Hi Ignite-users, We are looking for some advice on how best to approach a problem using Ignite. We’d be grateful if someone who has encountered this type of thing could share their experience.
We are looking for a recommended way of modelling a Parent->Child type relationship between caches such that any updates to the Parent or Child can be processed by a subscribing client in an atomic way. E.g. there is a parent object Company, with a List of Staff and a List of Customers. If fields from either of these tables are updated, we'd like the update in one go so client code can carry out logic on this consistent object graph. We have tried modelling this as a Company cache (all caches defined via DDL), with a separate Staff cache and a Customer cache. Staff and Customer have a foreign key which maps to the primary key of Company. We are now looking at executing reads and writes to this set of tables. Some requirements we have: 1. Users of the system (ideally) need to be able to edit each of these caches via an easy interface (E.g an SQL client). 2. A client subscribing to this data is notified upon changes as soon as they are made so a calculation can take place, involving entries from each of the N related tables. 3. Client code ultimately will operate on an instance of a java class Company, which has a List<Staff> and List<Customer> members so all three need to be tied together in a consistent way. (Client code will provide some sort of mapping function to convert from BinaryObject to internal domain objects). 4. The handling of this type of data structure is consistent between the initial query and continuous queries. Many thanks, Sham. -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
