data replication request at each of a plurality of data servers according to the replication layout. The client device receives, from each of the data servers, integrity information representing a state of a replication dataset at each of the servers. The client device transmits the integrity information to the replication controller.
1. A method, comprising:
receiving, in a client computing device having a processor and a memory, in response to a request, a replication layout from a replication controller;
initiating, in the client device, a data replication request at each of a plurality of data servers according to the replication layout;
receiving, in the client device, from each of the data servers, integrity information representing a state of a replication dataset at each of the servers; and
transmitting, from the client device, the integrity information to the replication controller;
wherein the integrity information includes a replication integrity assertion that is a mathematical result computed at a specific data server (RDS) for the layout (L); and
the integrity assertion is included in an ordered pair (IA L , S(IA L )), where IA L , represents the state of the replication dataset at a data server providing the integrity assertion, and S(IA L )) is a signature from the data server providing the integrity assertion that includes an encrypted authentication code.
2. The method of claim 1 , further comprising, in the replication controller, evaluating the integrity information to determine whether the replication request was performed successfully.
3. The method of claim 1 , wherein IA L is a linear hash value composed from recursively computed hash values of each data object modified under the layout.
4. The method of claim 1 , wherein IA L is based on is a file hierarchical checksum of objects L root represented in the layout at the data server providing the integrity assertion.
5. The method of claim 1 , wherein each of the data servers becomes stateless with respect to replication upon transmitting the integrity information.
6. A system, comprising:
a computing device that includes a processor and a memory, the device configured to:
receive, in response to a request, a replication layout from a replication controller;
initiate a data replication request at each of a plurality of data servers according to the replication layout;
receive, from each of the data servers, integrity information representing a state of a replication dataset at each of the servers; and
transmit the integrity information to the replication controller;
wherein the integrity information includes a replication integrity assertion that is a mathematical result computed at a specific data server (RDS) for the layout (L); and
the integrity assertion is included in an ordered pair (IA L , S(IA L )), where IA L , represents the state of the replication dataset at a data server providing the integrity assertion, and S(IA L )) is a signature from the data server providing the integrity assertion that includes an encrypted authentication code.
7. The system of claim 6 , wherein the replication controller is configured to evaluate the integrity information to determine whether the replication request was performed successfully.
8. The system of claim 6 , wherein IA L is a linear hash value composed from recursively computed hash values of each data object modified under the layout.
9. The system of claim 6 , wherein IA L is based on is a file hierarchical checksum of objects L root represented in the layout at the data server providing the integrity assertion.
10. The system of claim 6 , wherein each of the data servers becomes stateless with respect to replication upon transmitting the integrity information.
11. A computer-readable medium tangibly embodying instructions executable by a computer processor, the instructions comprising instructions for:
receiving, in a client computing device having a processor and a memory, in response to a
request, a replication layout from a replication controller;
initiating, in the client device, a data replication request at each of a plurality of data servers according to the replication layout;
receiving, in the client device, from each of the data servers, integrity information representing a state of a replication dataset at each of the servers; and
transmitting, from the client device, the integrity information to the replication controller;
wherein the integrity information includes a replication integrity assertion that is a mathematical result computed at a specific data server (RDS) for the layout (L); and
the integrity assertion is included in an ordered pair (IA L , S(IA L )), where IA L , represents the state of the replication dataset at a data server providing the integrity assertion, and S(IA L )) is a signature from the data server providing the integrity assertion that includes an encrypted authentication code.
12. The medium of claim 11 , wherein the replication controller is configured to evaluate the integrity information to determine whether the replication request was performed successfully.
13. The medium of claim 11 , wherein IA L is a linear hash value composed from recursively computed hash values of each data object modified under the layout.
14. The medium of claim 11 , wherein IA L is based on is a file hierarchical checksum of objects L root represented in the layout at the data server providing the integrity assertion.
15. The medium of claim 11 , wherein each of the data servers becomes stateless with respect to replication upon transmitting the integrity information.