Data Stream Management Systems (DMS) represent a fast-growing
research area. Most of current DMS projects take the approach
of using database query languages to process the fast-incoming
data streams. This approach can simplify the writing of applications,
inasmuch as these often need to query stored databases besides
streaming data.  However, database query languages, and SQL in particular,
encounter severe problems with continuous queries on data 
streams---after all,
they were designed for transient queries on persistent data, and not for
persistent queries on streaming data!

In this talk, I will focus on these problems and show that they compromise
the effectiveness of the DMS prototypes developed by most of  current
research projects. However, the Stream Mill system developed at UCLA
demonstrates a simple and general solution to the continuous query
language problem, and also supports mining queries and time series queries.