Every form of data use requires decisions on suitable technologies. In this article, we would like to ask questions that are crucial for a technology decision.
Where do we start?
When a company decides to merge data centrally or make it available, professional and technical decisions and activities follow. The business planning was described in the last article; it is the structuring of the issues and data sources in accordance with the data strategy from a business perspective. On the other hand, the solution must be technically implemented and operated. This topic is dealt with in this article, but no best technology decision is presented. There can be no such thing for different companies and would be different with every release of the thousands of tools. We would like to present some, but by no means all Approaches and experiences show to a decision in a specific case to get closer. As the Technology decisions are never detached from the technical requirements of stakeholders questions can therefore already be taken into account during the collection of ideas with the stakeholders. In any case, the SThe requirements are determined openly and independently of certain solutions and buzzwords such as big data.
The Data strategy of a company provides a glimpse into the future and is therefore a Important influencing factor in the choice of technologyas this is where the framework conditions for collecting, providing and using the data are set. The question of whether only one part of the company or even just one project should prepare and provide data should be asked at an early stage. However, the approach can also be to gradually build up a database for an entire company and make it available to a large number of users, which is part of the data strategy and leads to different technological decisions.
What kind of data do I have and what do I want to do?
This very obvious question is, of course, fundamental and should have a decisive influence on the decision for a technology. The simplest and often initial case is exclusively Structured data from various sourcesthat are linked and displayed. A lean solution that is not necessarily among the hottest, most expensive or best-known solutions in the technology hypecycle can be useful here. - Even if increasingly buzzwords such as "big data" fall.
In our data project, we initially opted for a Postgres database as a much leaner data storage solution after evaluating the costs, benefits and vision in order to build up the expertise required for the real application of a big data solution. The data is processed according to the technical questions using KNIME.
Are we able and willing to operate the technical solution ourselves?
The question of who develops and operates the solution found is not just a question of cost. Companies whose business content is far removed from the IT sector may not have any expertise in the development and operation of data landscapes and may not want to acquire any. Cloud providers see their great advantage here and offer carefree infrastructure packages in various gradations on. In addition to possible data protection issues, the focus is on the expected costs, not least what the commitment to a provider means.
Away from the infrastructure, the Permanent maintenance and support of your own data processes However, this should not be forgotten. Adjustments will always be necessary, for example due to changes in input data formats that hinder further processing of the data.
How confidential are my data sources?
If it becomes clear when defining the use cases that (strictly) confidential data sources are to be included, it is worth taking a look at the law or company processes. Storing and processing data with different levels of confidentiality may have limits that need to be taken into account. For example, the GDPR stipulates, among other things, that personal data may only leave Europe with explicit consent.
If not every user group is allowed to see everything, there may also be requirements for data visualization. In our project Tableau Software which enables both row level security (the user group may only see the figures for its own area) and column level security (the user group may only see certain KPIs).
For our company reporting, we decided that the data should not be processed in the cloud. We therefore preferred an on-premise solution for both data storage and provision via Tableau Server.
Should there be self-service?
The use case definition may show that each user should only be able to see the data that is necessary for their daily work and no more. In this case, it is not only necessary to define a Role-rights concept The requirements for the technology must also be taken into account. The technology must enable the implementation of the planned role-rights concept in accordance with the usual company processes. Depending on where the Self-service confidentiality must be implemented right up to the accessible data source. This may mean, for example, that only aggregated or anonymized data sources can be made available to users or that the selected tool enables very detailed assignment of rights. The introduction of Data quality processes is not only essential for self-service.
What should the reports look like for the user and how should the user reach them?
In our data project, the Requirements for the reports often based on the previous manually created reports in Excel and PowerPoint. If the visualization requirements are very numerous and mandatory, a visualization tool must offer a wide range of functions. And even then, not every wish can be implemented. If you have very specific requirements for the creation and provision of reports, you can also use a Specially developed solution should be considered.
The requirements for the provision of reports to the user can change over time. For the doubleSlash data landscape, provision in the browser was a requirement from the outset, so we immediately opted for Tableau Server. This meant that some reports could be made available to all employees and the Transparency in the company can be significantly increased. The price of Tableau Server is essentially determined by the number of users. The cloud providers' pricing is heavily dependent on usage, which can be interesting for a very large user group with little actual usage.
Conclusion
There will never be one right decision for a technology chain for all companies and projects. Last but not least Flexibility The ability to replace, omit or add new parts. This makes it possible to adapt to changing requirements of all kinds. The rarest way to achieve success is to commit to a specific technology before determining the requirements. The possibilities for choosing a combination of technologies are almost endless. This article presents examples of questions whose Comparison with the requirements has a major impact on the choice of technologies has. The strengths and weaknesses of a technology then determine how requirements can be implemented.
