BDsafe's focus is on the rapid storage & retrieval of event data i.e. data which represents historical events.
Some examples are,

  1. Telephone call CDRs,
  2. Web traffic records (IPDRs),
  3. Network Alarms,
  4. Log Files,
  5. Financial transactions/Utility billing records,
  6. Machine generated records (sensors, meters).

This focus allows BDsafe to act in one of two roles within a system architecture:

  • Data Distribution Engine:
    BDsafe provides long term storage for source data, distributes scheduled data feeds to northbound client system(s) & provides the capability to request either a later re-send of the data or to make ad hoc requests for specific data
  • Backup Store:
    BDsafe provides a Backup facility at low operational costs and no break in service for data coming in from outside a business.

BDsafe's low cost high data load/retrieval speeds & retention abilities are made possible by its focus solely on read-only data i.e. any form of event data.

  • Low Costs result from the "Frugal Engineering" approach which was applied to the original Datastore design:- a "back to basics" approach to the design , as opposed to just fixing or plugging gaps in an existing model
  • High Load/Retrieval Speeds result from DT2's use of "Inverted File Indexing" techniques - which avoids processing overheads associated with most other data management approaches
  • Long Term Storage is enabled via DT2's "Simple Virtual Filestore" technique which makes efficient use of physical resources.

The diagram below illustrates the type of data which BDsafe is designed to handle:

WIP - trying to
              find out how

The diagram illustrates that BDsafe's full emphasis is on Write-Once Read-Only data - this is the key to its very fast storage & retrieval capabilities.

The basic premise of the BDsafe approach is that data can be classified into three main categories:
  1. Data which supports a complex enterprise data model for which a relational model is often used
  2. Data which has been normalised for specific business needs e.g. reports & dashboards
  3. Data which represents the original business events: telephone calls or data volume usage.
Category 1 data is a representation of the business and therefore has to comply with strictly defined models.

Category 2 data has to be in a common format largely defined by the reporting requirements. This may only require a subset of the original data.

Category 3 data should be in a form as close to the original data as possible, with the emphasis being retrievability of the original data.

Fundamental to the design of BDsafe is the view that optimal performance is gained for event data by focusing solely on providing a robust platform for the data management of event data. This platform then offers the potential for a central event data management platform to support the sourcing of data to a multitude of enterpise systems.

WIP - trying to
              find out how