Data storage system with multiple durability levels

Patent metadata

Field	Value
Publication number	US-11467732-B2
Application number	US-201916723391-A
Country	US
Kind code	B2
Filing date	Dec 20, 2019
Priority date	Dec 28, 2016
Publication date	Oct 11, 2022
Grant date	Oct 11, 2022

Abstract

A data storage system includes multiple head nodes and multiple data storage sleds mounted in a rack. For a particular volume or volume partition one of the head nodes is designated as a primary head node for the volume or volume partition. The primary head node is configured to store data for the volume in a data storage of the primary head node and cause the data to be replicated to a secondary head node. The primary head node is also configured to cause the data for the volume to be stored in a plurality of respective mass storage devices each in different ones of the plurality of data storage sleds of the data storage system.

First claim

What is claimed is:

1. One or more non-transitory computer readable media storing program instructions, wherein the program instructions when executed on or across one or more processors implement:
a head node, wherein the head node is configured to:
serve, when designated, as a primary head node for a volume partition; and
serve, when designated, as a secondary head node for the volume partition;

wherein, when designated as a secondary head node for the volume partition, the head node is configured to:
store write data, included with incoming write requests directed to another head node designated as a primary node for the volume partition, as replicated write data wherein the replicated write data is stored to a log-structured storage of the head node designated as the secondary head node such that the write data is stored in respective log-structured storages of both the other head node and the head node;
send an acknowledgement to the other head node designated at the primary head node that the write data has been replicated;
release, subsequent to respective parts of the write data stored in the log-structured storage of the primary head node being stored in mass storage devices, storage space of the log-structured storage of the head node designated as the secondary head node;

wherein, when designated as a primary head node for the volume partition, the head node is configured to:
replicate, in response to a failure of one or more of the mass storage device, data corresponding to data stored on the failed one or more mass storage devices to one or more other mass storage devices;
update an index of the head node designated as the primary head node to indicate new locations of the data for the volume partition; and
locate, based on the updated index, in response to a request targeting the volume partition, data of the volume partition stored in the head node and the one or more other mass storage devices storing the data that has been replicated.

2. The one or more non-transitory computer readable media of claim 1 , wherein:
the log-structured storage of the head node designated as the secondary head node comprises an index with pointers to where data is stored;
to perform said store the replicated write data, the program instructions when executed cause the one or more processors to store a replicated version of the write data to a head of a log of the head node designated as the secondary head node's log-structured storage; and
the program instructions when executed cause the one or more processors cause the head node designated as the secondary head node to add, to the index, an entry that indicates where the written data is stored in the log of the log-structured storage.

3. The one or more non-transitory computer readable media of claim 1 , wherein the program instructions when executed cause the head node designated as the secondary head node to:
write incoming replicated write requests to a head of the log of the head node designated as the secondary head node's log-structured storage;
add an index entry that indicates where the written data is stored in the head node designated as the secondary head node's log; and
update the index when the storage space corresponding to the replicated write data is released.

4. The one or more non-transitory computer readable media of claim 1 , comprising program instructions, that when executed:
detect a failed mass storage device in a particular data storage sled;
cause extents that include columns on the failed mass storage device to be replicated to other extents that include columns on other mass storage devices in other data storage sleds; and
update indexes of the head node designated as the secondary head node that includes an extent in the failed mass storage device to indicate the new locations of the data for the volume partition.

5. The one or more non-transitory computer readable media of claim 1 , comprising program instructions, that when executed:
in response to a failure of the primary or secondary head node for the volume partition, replicate the data included in the write requests to an additional head node of the data storage system designated as a replacement secondary head node for the volume partition.

6. The one or more non-transitory computer readable media of claim 1 , comprising program instructions, that when executed:
store the write data across at least six mass storage devices such that at least four of the mass storage devices store portions of the data, and at least two of the mass storage devices store coded data derived from the data.

7. The one or more non-transitory computer readable media of claim 1 ,
wherein the head node is included in a system that comprises data storage sleds each comprising a plurality of mass storage devices and a respective sled controller, and the mass storage devices that store the respective parts of the data are in different ones of the data storage sleds; and
wherein the program instructions when executed:
detect a failure of one or more of the mass storage devices in a particular one of the data storage sleds; and
copy data stored in remaining mass storage devices of the data storage sled to mass storage devices in one or more other data storage sleds.

8. A data storage system, comprising:
mass storage devices; and
head nodes, wherein for a volume partition of a volume to be stored in the data storage system, a first of the head nodes is designated as a primary head node for the volume partition, and a second of the head nodes is designated as a secondary head node for the volume partition, the primary head node configured to:
receive a write request from a client of the data storage system;
write data included with the write request for the volume partition to a log-structured storage of the primary head node;
cause the data included with the write request to be replicated to a log-structured storage of the secondary head node;
provide a write acknowledgement to the client in response to the write request subsequent to the data being replicated to the log-structured storage of the secondary head node, wherein the primary head node is configured to provide the write acknowledgement to the client prior to the data being flushed to one or more of the mass storage devices; and
cause, based on a trigger, respective parts of the data stored in the log-structured storage of the primary head node to be flushed to one or more of the mass storage devices.

9. The data storage system of claim 8 , wherein:
the log-structured storage of the primary head node and the log-structured storage of the secondary head node comprise respective indexes with pointers to where data is stored;
said write the data comprises write the data included with the request to a head of a log of the respective head node's log-structured storage; and
the respective head nodes are configured to add, to the respective head node's index, an entry that indicates where the written data is stored in the respective head node's log.

10. The data storage system of claim 9 , comprising:
an extent comprising columns of the mass storage devices to which the written data stored in the log-structured storage of the primary head node is flushed;
wherein the primary head node is configured to update the primary head node's index when the written data is flushed from the primary head node's log to the extent comprising the columns of the mass storage devices.

11. The data storage system of claim 8 , wherein:
said cause respective parts of the data stored in the log-structured storage of the primary head node to be flushed comprises cause the respective parts of the data to be era

Assignees

• Amazon Tech Inc

Inventors

• Kusters Norbert P
• Arumugam Nachiappan
• Watson Christopher Nathan
• Brooker Marc John
• Richardson David R
• Wei Danny
• Guthrie Ii John Luther

Classifications

• H04L67/568

  Storing data temporarily at an intermediate stage, e.g. caching · CPC title

• G06F2212/466

  Metadata, control data · CPC title

• G06F3/067 Primary

  Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS] · CPC title

• G06F2212/222

  Non-volatile memory · CPC title

• H04L67/1095

  Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes · CPC title