So what we have is a patent on...
... using a hash (globally unique identifier, or GUID) as a search key to identify a specific linked list, which contains a set of records, in which said records contain time-stamp/expiry information, and a collision detection algorithm thrown-in to make sure conflicting updates are not made on the linked list.
This is utter bollocks.
How to manage linked lists is part of Programming 101 (Structured Programming), using a unique key to identify a set of structured data is part of Programming 201 (Relational Database Management and Design), and collision detection and avoidance of simultaneous updates on a specific chunk of data is part of Programming 301 (Parallel/Threaded Programming).
Just about every operating system and relational database management system makes congruent use of all three methods; here are just a few that come quickly to mind:
-- Microsoft Windows: Active Directory object record updates in combination with AD replication conflict resolution...
-- Just about any clustered SQL server solution provided by Microsoft, Oracle, IBM, Sybase, Pervasive, etc....
-- DNS record updates and scavenging...
-- 4ESS/5ESS telephone switching equipment state tables...
I am sure there are many more.
**Note: I am not against patents per se, as long as couple conditions are met:
-- Patent holders should be required to produce an actual product based upon their patents within X years of initial award (4 to 7 years to allow for commercial development seems fair; maybe a bit longer in the case of medicines that are stuck in double-blind-study-limbo by government drug safety authorities), otherwise the patent enters the public domain. ("Put up or shut up.")
-- Patents on software which provides general methods of manipulating data in various ways should be invalidated; after all, that's what general-purpose computers (and the apps that run on them) are designed to do. ("How I arrange my LEGOs is my business.")