This text is a first draft. Further work is defered until it is clear for what (actual) purposes it will be used, so that we can make sure it is fit for such purposes.
According to ISO 42010, the core of any architecture description (of a "System of Interest") consists of a specification of (identified) stakeholders and their concerns, and a set of '(architecture) views' (that are governed by '(architecture) viewpoints') that address these concerns. A high quality architecture description will be complete in its list of stakeholders (or stakeholder roles) and their concerns. Also, it will have addressed each of them in one or more of its views.
In an SSI context, there typically is no single "System of Interest". Instead there are many such systems, run by different parties, that work together - i.e. exchange data (typically: (verifiable) credentials). In such contexts, individual parties can and will perform a variety of functions in various roles, the most prominent of which are issuer, holder, and validatorverifier.
The purpose of this document is
- to provide a (conceptual) framework that shows how individual "Systems of Interest" that various parties run can be conceived to work together, i.e. exchange data (typically: (verifiable) credentials) between them;
- to derive what this means in terms of capabilities that such systems need, and to highlight for each of them what this means in terms of
- design/define time: governance (and governance processes);
- operational/runtime: kinds of actions that (designated) employees of the parties will need to be capable of performing on their behalf;
- the creation and maintenance of the various policies for an individual capability, so that such employees will know how to operate; that is: when they actually execute such actions, they will do so in accordance with the intention of the party on whose behalf they do so.
Thus, this document can be used as a mental model that can help create and maintain coherence and consistence between governance architectures and technical architectures.
2. ContextSSI is all about the exchange of data, typically in the form of (verifiable) credentials (data that represents a set of claims, each of which is a statement about a particular entity (the subject of the claim)).
The context within which SSI is used consists of various parties, each of which has set its own objectives (its mission being the most prominent one). They have processes that seek to achieve the realization of these objectives. We only focus on the information processes (production processes are out of scope), i.e. processes in which various kinds of data are manipluated. In order to exchange data, information processes need a number of capabilities that enable them to:
- formulate requests (to other parties) for sending data, and receiving responses and obtaining such data that can easily be validated, i.e.: that it an agent that receives/obtains such data can easily determine whether or not it is valid (in the eyes of its principal) to be used for further processing and the purposes that such processing serves. We provisionally call this the 'usage capability'.
- offer data that it has to other parties, i.e. create and publish advertisements for such offerings, and means by which such other parties can request for such data and obtain it (all, of course, compliant with the policies that this party has specified). We provisionally call this the 'provisioning capability';
- store data, for later use. We provisionally call this the 'storage capability'.
Figure 1. Context - Information processes and Capabilities.
A party is said to have a capability if it can (operationally, i.e. at runtime) have particular things done according to the policies that specify the party-specific details of how that should be done, constraints that should be adhered to, etc. This implies that the party has (digital and/or human) employees that have been assigned tasks to do the associated work, and/or outsource such tasks.
The following subsections provide a draft description of the capabilities mentioned in the figure above.
3.1. Usage Capability
- find parties that can provide such data;
- formulate criteria (that are specific for each purpose) to distinguish between data that is (not) valid for the purpose(s) that the data will be processed for;
- provide the circumstances and/or satisfy the conditions for its (designated) employees so that they can request such data, obtain and verify data from responses to such requests, to validate the data for the purpose(s) the data was requested, and to process the (validated) data in the appropriate information processes.
This capability consists of
- runtime/operational parts, where agents request, verify, and validate such data (the further processing is done using capabilities of the information processes);
- a design-time part, in which the party determines for every information process when data is needed, what kinds of data that would be, what the verification criteria are, what the validation criteria are, and how this data is to be subsequently processed.
- a mapping part, in which such criteria are converted into policies for employees (that are tasked to do the operational work) can read and understand, and use to perform the verification, validation and other, related kinds of actions in the way as intended by the party on whose behalf they will be acting.
3.2. Storage Capability
The purpose of the storage capability is to ensure that a party can cache data (typically: credentials) in such a way that the confidentiality (privacy) and integrity of the stored data is guaranteed, and that only designated employees (either of that party itself, but also of other parties) are provided with rights and/or duties to access (read, write) that cache.
This capability consists of
- runtime/operational parts, such as
- a wallet (the actual store/cache), that is capable of servicing requests for storing, retrieving, modifying and/or deleting credentials (perhaps also other data) from the store/cache, according to the access control policy of its owner;
- (at least) one holder (i.e. an employee that is capable of communicating with other employees for the purpose of
- a design-time part, in which the party manages the mandates (including consents) for its own employees, as well as for employees of other parties.
3.3. Provisioning Capability
- create and subsequently issue credential of specific kinds upon request, and
- advertise the kinds of credentials that it is capable and willing to issue, in such a way that other parties can find these advertisements and use their contents to determine whether or not such credentials contain data that are valid for being processed by (or on behalf of) tht party for some specific purpose.
- deploy and run components, or outsource the deployment and running of such components this to other parties (which we call service providers), and provide them with the policies that enable them to fulfill their task, so that they - in our parlance - are onboarded and hence qualify as employees of the party. These services will then, as agents of this party, provide services such as (but not limited to):
- supplying credentials upon request;
- revoking credentials whenever (parts of) their contents is no longer considered to be true in the knowledge of the party;
- signing credentials on behalf of the party;
- presenting the party's offerings for the kinds of credentials that it is willing and able to issue (such that other parties can decide whether or not the contents of such credentials are valid for one or more of its purposes);
- deploy and run (design-time) processes, in which it
- constructs the advertisements for the kinds of credentials it will be issuing;
- establishes the criteria that the components that it has running (and act on its behalf), must use to determine whether or not to service requests that they receive;
- decides what related services to use (such as services for revocation3), and
- a mapping part, in which
- the service request criteria are converted into policies that employees (that are tasked to do the operational work on the party's behalf) can read and understand, and use to decide whether or not to service issuing requests in the way as intended by the party on whose behalf they act.
- additional (provisioning-related) services are
- it is determined at what endpoint(s) requests for supplying such credentials will be serviced;
Provisioning also includes the issuing of credentials that contain claims that mandate4 specific agents (the claim's subjects) for specific kinds of access (read, update, write, ...) to functions that are provided by designated agents of the party with that provisioning capability.
4.4. Revocation Capability
- to ensure that that credentials that have been issued on behalf of the party that runs this component, are revoked whenever a situation is detected (in some information process) that shows that one or more of the claims in the credential are no longer true (in the knowledge of that party)
- parties that intend to use (data from) a credential that has been issued by the party that has the revocation capability, are enabled to learn whether or not that (data from) that credential has been revoked.
This capability consists of
- runtime/operational parts,
- a design-time part, in which the party maintains criteria for signalling credentials to be revoked, and determines which kind(s) of revocation mechanisms should be used for which kind(s) of credentials.
- a mapping part, in which the party maintains the set of policies that employees (that are tasked to do the operational work on the party's behalf) can read and understand, and use to decide whether or not to revoke credentials, and do the actual revocation thereof.
4. Functional Components
While parties typically have all these capabilities, they each will have their own, subjective expectations regarding the particular ways in which such capabilities should operate. Such expectations must be made explicit, e.g. by means of policies that employees that are mandated to perform associated actions can read and interpret such that they execute these actions according to the intentions of the party on whose behalf they do so.
The following figure shows how parties can furnish the aforementioned capabilities in terms of some generic functional components.
Figure 1. Context - Information processes and Capabilities.
- a wallet component, that holds credentials (and perhaps other data), has a simple interface for doing CRUD operations (the basic operations on data stores), and a mechanism to ensure that requests will only be serviced if the requesting component presents a proper mandate (permission, consent, …) for the requested operation.
- a holder component, that can interact with issuer and validator components for exchanging credentials)
Note that every component in the figure has two parts:
- the IT that is running on some computer. Note that this IT may be controlled by a party (typically called 'service provider') that is not the party on whose behalf the IT will be functioning.
- a policy, authored by the party on whose behalf the component functions, where the policy specifies the party-specific ways of its functioning. In the Wallet example, it would contain e.g., the access control criteria for deciding whether or not to service the various requests.
What the figure does not show is that a mapping or translation activity may be called for to translate the 'business speak' that states the contents of the policies into 'employee-specific speak', i.e. data/texts that employees (that are tasked to execute the associated operational actions), can read and interpret such that when they use them to execute such actions on behalf of the party that authored the (business) policies, they will do so in accordance with the intensions of that party. Since employees can be human, non-human, and typically have different capabilities for understanding policy-data, a single business-policy might need to be translated into multiple operational policies.
- The SSI community typically talks about verification (and the associated role of verifier), and ignores validation. However, since data is exchanged so that the party that receives presentations can actually use that data for the specific purpose(s) that the presentations are requested. This not only means that such data must be verifiable (and verified), but particularly that it must also be validatable, and only used when it turns out to be valid for further processing.↩
- every such holder needs to be appropriately mandated; this would typically be part of the onboarding process of that holder.↩
- The party may want to also implement additional services that are related to the provisioning of credentials, such as services for↩
- A mandate for acces to the party's Credential Store can also be seen as a consent of that party to the mandatee (agent)↩