Granted Patent B2

US 12,561,436 · App. 18/115,211 · Granted Feb 24, 2026

Inventors: Siamak Nazari (Mountain View, CA); David Dejong (Fremont, CA); Srinivasa Murthy (Cupertino, CA); Shayan Askarian Namaghi (San Jose, CA); Roopesh Tamma (San Ramon, CA)

Assignee: Nvidia Corporation

G06F21/566G06F9/4403G06F9/4416G06F11/1446G06F2221/033

Abstract

Claims (55)

1 . A process for operating a storage system including a plurality of servers containing a plurality of service processing units (SPUs), the process comprising:

creating a first snapshot set, the first snapshot set including a plurality of first snapshots respectively of a plurality of volumes that the SPUs maintain for storage services, the first snapshots capturing respective states that the volumes have at a first time;

creating a second snapshot set, the second snapshot set including a plurality of boot volume snapshots respectively of a plurality of boot volumes at a second time when a verified operating system image is written to the plurality of boot volumes;

reporting, by the storage system, the first snapshot set and the second snapshot set to a cloud-based service;

receiving, from a user of the storage system, communication to the cloud-based service about a recovery of the storage system;

providing for presentation to the user a list of recovery points, the recovery points including a first recovery point corresponding to the first time and a second recovery point corresponding to the second time, each recovery point corresponding to one or more snapshots of the plurality of volumes;

receiving from the user a selection to recover the storage system to the first recovery point, the first recovery point corresponding to the first snapshots; and

in response to the selection from the user, promoting the first snapshots to replace the volumes that SPUs present for storage services.

2 . The process of claim 1 , wherein the process further comprises creating a third snapshot set, the third snapshot set including a plurality of third snapshots respectively of the plurality of volumes, the third snapshots capturing respective states that the volumes have at a third time, wherein the list of recovery points presented to the user further includes a third recovery point corresponding to the third time.

3 . The process of claim 2 , further comprising instructing each of the SPUs to perform a snapshot operation at times indicated by a schedule, wherein:

each performance of the snapshot operation including each of the SPUs snapshotting each of the volumes that the SPU exports;

the schedule includes the first time and the third time;

the first snapshot set is results of the SPUs performing the snapshot operation at the first time; and

the third snapshot set is results of the SPUs performing the snapshot operation at the third time.

4 . The process of claim 1 , wherein the process further comprises receiving communication from the user to communicate with the cloud-based service about recovery while the volumes are corrupted or encrypted.

5 . The process of claim 1 , wherein when the cloud-based service presents the list, the cloud-based service also provides to the user information indicating whether the first snapshots appear to be corrupted or encrypted.

6 . The process of claim 1 , wherein creating the first snapshot set comprises performing, by each of the SPUs, a snapshot operation that includes:

suspending incoming storage service requests;

completing pending storage operations for prior storage service requests to the SPU;

creating the first snapshots of the volumes that the SPU maintains; and

resuming acceptance of incoming storage requests to the SPU.

7 . The process of claim 6 , wherein each of the SPUs creates the first snapshots of the volumes without copying data of the volumes.

8 . The process of claim 1 , wherein the promoting of snapshots comprises promoting, by the SPUs, their respective volumes, thereby rolling back an entire cluster of storage nodes associated with the SPUs.

9 . A cluster storage system comprising: a plurality of storage nodes each storage node including:

a server;

a backend storage device; and

an SPU (storage processing unit) connected to the backend storage device, the SPU operating the backend storage to provide storage services for a plurality of virtual volumes, wherein the SPU is configured to execute a first process including:

performing a snapshot operation at times indicated by a cluster-wide schedule, each performance of the snapshot operation including the SPU snapshotting each of the volumes that the SPU exports and thereby creating a synchronized snapshot set associated with the time of the snapshotting, wherein at least one snapshot operation includes creating a plurality of boot volume snapshots at a time when a verified operating system image is written to the plurality of boot volumes;

the SPUs are configured to respond to instructions from a cloud-based service by promoting the snapshots of a selected one of the synchronized snapshot sets to recover the cluster storage system to a state it had at the time associated with the synchronized set, wherein the instructions are based on a user selection from a list of recovery points, the user selection indicating the time as a recovery point.

10 . The cluster storage system of claim 9 , wherein the SPUs are configured to send information regarding the synchronized snapshot sets to the cloud-based service.

11 . A system, comprising at least one processor and memory storing instructions for executing steps, the steps comprising:

creating a first snapshot set, the first snapshot set including a plurality of first snapshots respectively of a plurality of volumes that a plurality of SPUs (storage processing units) maintain for storage services, the first snapshots capturing respective states that the volumes have at a first time;

creating a second snapshot set, the second snapshot set including a plurality of boot volume snapshots respectively of a plurality of boot volumes at a second time when a verified operating system image is written to the plurality of boot volumes;

reporting the first snapshot set and the second snapshot set to a cloud-based service;

receiving, from a user of a storage system, communication to the cloud-based service about a recovery of the storage system;

providing, for presentation to the user, a list of recovery points, the recovery points including a first recovery point corresponding to the first time and a second recovery point corresponding to the second time, each recovery point corresponding to one or more snapshots of the plurality of volumes;

receiving, from the user, a selection to recover the storage system to the first recovery point, the first recovery point corresponding to the first snapshots; and

in response to instructions from the cloud-based service associated with the selection, promoting the first snapshots to replace the volumes that SPUs present for storage services.

12 . The system of claim 11 , wherein the steps further comprise creating, by the SPUs, a third snapshot set, the third snapshot set including a plurality of third snapshots respectively of the plurality of volumes, the third snapshots capturing respective states that the volumes have at a third time, wherein the list of recovery points presented to the user further includes a third recovery point corresponding to the third time.

13 . The system of claim 12 , wherein the instructions further comprise instructing each of the SPUs to perform a snapshot operation at times indicated by a schedule, wherein:

each performance of the snapshot operation includes each of the SPUs snapshotting each of the volumes that the SPU exports;

the schedule includes the first time and the third time;

the first snapshot set is results of the SPUs performing the snapshot operation at the first time; and

the third snapshot set is results of the SPUs performing the snapshot operation at the third time.

14 . The system of claim 12 , wherein the steps further comprise receiving communication from the user to communicate with the cloud-based service about recovery while the volumes are corrupted or encrypted.

15 . The system of claim 11 , wherein when the cloud-based service presents the list, the cloud-based service also provides to the user information indicating whether the first snapshots appear to be corrupted or encrypted.

16 . The system of claim 11 , wherein creating the first snapshot set comprises performing, by each of the SPUs, a snapshot operation that includes:

suspending incoming storage service requests;

completing pending storage operations for prior storage service requests to the SPU;

creating the first snapshots of the volumes that the SPU maintains; and

resuming acceptance of incoming storage requests to the SPU.

17 . The system of claim 16 , wherein each of the SPUs creates the first snapshots of the volumes without copying data of the volumes.

18 . The system of claim 11 , wherein the promoting of snapshots includes all the SPUs promoting their respective volumes in response to the instructions from the cloud-based service, thereby rolling back an entire cluster of storage nodes associated with the SPUs.

19 . The system of claim 11 , wherein the steps further comprise deleting one or more snapshots of volumes that are not boot volumes.

20 . The system of claim 19 , wherein the one or more deleted snapshots are disqualified from providing a possible recovery point.

Assignments (2)

Continuity (5)

References Cited (40)