多模数据库.pdf
免费下载
Multi-Model Database Management
Systems - a Look Forward
Zhen Hua Liu
1
, Jiaheng Lu
2
, Dieter Gawlick
1
, Heli Helskyaho
2,3
Gregory Pogossiants
4
, Zhe Wu
1
1
Oracle Corporation
2
University of Helsinki
3
Miracle Finland Oy
4
Soulmates.ai
Abstract. The existence of the variety of data models and their associated data processing
technologies make data management extremely complex. In this paper, we envision a single
Multi-Model DataBase Management Systems (MMDBMS) providing declarative accesses to a
variety of data models. We briefly review the history of the evolution of the DBMS technology
to derive requirements of MMDBMSs and then we illustrate our ideas of building MMDBMSs
satisfying those requirements. Since the relational algebra is not powerful enough to provide a
mathematical foundation for MMDBMSs, we promote the category theory as a new theoretical
foundation, which is a generalization of the set theory. We also suggest a set of shared data
infrastructure services among data models to support “Just-In-Time” multi-model data access
autonomously.
1 INTRODUCTION – why MMDBMS?
Here is a short history of databases: Initially, database management systems sup-
ported the hierarchical and the network model (e.g., IBM’s IMS and GE’s IDS re-
spectively). These databases evolved very fast and developed the core infrastructures,
such as journaling, transactions, locking, 2PC (group and fast commit), recovery,
restart, fault tolerance, high performance, TP-monitors, messaging, main storage da-
tabases, and much, much more. We still use these concepts today. In the 80’ and 90’,
these databases were widely replaced by the relational database management systems
(RDBMS). The main argument is its solid theoretical foundation: set based relational
data model and declarative query language (SQL) based on abstract algebra over set
processing.
However, the demands to simplify the interaction between applications and databases
with simple storage and querying interfaces are not always possible using only the
relational model. Object databases ODBMS (Object Database Management Systems)
filled this gap by providing easy access to objects with object-oriented programming
languages. With additional OO features in RDBMSs, ORDBMSs are able to support
many domain data types, such as text, spatial, and images data. Interestingly, the last
decade has witnessed the re-emergence of hierarchical data models in the form of
XML and JSON data and the re-emergence of the network data model in the form of
RDF semantic graph and property graph data. This has led to native XML, JSON,
graph database systems and ORDBMSs providing XML, JSON, RDF and graph data
support via SQL/XML, SQL/JSON standards and ongoing standard development to
provide graph access via SQL. More applications are adopting graph modelling and
graph query since graphs provide a flexible way to structure application data and
2
adapt them dynamically to changes [15, 27, 29]. The source of graph data could come
from relational, XML, or JSON that exist in the different databases.
The history of the database evolution has shown that new applications often require
new data models leading to extended infrastructure of DBMSs with new query lan-
guages over these new data models. One existing solution is the polyglot persistency
approach, which leverages numerous DBMSs to support different data models and
integrates them programmatically at the application layer. The biggest issue of poly-
glot persistency is that the combined DBMSs is neither declarative nor unified. It
leaves database application to procedurally join data among multiple data models and
manually transform among data model instances. Instead of putting the burden on
applications, it is more desirable to have a unified single DBMS, which hides the
complexity of multiple data models by providing declarative approach of querying
multi-model data instances and just-in-time data model transformation.
In this position paper, we advocate a multi-model database management system
(MMDBMS) that has the ability to incorporate any data model and allows users to
manipulate all data models declaratively. Users are able to explore the real power of
an MMDBMS by leveraging its ability to autonomously transform data from one data
model to another. MMDBMSs allow data providers and data consumers to look at the
same data using different models depending on their most effective view.
MMDBMSs accomplish these data model transformation autonomously on behalf of
users.
We argue that the design of a full-fledged MMDBMS requires a more powerful math-
ematical foundation. The last few decades have witnessed a tremendous success of
RDBMSs leveraging the relational algebra as theoretical foundation and therefore
limiting this foundation to relational data. We recognize the same data can be repre-
sented relationally, hierarchically, graphically and are thus queryable by SQL,
XQuery, Property-Graph Query Language respectively. Therefore, we feel the need of
having a new theoretical foundation to provide transparent data model and query lan-
guage transformations among those data models and languages. In other words,
MMDBMSs require a powerful mathematical foundation to reason about declarative
data model transformation among multiple data models. In this paper, we promote
category theory [5,14] shall be able to play the role of the new mathematical founda-
tion to reason declarative construction and transformations among various data mod-
els.
In addition, this paper describes a set of shared data infrastructure services. The
shared services not only include essential common data services, such as transaction,
recovery, security, high availability but also include integrating artificial intelligence
to provide “Just-In-Time” data model access and telemetry service to promote multi-
model situation awareness service [4].
Organization The remainder of this paper is structured as follows: Section 2 intro-
duces the preliminaries on categories and examples of model transformation. Section
3 presents category theory as the mathematical foundation for MMDBMS. Section 4
illustrates MMDBMS infrastructure services. Section 5 shows related work and sec-
tion 6 concludes the paper.
of 15
免费下载
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文档的来源（墨天轮），文档链接，文档作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
下载排行榜
评论