CCSDK-2872 DP: Agree and document proposed long-term Persistence SPI(s)

Jira Ticket:

CPS-37 - Getting issue details... STATUS

Decisions/Open Issues

Decoupling

It is very important that the SPI doesn't expose any database implementation for example database ID's

	Description	Details	Decisions
1	Assume flat attributes (no complex data types) but data could be stored as json object in SPI impl. for convenience/PoC
2	SPI Implementation NOT using Model (service) current ENM SPI does!
3	Are we going to have XPath/query builder?	We should decide based on what the users want to use/is easy to use We don't want a query builder that eventually mimics XPaths functionality, use what is defined already XPATH as a based but maybe supplement with a xpath-builder
4	When do we get attributes?	How do control how many attributes we get back?	We will have an (optional) output specification with the following options Default is no attributes All Attributes Specify attributes - output specification (filter)
5	Should we use the name node or fragment?		We decided on the name DataNode
6	Do we want to split the interface?	Maybe we could just have a 3 Alternatives all in one interface Limited number of interfaces e.g Data Objects/ Data Persistence/Query Very fine grained (see below)	DataStore SPI - methods to add data. e.g dataStore.addChild(parentDataObject, childDataObject) CpsAdmin Interface will handle the dataspaces and anchors Module SPI - responsible for handling modules and module sets The above will be in separate maven modules. Any data object they need will be in the same module. We will not have a querybuilder in SPI but the SPI will have to understand the cpsPath. The java API will use a builder pattern to create a cpsPath
7	Do we need objects for very simple concept like dataspaceName which are just a string?		We will not use objects for name fields. We will just use Strings. We may need an object for AnchorName
8	Should we share common objects between the Java API and SPI?	Alternatives API depends on model objects defined in SPI SPI depends on model objects defined in API Extract common model API in separate module (deliverable) Use separate models with slightly different names e.g Anchor in API and AnchorDetails in SPI (CPService will transfer when needed)

ENM SPI Study

Details about the ENM SPI can be found here: ENM Study

Proposed Future Implementation of CPS SPI

DataStore SPI

	Name	Capabilities
1	DataStoreService	DataNode getDataNode(cpsPath) Collection<DataNode> queryDataNodes(cpsPathQuery) DataNode getDataNode(cpsPath,outputSpecification) int count(cpsPath) Void addDataNode(anchorName,dataNode) Void addChild(cpsPathParent,dataNode) Void addChildren(cpsPathParent,dataNode...) Void updateDataNode(anchorName,dataNode) Void setAttribute(cpsPath,attributeName,value) Void deleteDataNode(cpsPath)
2	DataNode Object	DataNode createDataNode(DataNode) String getCpsPath() DataNode getAttributes() DataNode getAttribute(name) Void setAttributes() Void setAttribute(name) Module getModule(moduleName)

CpsAdmin SPI

	Name	Capabilities
1	DataspacePersistenceService	Void createDataspace(dataspaceName) Void getDataspace(dataspaceName) Void getDataspaces() boolean exists(dataspaceName)
2	AnchorPersistenceService	Void createAnchor(dataspaceName, anchorName) Anchor getAnchors() Anchor getAnchor(anchorName) Void deleteAnchor(anchorName) Void associateAnchor(anchorName, moduleSetName) Void associateAnchor(anchorName, moduleRef) Void addDataNode(anchorName, dataNode)
3	~~Dataspace (object)~~	~~Void createDataspace(dataspaceName)~~ ~~String getName(Dataspace)~~
4	~~Anchor Object~~	~~Void createAnchor(anchorName)~~ ~~String getName(Anchor)~~

Module SPI

	Name	Capabilities
1	ModulePersistenceService	YangModule getModule(moduleRef) Collection<YangModule> getModules(moduleSetName)
2	YangModule	ModuleRef getModuleRef(moduleRef) String getModuleContent(moduleRef)
3	ModuleRef	String getNamespace(moduleRef) String getRevision(moduleRef)
4	ModuleSet	String getName(moduleRef) Collection<ModuleRef> getModuleRefs()