Contributors

{"order":"name","contextEntityId":45300448}

Glossary

TermDefinition
AAI-internal-clients

Sub-projects of AAI that are direct clients of schema services (currently have dependency on aai-schema-ingest library) including:

  • data-router, gizmo, resources, router-core, sparky-be, spike, traversal, validation

Sub-projects of AAI that are indirect clients of schema services (via aai-common library) or are schema-agnostic including:

  • babel, cacher, chameleon, champ, eis, enricher, esr, gallifrey, graphadmin, graphgraph, logging-service, model-loader, rest-client, search-data-service, sparky-fe, tabular-data-service
ONAP-project-clients

Projects of ONAP that are definitely clients of AAI including:

Projects of ONAP that are likely clients of AAI including:

  • CLAMP, DMaaP, ExtAPI, Holmes, MSB, MultiVIM, OOM, Policy, Portal, SDC, SO, SDNC, VFC, VNFSDK, UUI
producer-consumer pairings

Apart from "internal" AAI data, the data in AAI is produced by a non-AAI component and then consumed by another non-AAI component. Therefore the schema definitions and intended meanings of each schema element are not controlled by AAI itself.

See examples in AAI Schema Producer-Consumer Pairings

deployment tools

Tools (e.g. Helm, ansible, Docker Compose, etc) used to:

  • place the microservices into their run-time environment
  • distribute data files for initial configuration
  • perform the initial configuration of microservices
  • get the ONAP system running as intended
IDL Files

Some form of "Interface Definition Language" which unifies and replaces the OXM files and EdgeRule files as definition of AAI schema

Purple highlightsHighlight change from predecessor ONAP Release

References

Use Case Outlines

ONAP Release TargetUse Case SummaryDeprecated Functions
Amsterdam, Beijing

AAI Schema is:

  • single-data-file configuration per version, i.e. one OXM file and one EdgeRule file
  • distributed via source-code repository to AAI-internal-clients and ONAP-project-clients
  • built into and deployed as microservice JAR file
  • effective only upon microservice startup

Casablanca

AAI Schema is:

  • multiple-data-file configuration, i.e. multiple OXM files and multiple EdgeRule files
  • distributed via source-code repository to AAI-internal-clients and ONAP-project-clients
  • built into and deployed as microservice JAR file
  • effective only upon microservice startup
  • single-data-file configuration per version
Dublin

AAI Schema is:

  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple OXM files and multiple EdgeRule files, without change to JAR files
    • effective only upon AAI-internal-clients startup
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple OXM files and multiple EdgeRule files
    • effective only upon AAI Schema Service startup
  • distributed via run-time API to AAI-internal-clients
    • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients, i.e. XSD files and generated POJOs
  • built into and deployed as ONAP-project-clients JAR file
    • effective only upon ONAP-project-clients startup
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
Dublin and beyondEach of the AAI-internal-clients and ONAP-project-clients will need to identify the relevant pieces of AAI schema that relates uniquely to their operations. The multiple-data-file configuration to be grouped by producer-consumer pairings.
Dublin and beyondEach of the AAI-internal-clients will need to decide when/how to proceed with changing over to the updated methods of distributing the AAI schema.
El Alto

AAI Schema is:

  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple OXM files and multiple EdgeRule files, without change to JAR files
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple OXM files and multiple EdgeRule files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time API to AAI-internal-clients
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to ONAP-project-clients without change to JAR files
    • effective only upon ONAP-project-clients startup
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
  • effective only upon AAI Schema Service startup
  • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients
  • XSD files and generated POJOs
El Alto and beyondEach of the ONAP-project-clients will need to decide when/how to proceed with changing over to the updated methods of distributing the AAI schema.
  • Needing to deal with schema change fragility (see JIRA for "UnrecognizedPropertyException" or "Unrecognized field" error messages)
Frankfurt

AAI Schema is:

  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple IDL files, without change to JAR files
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple IDL files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time SDC API to AAI Schema Service, i.e. multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time API to AAI-internal-clients
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to ONAP-project-clients without change to JAR files
    • effective only upon ONAP-project-clients startup
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
  • effective only upon AAI Schema Service startup
  • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients
  • XSD files and generated POJOs
  • multiple OXM files and multiple EdgeRule files
G release

AAI Schema is:

  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple IDL files or multiple TOSCA files and CSAR files, without change to JAR files
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple IDL files or multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time SDC API to AAI Schema Service, i.e. multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time API to AAI-internal-clients
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to ONAP-project-clients without change to JAR files
    • effective only upon ONAP-project-clients startup
  • distributed via run-time API to ONAP-project-clients
    • effective only upon ONAP-project-clients startup
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
  • effective only upon AAI Schema Service startup
  • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients
  • XSD files and generated POJOs
  • multiple OXM files and multiple EdgeRule files
G release and beyondEach of the ONAP-project-clients will need to decide when/how to proceed with changing over to the updated methods of distributing the AAI schema.
H release and beyond

AAI Schema is:

  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple IDL files or multiple TOSCA files and CSAR files, without change to JAR files
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple IDL files or multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time SDC API to AAI Schema Service, i.e. multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time API to AAI-internal-clients
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to ONAP-project-clients without change to JAR files
    • effective upon ONAP-project-clients run-time refresh
  • distributed via run-time API to ONAP-project-clients
    • effective upon ONAP-project-clients run-time refresh
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
  • effective only upon AAI Schema Service startup
  • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients
  • XSD files and generated POJOs
  • multiple OXM files and multiple EdgeRule files
  • effective only upon ONAP-project-clients startup
I release and beyond

AAI Schema is:

  • federation of AAI instances with disjoint/partially-overlapping schemas and databases
    • facade to stitch disjoint schema and data back into a unified representation
    • schema change and version upgrade by deploying new AAI instance to join federation
    • rolling upgrade and data migration between AAI instances within federation
    • support for multiple-sources-of-truth scenario: AAI-1343 - Getting issue details... STATUS
  • multiple-data-file configuration
  • distributed as data files via deployment tools to AAI-internal-clients, i.e. multiple IDL files or multiple TOSCA files and CSAR files, without change to JAR files
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to AAI Schema Service, i.e. multiple IDL files or multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time SDC API to AAI Schema Service, i.e. multiple TOSCA files and CSAR files
    • effective upon AAI Schema Service run-time refresh
  • distributed via run-time API to AAI-internal-clients
    • effective upon AAI-internal-clients run-time refresh
  • distributed as data files via deployment tools to ONAP-project-clients without change to JAR files
    • effective upon ONAP-project-clients run-time refresh
  • distributed via run-time API to ONAP-project-clients
    • effective upon ONAP-project-clients run-time refresh
  • single-data-file configuration per version
  • distributed via source-code repository to AAI-internal-clients
  • built into and deployed as microservice JAR file
  • effective only upon AAI Schema Service startup
  • effective only upon AAI-internal-clients startup
  • distributed via source-code repository to ONAP-project-clients
  • XSD files and generated POJOs
  • multiple OXM files and multiple EdgeRule files
  • effective only upon ONAP-project-clients startup
  • single unified AAI schema and database
J release and beyond
  • ???

  • No labels

6 Comments

  1. Keong Lim

    Pavel Paroulek, James Forsyth, Manisha Aggarwal, Venkata Harish Kajur, William Reehil, CT Paterson, Steven Blimkie, Robby Maharajh, please review, comment and update as needed.


  2. Keong Lim

    Does the SDC modelling include the equivalent of EdgeRule metadata?

    Update the AAI-internal-clients to list only the ones having schema-ingestor library as a dependency.

    • Probable design: upgrade the schema-ingestor to take input from files or REST API or something else.
    • Merge the two Dublin options into a single option that covers all variations of input via schema-ingestor
    • however, note that discussion of schema-ingestor design decisions seems premature, given that the use case direction has not even been confirmed

    Add an item for an IDL-abstraction layer, replace OXM/EdgeRule files, probably targeted to a release before using TOSCA files as input

    1. Keong Lim

      Updated.

  3. Keong Lim

    Chesla Wechsler appreciate if you could review and comment on this page to see how well aligned it is with your Service Instance thoughts.


    1. Chesla Wechsler

      Two initial things:  I really like how this page is laid out and presents an evolution path.  I also read the whole thread on the schema service to understand more about it. 

      My use case seems like it would leverage the schema service, particularly in the "TOSCA files and CSARs" which you have for a later release (but see caveat below). 

      Being able to drive AAI from SDC has always been a goal of mine.  CT Paterson and I are aligned that the best of all worlds would have been for ONAP to be a complete framework that interpreted the models and any models could be dropped in and the platform would just work.  I think we're far away from that and I'm not sure when ONAP would get there, or even if there is mind share to be that general.  That being said, I still believe that converting AAI from being defined as "the superset of all possible things SDC can design and the platform can instantiate" to AAI being driven by the individual models is where we should head.  I see the platform as having a need, for the foreseeable future, for some global types around which the platform knows how to operate (CT Paterson is sighing now).   Those global types get extended by specific models, and producers of data that fit the syntax and semantics (think AAI edges here) have to have a place to share the results for consumers (as well as for themselves if they need to reconstitute something at a later date). 

      I would want all platform components to have a shared knowledge, driven from the models, as to how to interact with AAI.  If this schema service is consuming the models and using them to define the valid information that can be made available through AAI, in a way we know is going to be quite dynamic, then I'm all for it. 

      Caveat:  I think the original intention of TOSCA files and CSARs may have been to construct TOSCA that matches the OXM of AAI into inventory types.  I think that continues the "superset" approach.  What I would want to see is what I portrayed in my slides, which is that each node type defined in a model becomes a vertex type in AAI which knows its inheritance hierarchy as well as its composition.    The inheritance hierarchy permits queries based on abstraction.

      Note:  I believe AAI has at least some logic to say "if version 2 of this API included 2 fields and version 3 included 3, a client doing a PUT using the version 2 API cannot affect the new field added with version 3".  It's possible logic like that may help the backwards compatibility issues identified on the call regarding the concern that some clients may migrate to the model-driven approach and others may not.

  4. Keong Lim

    Sharon Chisholm appreciate if you could review and comment on this page to see how well aligned it is with your visions of a future AAI.