Are We Ready For Learned Cardinality Estimation?

A Research Proposal on Learned Cardinality Estimation
Student Name
Institutional Affiliation
Instructor
Date
Problem statement
 Are we ready for learned Cardinality Estimation? This topic is essential, especially in conceptualizing query optimization. Studies such as “ Myrocks: Lsm-Tree Database Storage Engine Serving Facebook's Social Graph” by Matsunobu et al.  (2020) and “Hadoopdb: An Architectural Hybrid Of Mapreduce And Dbms Technologies For Analytical Workloads” by Abouzeidet al. (2009) reveal that learned models portray the capability of replacing ordinary estimators.
Research Objectives. 
The primary objective of this study is to examine if technology users are ready to Leaned Cardianality models. The study will address this topic in three sections. Firstly, the focus will shift to understanding static environments. The latter observes research elements without data updates (Abouzeid et al., 2009). At this stage, a comparison is drawn between newly learned methods and traditional ones. In comparison to “ Myrocks: Lsm-Tree Database Storage Engine Serving Facebook's Social Graph” by Matsunobu et al.  (2020), this process will rely on real-world databases with unified settings. Essentially, the primary purpose of this stage will determine if the results are accurate in comparison to traditional setups. On the same note, the experiment also seeks to define how challenges such as inference and training costs impact modern strategies.
Secondly, the research will determine if learned models can function in dynamic environments. The latter describes circumstances that involve frequent data updates. While addressing this objective, the process will involve previous studies such as Hadoopdb: An Architectural Hybrid Of Mapreduce And Dbms Technologies For Analytical Workloads” by Abouzeid et al. (2009) to define why learned find it challenging to incorporate data updates. Scholars argue that when these models face the latter, they return significant errors.
Thirdly, the research will focus on understanding the functionality of learned models and explore circumstances that make them go wrong (Woltmann et al., 2019). This course action aims to acquire results that showcase how the performance of this strategy relates to modifications in domain size, skewness, or correlation. Contrary to Hadoopdb: An Architectural Hybrid Of Mapreduce And Dbms Technologies For Analytical Workloads” by Abouzeidet al. (2009), portraying the relationship will highlight the reasons behind unpredictability and challenging interpretation. At this point of research, it will be easier to identify promising directions. For example, the outcome will ease the process of controlling the cost of learned models and their trustworthiness (Wang et al., 2020). On the same note, these objectives provide a clear future outlook in terms of research opportunities. Collectively, this research aims to provide extensive insight into the topic and offer a guide to practitioners and researchers. The primary objectives are fundamental in transforming learned cardinality estimators into functional elements of database systems.
 Research Methodology.
To undertake the objectives, the...