US 12,189,573 B2
Technique for creating an in-memory compact state of snapshot metadata
Abhishek Gupta, Ghaziabad (IN); Freddy James, Maharastra (IN); Pranab Patnaik, Cary, NC (US); and Ranjan Mn, Karnataka (IN)
Assigned to Nutanix, Inc., San Jose, CA (US)
Filed by Nutanix, Inc., San Jose, CA (US)
Filed on Jul. 15, 2021, as Appl. No. 17/376,822.
Claims priority of application No. 202141020922 (IN), filed on May 8, 2021.
Prior Publication US 2022/0358087 A1, Nov. 10, 2022
Int. Cl. G06F 16/11 (2019.01); G06F 16/16 (2019.01)
CPC G06F 16/128 (2019.01) [G06F 16/164 (2019.01)] 30 Claims
OG exemplary drawing
 
1. A method comprising:
scheduling creation of a first snapshot to support application operations processed by a node;
retaining a first compact state for the first snapshot in a memory of the node based on a predicted use of one or more fields of metadata of the first snapshot by a first application operation, wherein the one or more fields of the first compact state represent a filtered subset of metadata including frequently referenced attributes of the first snapshot, wherein the filtered subset of metadata is retrieved from a backing store based on frequency of use of the referenced attributes to demand-page the metadata from the backing store to reduce memory consumption;
completing the first application operation using the first compact state; and
in response to predicting that the first compact state is not needed by a future application operation, evicting the first compact state from the memory.