In higher education and the public sector, the drive to moving IT infrastructure and applications into the cloud is accelerating at a rapid pace. Vendors are providing various solutions for Software as a Service (SaaS), Platform as a Service (PaaS), Infrastructure as a Service (IaaS), and even Desktop as a Service (DaaS or VDI – Virtual Desktop Infrastructure). All of these platforms offer great promise for reducing costs and increasing productivity, but there are many questions that arise about the practicality and viability of these approaches in the context of a diverse, heterogeneous applications environment.
If one is to assume using data for analysis and informed decision making is a strategic asset to the organization, then several important issues arise regarding access to the data and the ability to integrate it for reporting and analysis across the enterprise. Many organizations seem to be rushing headlong into the cloud without considering some very important questions. The answers to these questions can significantly impact the ability to leverage your own data for decision making.
Here are a few to consider and should be addressed by vendors when considering cloud versus on premise solutions:
- Do you really have complete and total access to the entirety of your data? Often vendors will answer “yes” to an RFP requirement when in fact they only provide interfaces that give “one record at a time” query, or limited access to some of the data but not all. This is entirely inadequate to support a comprehensive data warehouse or integrated data environment that typically needs access to all new or changed data regardless of time frame.
- In what format is data access or output provided?
Query access via an API that can export multiple records? Flat file text extract? Excel export? Third party ETL tools? The ideal scenario is to have API query access or the ability to use third party data integration tools so that you can bulk load data into a data warehouse or other analytics platform in near real time. Flat file and excel may be acceptable depending on the latency and scope of access, but still may introduce additional size and performance limitations that can undermine successfully integrating data.
- Is there latency between requesting a data set and receiving it? For example, some cloud multi-tenant providers cannot easily provide an extract of one client’s data from the system and must do so as a manual extract that can take days to produce. This latency can significantly affect the frequency, accuracy, and usefulness of data integrated for analysis purposes.
- Does the cloud implementation allow for secure, seamless, direct access back to on premise applications and databases or other cloud applications that may be in a different cloud provider? Not all SaaS applications are available in the same infrastructure or zones which can present problems for communicating across applications and the data integration platform that may be used for an on premise or hosted data warehouse.
- Is it possible to modify the application (and resulting physical data model) to include additional data and tracking of metrics that are specific to the requirements and management strategies of the organization? The issue of flexibility and modification is a particular concern in multi-tenant SaaS solutions which generally do not allow significant modification, if at all, thus limiting the ability to capture important data at its entry points. This limits the ability to provide important metrics or data to predictive models that can provide essential insights to business processes.
There are real benefits of the cloud: tax and cash flow benefits by leveraging operational budgets for software and hardware, easier software updates, reduction in IT infrastructure to manage, more robust security, and even environmental benefits. However, the fact remains that moving to the cloud does not necessarily simplify the technology landscape. Instead it gets significantly more complex since some of the control and choices are taken away. Any consideration of cloud computing must be thought through carefully, particularly with respect to access to data for comprehensive reporting and analytics.