User interface and API design for Streaming Queries (continued from previous post ...)

 The Proposal  

Query Interface is not about keywords. 

When it comes to a user interface for querying data, people often associate Google user interface with search terms. Avoid Google. This is not necessarily the right interface for Streaming queries. If anything, it is simplistic, and keywords based. It is not operators-based. Splunk interface is a much cleaner interface since it has seasoned from search over machine data. Streaming queries are not table queries. They also don’t operate on big data on the Hadoop file system only. They do have standard operators that can directly translate to operator keywords and can be chained the same manner in a java file as pipe operators on the search bar. These translations of query operator expressions on the search bar to a java language FLink streaming query is a patentable design for streaming user interface and one that will prove immensely flexible for direct execution of user queries as compared to package queries. 

Some truths about user interface development: 

1) User Interface development estimates slip too often. It is not necessarily due to user interface test automation frameworks such as Selenium. Instead, it is caused by the user interface becoming monolithic and clunky. It takes on a large maintenance tail as different browsers and timing issues must be handled or special-cased. 

2) With the number of resources working on user interface increasing, there is a machine efficiency curve generated to user interface library development.  

 

 

3)  User interface should be more like a library rather than an all-inclusive customized application. It should easily switch as a layer over one data source to another or both 

4) It may be tempting to copy the existing solution from competitors such as 

Splunk 

SQL 

Tableau 

Machine Learning User Interface 

But it is far more important to streamline the translation of search expressions in plain English on the Search bar to Java Flink queries that can run in a fast serverless computing manner. 

5) If a wire diagram becomes more appealing and the go-to resource for all problem solving, it will likely not materialize. Instead, the comfort and pain of using the existing technologies for improving the user experience and usability must be honed. 

6) Plug-n-play plugin JavaScript framework for Sharepoint is a great start for standard query operators. The essence of stream querying and the continuous availability of results and their paging is something that will not be found elsewhere and should be salient for a user interface over streams. 

7) Security is not a functionality of the user interface. It is a non-functional aspect and should not require scattered checks all over the place but rather at the admission of control. 

8) User interface is great to build on REST interface and with thin clients, but the functionality should be composable and inheritable  

9) Separation of concerns especially between dashboard elements and the logic pertaining to one or more search queries will tremendously improve the offerings to the customer.