=Paper=
{{Paper
|id=Vol-1670/paper-14
|storemode=property
|title=None
|pdfUrl=https://ceur-ws.org/Vol-1670/paper-14.pdf
|volume=Vol-1670
}}
==None==
<pdf width="1500px">https://ceur-ws.org/Vol-1670/paper-14.pdf</pdf>
<pre>
      Challenges with Continuous Deployment
      of NoSQL-backed Database Applications

               Meike Klettke1 , Stefanie Scherzinger2 , Uta Störl3 ,
                 Stephanie Sombach2 , and Katharina Wiech2
                          1
                         University of Rostock, Germany
                              2
                           OTH Regensburg, Germany
               3
                 Darmstadt University of Applied Sciences, Germany

    We address a practical challenge with the continuous deployment of database
applications, which actually constitutes a data integration problem: Upon a new
deployment of the application code, entities already persisted in the production
database no longer match what the application code expects. Apart from mi-
grating all legacy entities eagerly at the time of the release, lazy migration is an
alternative popular with NoSQL data stores: A schema-flexible database stores
entities with legacy structure, as well as up-to-date entities. When a legacy en-
tity is loaded into the application, all pending structural changes are applied.
Thus, from the viewpoint of the application, entities are always up-to-date.
    Yet lazily migrating legacy data from several releases back, involving more
than one entity at-a-time, is not a trivial task. At LWA 2015 [3], we presented
our vision of a schema management unit for NoSQL data stores that carries
out schema evolution lazily: This involves an internal, Datalog-based model for
reading, writing, and migrating data [2]. However, we use Datalog not only to
specify the semantics of schema evolution operations, but Datalog is our actual
vehicle for carrying out data migrations: In this overview talk, we introduce
Datalution [1], a tool that alternatively evaluates our Datalog rules bottom-up
(for eager data migration) or top-down (for lazy data migration). In particular,
our tool allows for an easy comparison of both approaches in terms of the number
of physical writes to the data store.
    We demonstrate Datalution, provide insight into its mechanics, and outline
our next steps in integrating the Datalution engine with an industrial-strength
NoSQL data store.

Keywords: Schema evolution, NoSQL data stores, Datalog

References
1. Scherzinger, S., Sombach, S., Wiech, K., Klettke, M., Störl, U.: Datalution: A Tool
   for Continuous Schema Evolution in NoSQL-backed Web Applications. In: Proceed-
   ings QUDOS’16 (2016), Tool Demo.
2. Scherzinger, S., Störl, U., Klettke, M.: A Datalog-based Protocol for Lazy Data Mi-
   gration in Agile NoSQL Application Development. In: Proceedings DBPL’15 (2015)
3. Störl, U., Klettke, M., Scherzinger, S.: Kontrolliertes Schema-Evolutions-
   management für NoSQL-Datenbanksysteme. In: Proceedings LWA 2015 Workshops:
   KDML, FGWM, IR, and FGDB. (2015)

</pre>