System and method for transaction continuity across failures in a scale-out database

Patent metadata

Field	Value
Publication number	US-11983170-B2
Application number	US-202318117810-A
Country	US
Kind code	B2
Filing date	Mar 6, 2023
Priority date	Oct 14, 2020
Publication date	May 14, 2024
Grant date	May 14, 2024

Abstract

One or more engine instances are executed on each host to form an engine cluster. A plurality of control instances are executed on a first set of hosts to form a control cluster and comprise a control instance leader and one or more control instance followers. In response to a first host indicating a failure of a neighbor host, a pair-wise focused investigation is initiated to check peer-to-peer connections between the first host and the neighbor host. In response to one or more additional hosts indicating failures of neighbor hosts while the pair-wise focused investigation is being performed, a wide investigation is performed to check connections between the control cluster and the plurality of hosts. One or more hosts are added to an eviction list and an eviction protocol is performed to evict the one or more hosts from the engine cluster using the eviction list.

First claim

What is claimed is:

1. A method comprising:
on each host of a plurality of hosts, executing one or more engine instances of a plurality of engine instances to form an engine cluster, wherein each host of the plurality of hosts is a computing device;
establishing neighbor relationships among the plurality of hosts;
executing a plurality of control instances on a first set of hosts of the plurality of hosts to form a control cluster, wherein:
each control instance executes on a distinct host of the first set of hosts,
the plurality of control instances maintain data indicating the neighbor relationships between the plurality of hosts, and
the plurality of control instances comprise a control instance leader and one or more control instance followers;

in response to a first host within the plurality of hosts indicating a failure of a neighbor host, initiating a pair-wise focused investigation to check peer-to-peer connections between the first host and the neighbor host;
in response to one or more additional hosts within the plurality of hosts indicating failures of neighbor hosts while the pair-wise focused investigation is being performed, performing a wide investigation to check connections between the control cluster and the plurality of hosts;
adding one or more hosts to an eviction list based on results of the pair-wise focused investigation or the wide investigation; and
performing an eviction protocol to evict the one or more hosts from the engine cluster using the eviction list.

2. The method of claim 1 , wherein the pair-wise focused investigation comprises:
determining whether the first host can reach the control instance leader; and
in response to a determination that the first host cannot reach the control instance leader, adding the first host to the eviction list.

3. The method of claim 2 , wherein the first host ceases to process client requests for data until recovery is performed to add the first host back into the engine cluster.

4. The method of claim 2 , wherein the pair-wise focused investigation comprises:
in response to a determination that the first host can reach the control instance leader, attempting contact, by the control instance leader, with the neighbor host; and
in response to a determination that the control instance leader cannot communicate with the neighbor host, adding the neighbor host to the eviction list.

5. The method of claim 4 , wherein the pair-wise focused investigation comprises:
in response to a determination that the control instance leader can communicate with the neighbor host, determining whether the first host and the neighbor host have access to the same network;
in response to a determination that the first host and the neighbor host do not have access to the same network, using a tie-breaking heuristic to determine which of the first host and the neighbor host is to be evicted; and
adding the first host or the neighbor host to the eviction list based on results of the tie-breaking heuristic.

6. The method of claim 5 , wherein the pair-wise focused investigation comprises:
in response to a determination that the first host and the neighbor host have access to the same network, performing, by at least one control instance follower from the one or more control instance followers, a probing operation on the first host and the neighbor host;
determining a connectivity score for each of the first host and the neighbor host based on results of the probing operation; and
adding the first host or the neighbor host to the eviction list based the connectivity scores of the first host and the neighbor host.

7. The method of claim 6 , wherein the connectivity score is based on a number of members of the control cluster can communicate with the first host or the neighbor host.

8. The method of claim 6 , wherein the pair-wise focused investigation comprises:
in response to a determination that the connectivity score of the first host and the connectivity score of the neighbor host are not equal, adding the first host or the neighbor host having a lower connectivity score to the eviction list; and
in response to a determination that the connectivity score of the first host and the connectivity score of the neighbor host are equal, using the tie-breaking heuristic to determine which of the first host and the neighbor host is to be evicted; and
adding the first host or the neighbor host to the eviction list based on results of the tie-breaking heuristic.

9. The method of claim 1 , wherein the wide investigation comprises:
classifying connections between the control instance leader and each host of the plurality of hosts; and
adding a subset of the plurality of hosts to the eviction list based on results of classifying.

10. The method of claim 9 , wherein evictions of the subset of the plurality of hosts are batched.

11. The method of claim 9 , wherein classifying the connections between the control instance leader and each host of the plurality of hosts comprises classifying hosts that are unreachable into a subset of unreachable hosts, wherein the subset of the plurality of hosts includes the subset of unreachable hosts.

12. The method of claim 9 , wherein classifying the connections between the control instance leader and each host of the plurality of hosts comprises:
classifying one or more hosts to which the control instance leader has indirect access through a single network within a plurality of networks as a subset of fringe hosts;
identifying, by the control instance leader, at least one host within the subset of fringe hosts to be evicted; and
adding the identified at least one host to the eviction list.

13. The method of claim 12 , wherein identifying the at least one host within the subset of fringe hosts to be evicted comprises identifying at least one host that does not include a control instance.

14. The method of claim 12 , wherein:
the first set of hosts in the control cluster are connected using a first network,
the plurality of hosts in the engine cluster are connected using a second network, and
identifying the at least one host within the subset of fringe hosts to be evicted comprises identifying at least one host connected to the control instance leader using the first network.

15. The method of claim 14 , wherein classifying the connections between the control instance leader and each host of the plurality of hosts comprises classifying at least one host connected to the control instance leader using the first network and the second network as a subset of fully connected hosts.

16. One or more non-transitory storage media storing instructions which, when executed by one or more computing devices, cause performance of a method comprising:
on each host of a plurality of hosts, executing one or more engine instances of a plurality of engine instances to form an engine cluster, wherein each host of the plurality of hosts is a computing device;
establishing neighbor relationships among the plurality of hosts;
executing a plurality of control instances on a first set of hosts of the plurality of hosts to form a control cluster, wherein:
each control instance executes on a distinct host of the first set of hosts,
the plurality of control instances maintain data indicating the neighbor relationships between the plurality of hosts, and
the plurality of control instances comprise a control instance leader and one or more control instance followers;

in response to a first host within the plurality of hosts indicating a failure of a neighbor host, initiating a pair-wise focused investigation to ch

Assignees

• Oracle Int Corp

Inventors

• Mylavarapu Ajit
• Krishnaswamy Vasudha
• Pendse Sukhada
• Kolahi Solmaz
• Kumar Ankita
• Swart Garret F
• Lahiri Tirthankar
• Loaiza Juan R

Classifications

• G06F16/2379 Primary

  Updates performed during online database operations; commit processing · CPC title

• G06F16/285

  Clustering or classification · CPC title

• G06F16/278 Primary

  Data partitioning, e.g. horizontal or vertical partitioning · CPC title