Replication of log-structured data

Patent metadata

Field	Value
Publication number	US-10346387-B2
Application number	US-201615180720-A
Country	US
Kind code	B2
Filing date	Jun 13, 2016
Priority date	Jun 13, 2016
Publication date	Jul 9, 2019
Grant date	Jul 9, 2019

Abstract

A method includes a primary storage unit receiving a first write request including a first key and a first value; persisting the first value in a first non-volatile memory in association with the first key; broadcasting the first write request and a first set of globally-durable keys to secondary storage units; receiving, from the secondary storage units, an acknowledgement of the first write request and a first set of locally-durable keys, each of the first sets of locally-durable keys including the first key; the primary storage unit receiving a second write request including a second key and a second value; persisting the second value in the first non-volatile memory in association with the second key; and broadcasting the second write request and a second set of globally-durable keys to the secondary storage units, the second set of locally-durable keys including the first key. A system is also disclosed.

First claim

What is claimed is:

1. A system comprising a primary storage unit and one or more secondary storage units:
a primary storage unit comprising:
a first non-volatile memory;
a first memory storing processor-executable process steps;
and a first processor to execute the processor-executable process steps to cause the primary storage unit to:
define two or more epochs from a directory of available storage units and storage cluster configuration information, each of the epochs specifying a primary storage unit and a respective set of secondary storage units of the one or more secondary storage units;
set each of the one or more secondary storage units to a respective epoch selected from the two or more epochs;
designate one of the two or more epochs as a current epoch;
receive a first write request from a client including a first key and a first value;
persist the first value in the first non-volatile memory in association with the first key;
broadcast the first write request and a first set of globally-durable keys to the one or more secondary storage units;
receive, from each of the one or more secondary storage units set to the current epoch, an acknowledgement of the first write request and a first set of locally-durable keys, each of the first sets of locally-durable keys including the first key;
each of the one or more secondary storage units not set to the current epoch ignores or rejects the first write request;
in response to the reception of the acknowledgement of the first write request and the first set of locally-durable keys from each of the one or more secondary storage units, transmit an acknowledgement of the first write request to the client;
receive a second write request from the client including a second key and a second value; persist the second value in the first non-volatile memory in association with the second key;
and broadcast the second write request and a second set of globally-durable keys to the one or more secondary storage units, the second set of globally-durable keys including the first key;
wherein the primary storage unit stores a first configuration index associated with the storage cluster, wherein the first write request includes a configuration index, and wherein the first processor is to execute the processor-executable process steps to cause the primary storage unit to:
determine whether the received configuration index is identical to first configuration index;
and ignore the first write request if it is determined that the received configuration index is not identical to first configuration index.

2. The system according to claim 1 , comprising:
a secondary storage unit comprising:
a second non-volatile memory;
a second memory storing processor-executable process steps; and
a second processor to execute the processor-executable process steps to cause the secondary storage unit to:
receive the first write request from the primary storage unit;
in response to reception of the first write request, persist the first value in the second non-volatile memory in association with the first key, and transmit the acknowledgement of the first write request and the first set of locally-durable keys to the primary storage unit;
receive the second write request and the second set of globally-durable keys from the primary storage unit; and
in response to reception of the second write request and the second set of globally-durable keys, persist the second value in the second non-volatile memory in association with the second key, persist a listing of the second set of globally-durable keys, and transmit the acknowledgement of the second write request and a second set of locally-durable keys to the primary storage unit, the second set of locally-durable keys including the first key and the second key.

3. The system according to claim 2 , comprising:
a management host comprising:
a third memory storing processor-executable process steps; and
a third processor to execute the processor-executable process steps to cause the management host to:
determine to remove a storage unit from the storage cluster;
instruct each storage unit of the storage cluster to become read-only;
receive a list of globally-durable and locally-durable keys from one or more of the storage units of the cluster;
determine a recovery set of globally-durable keys based on the lists of globally-durable and locally-durable keys;
define a second storage cluster comprising two or more storage units, the two or more storage units comprising a prior storage unit which was a storage unit of the storage cluster and a prior storage unit which was not a storage unit of the storage cluster; and
transmit the recovery set of globally-durable keys to the two or more storage units of the second storage cluster.

4. The system according to claim 1 , the first processor to further execute the processor-executable process steps to cause the system to:
receive a third write request from a second client including a third key and a third value;
persist the third value in the first non-volatile memory in association with the third key;
broadcast the third write request and a third set of globally-durable keys to the one or more secondary storage units;
receive, from each of the one or more secondary storage units, an acknowledgement of the third write request and a third set of locally-durable keys, each of the third sets of locally-durable keys including the first key, the second key and the third key; and
in response to the reception of the acknowledgement of the third write request and the third set of locally-durable keys from each of the one or more secondary storage units, transmit an acknowledgement of the second write request to the client.

5. The system according to claim 4 , the first processor to further execute the processor-executable process steps to cause the system to:
receive a fourth write request from the second client including a fourth key and a fourth value;
persist the fourth value in the first non-volatile memory in association with the fourth key;
broadcast the fourth write request and a fourth set of globally-durable keys to the one or more secondary storage units;
receive, from each of the one or more secondary storage units, an acknowledgement of the fourth write request and a fourth set of locally-durable keys, each of the fourth sets of locally-durable keys including the first key, the second key, the third key and the fourth key; and
in response to the reception of the acknowledgement of the fourth write request and the fourth set of locally-durable keys from each of the one or more secondary storage units, transmit an acknowledgement of the third write request to the second client.

6. The system according to claim 1 , comprising:
a management host comprising:
a third memory storing processor-executable process steps; and
a third processor to execute the processor-executable process steps to cause the management host to:
determine to remove a storage unit from the storage cluster;
instruct each storage unit of the storage cluster to become read-only;
receive a list of globally-durable and locally-durable keys from one or more of the storage units of the cluster;
determine a recovery set of globally-durable keys based on the lists of globally-durable and locally-durable keys;
define a second storage cluster comprising two or more storage units, the two or more storage units comprising a prior storage unit which was a storage unit of the storage cluster and a prior storage unit which was not a storage unit of the storage cluster; and
transmit the recovery set of globally-durable keys to the two or more storage units of the second storage cluster.

7. The system according to claim 6 , wherein determination

Assignees

• Sap Se

Inventors

• Bumbulis Peter
• Pound Jeffrey
• Auch Nathan
• Goel Anil Kumar
• Ringwald Matthias
• Bodner Thomas
• Maclean Scott

Classifications

• G06F16/2365

  Ensuring data consistency and integrity · CPC title

• G06F11/2058

  using more than 2 mirrored copies · CPC title

• G06F11/2097

  maintaining the standby controller/processing unit updated (initialisation or re-synchronisation thereof G06F11/1658 and subgroups) · CPC title

• G06F16/2358 Primary

  Change logging, detection, and notification (replication G06F16/27) · CPC title

• G06F11/2066

  Optimisation of the communication load · CPC title