There are three primary and distinct roles to consider, whether
you're building or buying DaaS - regardless of the type or
characteristics of data that's being exchanged; big data, open data,
fast data, IoT/IoE data, metadata, microdata, multimedia content,
structured, non-structured, semi-structured...ALL DATA.
The DaaS Consumer - who needs not only to acquire data from
somewhere (in a way that shields them from the underlying technology
concerns), but also then may use it to develop information apps and
services, or repackage the data to share further with others. The
consumer assigns and realizes value from the service.
The DaaS Provider - who actually builds, markets and
operates the business service and categorized storefront (or catalog),
and brokers or stewards the data quality & availability, data
rights, licenses and usage agreements between the consumers and the
original data owners. The provider creates, shapes and deploys the
opportunities for value-enablement of specific data assets.
IT Services Management - who design, implement and operate
the information and data management infrastructure the DaaS Provider
relies upon - and manage the IT component and services portfolio this
infrastructure includes. For example the databases, virtualization
technologies, data access services, storage and middleware capabilities.
(Note that "IT Services Management" may be a wholly 3rd-party role, as
well as a role within the DaaS Consumer or Provider organizations -
there may be 3 or more IT Services Management domains).
There's also a less distinct, more broadly relevant role - the DaaS Enabler.
a.k.a. the "Enterprise Architect", which can be a person, a role, or an
organizational capability. The EA scope includes a heavy focus on
enterprise "universal" information management and governance, infused
(particularly in the Public Sector) with the currently vogue
philosophies of SOA, Open Data, Mobility, Privacy-by-Design (PbD)
and Cloud Computing. (Note that DaaS does not have to be delivered via a
"cloud" deployment model - it's equally-applicable delivered as a
private data services virtualization platform, for example).