Page compression strategy for improved page out process

A method comprising:

• identifying a set of uncompressed pages in a primary memory storage;
• assigning to each uncompressed page in the set of uncompressed pages a corresponding class from a pre-determined set of classes;
• compressing a plurality of pages in the set of uncompressed pages, a first set of pages of the plurality of pages being assigned to a first class are compressed into a single logical page, a second set of pages of the plurality of pages being assigned to a corresponding class different than the first class are compressed into a separate logical page different than the single logical page;
• determining a page weight factor for the single logical page;
• storing the single logical page in a secondary memory storage;
• paging in the single logical page into the primary memory storage;
• prompting a hypervisor to steal the single logical page, based, at least in part, on the pages of the plurality of pages being assigned to the first class;
• wherein:
• the page weight factor is made up of component weight factors assigned individually to pages within the single logical page;
• paging in the single logical page causes only a set of compressed pages assigned to the same class to paged into the primary memory storage;
• and
• the first class of pages is a member of a group consisting of:
• working segment pages;
• file segment pages;
• client pages;
• and
• klock pages.