Method and system for high availability topology for master-slave data systems with low write traffic
The present teaching relates to a system, method, medium for a data system, which comprises a plurality of data centers with one read/write master server designated to handle write and consistent read requests and a plurality of read master servers designated to handle normal read requests, wherein the master servers form a ring structure and each is connected to a corresponding slave server. When a data access request is received, a read master communication unit invokes one of the plurality of read master servers in the data system if the received data access request is a normal read request, a main master communication unit invokes the read/write master server if the received data access request is either a write request or a consistent read request. When a response is received from the invoked handling master server, a response processing unit responds to the data access request based on the received response.
1. A method implemented on a computer having at least one processor, a storage, and a communication platform for a data system, comprising:
detecting a fault in a server included in the data system, wherein the data system includes a read-write server and a plurality of read servers that are connected in a ring topology, wherein the read-write server is designated to process write operations and read operations of a first type, and the plurality of read servers are designated to process read operations of a second type that are different than the first type;
in response to detecting that the server that experiences the fault is the read-write server
selecting one of the plurality of read servers to serve as a new read-write server, and
forming a modified ring topology excluding the read-write server; and
in response to detecting that the server that experiences the fault is one of the plurality of read servers
forming the modified ring topology including the read-write server and one or more remaining read servers of the plurality of read servers.
2. The method of claim 1 , wherein each of the read-write server and the plurality of read servers is directly connected to a corresponding slave server.
3. The method of claim 1 , wherein in response to detecting that the server that experiences the fault is one of the plurality of read servers, the modified ring topology is formed by excluding the one of the plurality of read servers that experiences the fault.
4. The method of claim 1 , wherein one of the plurality of read servers that is selected to serve as a new read-write server is a first neighbor of the read-write server, and the step of forming the modified ring topology excluding the read-write server further comprises establishing a new connection from a second neighbor of the read-write server to the new read-write server.
5. The method of claim 4 , further comprising:
detecting a recovery of the read-write server;
re-forming the ring topology to include the read-write server;
replicating in the read-write server, transactions that occurred when the read-write server was in a fail state;
removing the established new connection; and
designating the new read-write server as a read server.
6. A machine readable non-transitory medium having information recorded thereon for a data system, where when the information is read by the machine, the information causes the machine to perform:
detecting a fault in a server included in the data system, wherein the data system includes a read-write server and a plurality of read servers that are connected in a ring topology, wherein the read-write server is designated to process write operations and read operations of a first type, and the plurality of read servers are designated to process read operations of a second type that are different than the first type;
in response to detecting that the server that experiences the fault is the read-write server
selecting one of the plurality of read servers to serve as a new read-write server, and
forming a modified ring topology excluding the read-write server; and
in response to detecting that the server that experiences the fault is one of the plurality of read servers
forming the modified ring topology including the read-write server and one or more remaining read servers of the plurality of read servers.
7. The medium of claim 6 , wherein each of the read-write server and the plurality of read servers is directly connected to a corresponding slave server.
8. The medium of claim 6 , wherein in response to detecting that the server that experiences the fault is one of the plurality of read servers, the modified ring topology is formed by excluding the one of the plurality of read servers that experiences the fault.
9. The medium of claim 6 , wherein one of the plurality of read servers that is selected to serve as a new read-write server is a first neighbor of the read-write server, and the step of forming the modified ring topology excluding the read-write server further comprises establishing a new connection from a second neighbor of the read-write server to the new read-write server.
10. The medium of claim 9 , further comprising:
detecting a recovery of the read-write server;
re-forming the ring topology to include the read-write server;
replicating in the read-write server, transactions that occurred when the read-write server was in a fail state;
removing the established new connection; and
designating the new read-write server as a read server.
11. A data system comprising:
a read-write server and a plurality of read servers that are connected in a ring topology wherein the read-write server is designated to process write operations and read operations of a first type, and the plurality of read servers are designated to process read operations of a second type that are different than the first type; and
a processing unit implemented by a processor and configured for
detecting a fault in a server included in the data system,
in response to detecting that the server that experiences the fault is the read-write server
selecting one of the plurality of read servers to serve as a new read-write server, and
forming a modified ring topology excluding the read-write server, and
in response to detecting that the server that experiences the fault is one of the plurality of read servers
forming the modified ring topology including the read-write server and one or more remaining read servers of the plurality of read servers.
12. The data system of claim 11 , wherein each of the read-write server and the plurality of read servers is directly connected to a corresponding slave server.
13. The data system of claim 11 , wherein in response to detecting that the server that experiences the fault is one of the plurality of read servers, the modified ring topology is formed by excluding the one of the plurality of read servers that experiences the fault.
14. The data system of claim 11 , wherein one of the plurality of read servers that is selected to serve as a new read-write server is a first neighbor of the read-write server, and the step of forming the modified ring topology excluding the read-write server further comprises establishing a new connection from a second neighbor of the read-write server to the new read-write server.
15. The data system of claim 14 , wherein the processing unit is further configured for
detecting a recovery of the read-write server;
re-forming the ring topology to include the read-write server;
replicating in the read-write server, transactions that occurred when the read-write server was in a fail state;
removing the established new connection; and
designating the new read-write server as a read server.