Transaction control

Skyve implements consistent transaction control pervasively and provides a Persistence utility class for specific transaction management where this may be required.

Within the context of user interactions, Skyve supports n-level zoom contained within a single transaction. The context for the transaction is the document instance at the lowest zoom level (usually from the list view), i.e. the starting document. Skyve automatically modifies default actions depending on the zoom level or transaction context.

Concurrent conversations and transactional demarcation

Skyve supports conversation level transaction demarcation, meaning that the user can have any number of concurrent browser conversations.

Conflicting actions between conversations (e.g. two attempts to update the same record from within different browser windows, or by different users) are handled within the conversation context, and users will be notified if the action they are performing conflicts within an action from another conversation. The usual rule of first-in-wins applies; the first conversation to complete a transaction and commit will be successful, and any subsequent concurrent transaction involving the same record will be notified that their attempts to update the same record are no longer valid.

Skyve application databases persist two columns to manage concurrent conversation and optimistic locking protocols:

  • bizLock, and
  • bizVersion

The bizLock and bizVersion columns exists for each tuple. BizLock records the time of the last successful transaction and the user principal responsible for that transaction. BizVersion is compared by Skyve against the value of the version at the time when the record was accessed within the current user gesture. If the version value in memory does not match the persisted version value Skyve determines that another conversation has modified the record, and therefore the current conversation cannot commit without loss of information.

Authorisation model - summary of options and discussion

There are three common authorisation models:

  • Impersonation model
  • Trusted subsystem model
  • Hybrid model

Impersonation model

The impersonation model offers the most granular authorisation, however “the approach suffers from poor application scalability because effective connection pooling for database access is not possible.” (Refer to MSDN Library - Improving Web Application Security: Threats and Countermeasures, Chapter 4 Design Guidelines for Secure Web Applications, extracted from http://msdn.microsoft.com/en-us/library/ff649874.aspx on 2013-09-17)

The impersonation model is shown below.

Impersonation model

Skyve does not adopt this model for a number of reasons:

  • This model does not suit agile development methods where
    • user access requirements are not well known
    • identification of the end user set is unclear, and
    • an overhead exists in managing authorisation levels in multiple places,
  • The method does not scale well (as per the Microsoft quote above),

  • This model is not feasible for public web applications and SaaS environments where user populations are constantly changing and not known in advance, and

  • This model requires a large overhead in terms of managing implementation specific database permissions and access which does not confom to the Skyve principle of portability.

Trusted subsystem model

This model uses the application’s process identity to access the persistence layer. This model was selected for Skyve because:

  • Database connection pooling is supported which means that the application is scalable,

  • This model has the least overhead in terms of permissions management making it the most effective for agile development where requirements and the user population is not clearly defined, and

  • This model is widely used and therefore assumed to be acceptable in an otherwise unknown policy and strategy environment.

Trusted subsystem model

Hybrid model

Under the hybrid model, a combination of both of the above models is implemented. Skyve does not adopt this method natively because:

  • According (for example) to Microsoft - “creating multiple thread access tokens used to establish different security contexts for downstream resource access using Windows authentication is a privileged operation that requires privileged process accounts. This is counter to the principle of least privilege.”, and

  • User access requirements for SaaS are not clearly identified so decisions about the specifics of the Trusted service access was not able to be ascertained.

Hybrid model

Additionally, Skyve supports sophisticated declarative group, row and column level security modelling and enforces these declarations pervasively and consistently throughout any Skyve application.

⬆ back to top


Next Exception Handling
Previous Concepts