SQL and Temporal Database Research: Unified Review and Future Directions

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 09 | Sep -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1160
SQL and Temporal Database Research: Unified Review and Future
Directions
Rose-Mary Owusuaa Mensah1, Vincent Amankona2
1Postgraduate Student, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
2 Postgraduate Student, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Several attempts to incorporate temporal
extensions into the Structured Query Language, SQL, one of
the most popular query languages for databases date back
to the nineteenth and twentieth century. Although a lot of
work and research has been done on temporal databases
and SQL, there exist very limited literature clearly outlining
the various events which have taken place with regards to
temporal extensions of SQL over the years till the present
state in a concise document. Consequently, researchers need
to gather several pieces of literature before they can obtain
a vivid pictorial timeline of the history and the current state
of these temporal extensions for research and software
development purposes. In order to align current research
with that from the past, there is the need for a consolidation
of information. In this paper, we seek to present the past and
current state of temporal support in relational databases
and also describe the various attempts and proposals to
introduce temporal extensions into SQL over the years,
discuss their level of success and impact on the most recent
SQL standard. We also identify desirable temporal features
which have not yet been introduced into any of the SQL
standards and also determine the level of acceptance of the
available SQL temporal extensions by some commercial
database management system vendors.
Key Words: SQL, temporal, databases, SQL:2011, DBMS,
TSQL2,SQL-92
1. INTRODUCTION
Different users and applications have diverse
requirements and as such wish to access data at different
points in time; this requires a form of database
management system which is able to keep track of past,
current and future states of data. Many computer
applications, such as banking, inventory control, law,
medical records and airline reservations deal with time
varying information: hence it is important that database
management systems, DBMSs, are designed to be able to
handle the temporal information. A database which
provides in-built support for time is referred to as a
temporal database. Researchers have engaged in the study
of temporal concepts in databases over the years and have
written various scientific papers, articles and books in this
area.
In the early 1990s, a renowned researcher in the database
community, Richard Snodgrass, proposed that temporal
extensions to SQL be developed because few of them
existed. In response to his proposal [1], a committee was
formed to consolidate past research and suggestions from
the research community in order to design temporal
extensions to the 1992 edition of the SQL standard. Those
extensions, known as TSQL2, were then developed by this
committee and in 1993, they presented proposals to the
ANSI SQL Technical Committee. Based on responses to the
proposals, changes were made to the Language, and the
definitive version of the TSQL2 Language Specification
was published as a technical report in September, 1994
[2].
Almost five years later, a new standard SQL:1999 [3] was
to be released and attempts were made to include a
substandard, SQL/Temporal [4], which comprised parts of
the TSQL2 specification. SQL/Temporal failed to make it to
the SQL:1999 standard because the TSQL2 approach was
heavily criticized. After the unsuccessful attempt,
researchers continued to propose features to extend the
SQL syntax for temporal support. A decade later, a new
standard SQL:2011[5] has been developed, which
provides a substantive support for temporal data
management by introducing system-versioned tables,
application-time period tables and bitemporal tables.
Although a lot of work and research has been done on
temporal databases, there exists very limited literature
which clearly outlines the various events which have
taken place with regards to temporal extensions of SQL
over the years till the present state in a single document.
Researchers need to gather several pieces of literature
before they can obtain a vivid pictorial timeline of the
history and the current state of these temporal extensions,
for research and software development purposes. It is for
this reason that this paper seeks to provide a state-of-the-
art survey on the temporal extensions of SQL, identify any
drawbacks of the current extensions and present further
research being conducted in this area. The survey
conducted addresses the questions below:
 Which temporal features are available in the
various SQL standards?
 What are the key activities and research which
have been undertaken in the field of temporal

databases and their consequent impact on the
most recent temporal extensions of SQL?
 How are the various database vendors adopting
temporal extensions of SQL to their database
management system products?
 Are there any shortcomings of the most recent
SQL standard, SQL: 2011 and which further
research is being conducted to solve these
problems?
By gathering and analyzing various literature, other
sources of information and extensive research, we attempt
to understand the rationale behind the temporal concepts
in SQL over the years.
2. TEMPORAL DATABASES
Conventional databases represent the state of information
at a single point in time; new data can be added and
modifications to existing data can be easily performed.
However, a drawback of these conventional databases
was the fact that old values were usually lost whenever
changes occurred and as such it was impossible to track
the change of information over time in the database. Also,
in the early years, DBMSs provided very limited support
for handling attributes involving time; Application
programs had to implement functionalities for this
purpose. Keeping track of data over time was therefore a
major problem which needed to be addressed in
databases.
Although the relational data model had gained popularity
over the years, it provided very little support for
addressing the temporal dimension of data. The need to
handle time more comprehensively arose in the early
1970s especially in the area of medical information
systems, where a patient’s medical history is particularly
important [6]. Other application areas which handle time
related information in databases include banking,
inventory management and airline reservation systems.
Today, handling data that evolves over time can be
effectively done using temporal databases. A temporal
DBMS therefore has built-in support for time-varying
data.
In temporal databases, timestamps can be attached to
data or information. To establish the lifetime of a piece of
information, we can represent its start time and end time
in a database. The three notions of time which are
relevant to temporal databases are user-defined time,
valid time and transaction time. Valid and transaction
times were coined by a very active researcher in the
database community called Richard Snodgrass, together
with his doctoral student, Ahn [7]. Valid time refers to
when a fact is true in the real world. Transaction time is
the time a fact is stored or modified in a database. User-
defined time is used to include temporal information
which is not handled by transaction time or valid time;
and values of user-defined time are not interpreted by a
DBMS.
As an example, consider Mr. X who registers for his health
insurance on 4th June, 1995. However, the health
insurance company could not record this information in
their computer system because of technical challenges. On
8th July, 1995, the company is able to record Mr. X’s
information in their database. In this case, 4th June, 1995
is the valid time start event of his insurance and 8th July,
1995 becomes the transaction time. The need for
temporal support in databases keeps increasing over the
years: From the years 1982 to 1986, about 25 groups
were also studying time in databases and 60 articles on
temporal databases had been released [8]. It therefore
became necessary to develop a common consensus
glossary on temporal database concepts: this goal was
successfully achieved [9].
Many years now, numerous temporal extensions have
been proposed and therefore there is the need to further
consolidate information. Unlike temporal databases,
conventional databases model the dynamic real world as
a snapshot at a particular point in time, and it may not
reflect the status of the information in the real world.
Therefore, as soon as an update is performed on such a
database, the old information is lost and answering
queries about past states became a major problem. These
databases were referred to as snapshot databases [8]. To
make such databases up-to-date, DBMSs provide various
mechanisms. One of those was to store all the past states
of the snapshot database and index them by the times
they change (their transaction times).
Although this method was a step forward, we could only
keep track of the transaction time history of the database
without any support for history of the real world. Another
approach to store past information is by using historical
databases. These databases record a single historical state
per relation, thereby taking into account the valid time
[8]. In 1987, when Snodgrass introduced a new temporal
relational query language, TQuel [8], he also proposed
that the approaches to historical and rollback databases
can be adopted in temporal databases which require
support for both valid time and transaction time.
The effort of the database community to provide support
for temporal data has really paid off. Today most DBMS
vendors provide temporal support in their database
products. Temporal databases can now be easily created
to provide support for valid time, transaction time or
both. The widely used query language for relational
databases, SQL, has also undergone major changes to
provide improved support for temporal data.

3. STRUCTURED QUERY LANGUAGE (SQL)
Structured Query Language (SQL) is the standard
language used to retrieve information in relational
databases. It was initially developed at IBM by Donald D.
Chamberlin and Raymond F. Boyce in the early 1970s [6].
This initial version was referred to as Structured English
Query Language, SEQUEL, and was designed purposely to
manipulate data in IBM’s first relational DBMS prototypes
referred to as System R [6]. Subsequently, SQL was later
introduced as a commercial database system in 1979 by
Oracle Corporation.
In 1986, ‘‘Database Language SQL’’ was formally adopted
by the ANSI and ISO standards groups [10]. The official
name of SQL as specified by the ISO is ISO/IET 9075
Standard:”Information Technology- Database Languages-
SQL”. According to Donald Chamberlin [6], making SQL a
standard provided a mechanism for controlled evolution
of the language and also created a forum for users and
implementers to share ideas on the Language. From that
time till today, the SQL standard has undergone major
revisions and new features have been introduced
(including e.g., outer joins, table expressions, recursion,
triggered actions, user-defined types and functions, and
online analytic processing (OLAP) functions). One of the
most important features of SQL is its improved support
for temporal data over the years. New versions of the SQL
standard were released in 1989, 1992, 1999, 2003, 2006,
2008 and 2011.
4. TEMPORAL FEATURES OF SQL STANDARDS
OVER THE YEARS
The SQL standard has undergone eight main revisions and
the latest version was released in 2011. Each standard
came with its own set of features by correcting problems
in a previous standard and introducing entirely new
features. However, not all the eight revised versions
provided temporal support for SQL. The SQL standard has
been divided into multiple parts to enable the relevant
pieces to progress at different rates. Parts have been
rejoined and split over the years based on extensive
research and recommendations from the database
community. Figure 1 depicts the timeline of events with
respect to incorporating temporal extensions into SQL.
Figure 1: Timeline of history of temporal extensions of
SQL
4.1SQL-92
The first major revision on the standard since its
introduction occurred in 1992 and it was called SQL2 It is
also known by names such as SQL/92, SQL-92, ISO/IEC
9075:1992, ANSI X3.135-1992, or FIPS 127-2 [11]. SQL-92
introduced significant features including some initial
support for temporal data. Four data types were
introduced, three of which were used in order to represent
a point in time on the time axis referred to as an instant,
and the other used to represent duration of time. These
three data types were: DATE (a particular day, with a year
in the range ad 1–9999), TIME (a particular second within
a range of 24 hours), and TIMESTAMP (a particular
fraction of a second, defaulting to microsecond, of a
particular day) [12]. A DATE field consists of 10 characters
which are used to denote day, month and year. DATE
literals are specified in SQL by the keyword DATE,
followed by a string of the form YYYY: MM: DD. In SQL,
TIME is represented using the Universal Coordinated
Time. TIME values are represented by the keyword TIME,
followed by specifying the hour, minutes and seconds in
the form, hh:mm:ss. TIME fields are usually 8 characters
long but it is possible to specify precisions higher than a
second: In this case, a dot is introduced after the default 8
character string and the number of digits after the dot
must be placed in parenthesis after the keyword TIME.
TIMESTAMP data type contains year, month, day,
hour, minute, and second values represented in the form
YYYY-MM-DD hh:mm:ss. Just as in TIME data type, it is
possible to specify higher precisions by introducing
fractions of a second. The fourth data type introduced was
the INTERVAL. A time interval is a directed duration of
time with a known length, but not specific starting or
ending instants. If we wish to specify the duration
between two instant time points, the term PERIOD is used.
By the definitions of both INTERVAL and PERIOD, it is
clear that they represent entirely two different things.
SQL-92 supports two kinds of intervals, month-year and
second-day intervals [12]. There are three variants of
month-year intervals namely INTERVAL <year literal>
YEAR, INTERVAL <month literal> MONTH and INTERVAL
<year-to-month literal> YEAR TO MONTH [13]. Second-
day intervals can also be written in different forms such as

e.g., INTERVAL <day literal> DAY and INTERVAL <day-to-
second literal> DAY TO SECOND. More detailed
information about the description and use of intervals can
be obtained from [13].
4.2TSQL2
Although SQL-92 introduced some form of temporal
support, this was still not enough with respect to
managing time in databases. The database community
continued to work towards improving the old and
developing new mechanisms for temporal support in
databases. In 1994 [2], a temporal extension to SQL2
called TSQL2 was developed. TSQL2 is based on and highly
compatible with SQL-92; legacy applications in the latter
can easily be supported without any statement
modifications or introduction of additional codes.
The language specification inherits the DATE, TIME,
TIMESTAMP and INTERVAL data types from SQL-92. A
new temporal data type PERIOD was also introduced,
making it easier to specify range and precision. Other
results from the TSQL2 project included the introduction
of the three timelines: user-defined, valid and transaction
times, temporal aggregates and surrogate data. In SQL-92,
only snapshot tables existed; TSQL2 provided more
support for introducing temporal elements into tables
enabling users to include different timestamps into
snapshot tables, event tables, valid time tables, transaction
time tables and bitemporal tables. Valid time and
transaction time which represent the time a fact occurred
in the real world and the time a fact was stored in a
database respectively. The document containing the
TSQL2 Language specification can be obtained from [2].
This document contains about 71 pages of information on
the concepts, syntax and implementation of TSQL2.
4.3SQL:1999 and SQL/Temporal
The fourth revision of the SQL standard occurred in
1999; It was known as SQL3 or SQL:1999 [14]. An
interesting change occurred in the naming convention of
this standard. The hyphen which was used in the previous
standards such as SQL-89, SQL-92 was replaced with a
colon in SQL:1999. This change was needed for two main
reasons. The first was to ensure consistency with the
names of other ISO standards [15]. Secondly, unlike the
previous years where shortened form of the years, for e.g
‘86 and ‘92 were used, 1999 was maintained in full when
the standard was named. This decision was taken because
of the year 2000 computer problem, Y2K [16] and the
impact of dropping the century indicator from dates [17].
SQL:1999 consisted of eight different parts which were
published in two groups [17]. Though SQL:1999 was a
significant improvement on the previous SQL standards,
there were no new features introduced to improve
handling temporal data.
Due to the fact that SQL:1999 provided no new
support for temporal data, the database community made
attempts to introduce some parts of the TSQL2 into the
SQL:1999 standard. This new part (part 7) was called
SQL/Temporal and it was to be included as a new
substandard of SQL:1999. However, this attempt was
unsuccessful because the TSQL2 approach was heavily
criticized by Chris Date and Hugh Darwen [18]. In their
words, TSQL2 contained major “flaws” and one of the
obvious ones was related to “hidden attributes” [18].
Consequently SQL/Temporal was withdrawn and the ISO
project for temporal support was eventually cancelled in
2001 because people were no longer willing to dedicate
more efforts to the topic [18].
The fifth revision to the SQL standard occurred in 2003
[19]. SQL:2003 made revisions to all parts of SQL:1999
and added a brand new part, Part 14: SQL/XML, XML-
Related Specifications. New data types; BIGINT,
MULTISET, and XML were also introduced. [19] gives a
good overview of the features that were introduced in this
standard. There was no temporal extension to SQL in this
Standard. SQL:2006 was introduced after SQL:2003 and
the major change that it contained was revision to Part 14,
SQL/XML, of the SQL:2003 standard. In July, 2008,
SQL:2008 was introduced but also did not offer any new
form of support for temporal databases.
4.4SQL:2011
About three years after SQL:2008, a new standard,
SQL:2011 was released [5]. This standard was to some
extent based on some concepts from SQL/Temporal and
provided extensive support for temporal data. Out of the
nine parts which existed in SQL:2008, five of them were
modified. The remaining four remained in effect. SQL:2011
provides substantive support for handling temporal
information in databases. The new temporal features are
now part of SQL:2011 Part 2, SQL/Foundation [20],
instead of appearing as a new part as was in the case of
SQL/Temporal. New terminologies such as system-
versioned tables, application time period tables and
system-versioned application time period tables have
been introduced for the old concepts which existed in
TSQL2 namely transaction time tables, valid time tables
and bitemporal tables respectively.
SQL:2011 is the most recent of the various SQL
standards. After the last major project, SQL/Temporal,
which was to add temporal extensions to SQL was
cancelled in 2001, it took about 10 years before new
temporal extensions were added to SQL. In between the
SQL/Temporal and SQL:2011, the SQL standard
underwent three major revisions namely SQL:2003,
SQL:2006 and SQL:2008, but the revisions to the standard
did not include new features for handling temporal data
aside the pre-existing datatypes DATE, TIME, TIMESTAMP
and INTERVAL. Consequently, it was challenging for users

to represent the space of time between two instants which
is denoted as period because no datatype existed in SQL
which could serve that purpose. Users had to “define”
periods in their databases by creating two table columns,
for the start time and the end time. Many applications had
to implement basic temporal operations like time travel by
introducing extra codes in their software applications: this
often led to development overhead, faulty semantics and
sub-optimal performance [21].
In December 2011, SQL:2011 [22] was formally
adopted, replacing SQL:2008 as the most recent version of
the standard. This new standard is also referred to as
ISO/IEC 9075:2011 ("Information technology – Database
languages – SQL") and the official document contains more
than a thousand pages. One of the main new features of
this new standard is improved support for temporal
databases [23]. A new set of language extensions for
temporal data support are now part of SQL:2011 Part 2,
SQL/Foundation instead of appearing as a new part [24].
Almost four years after the introduction of this standard,
the amount of literature covering its temporal aspects is
quite limited. Several informative sources gathered during
our research include but are not limited to the following
[23] [24] [25].
Temporal concepts in SQL:2011 are largely based on
concepts and constructs from SQL/Temporal, but there
are significant differences [23]. The PERIOD datatype
which was proposed in TSQL2 and SQL/Temporal did not
make it into SQL:2011. However, SQL:2011 provides a
means of specifying distance between two instants, which
is called a period. Perhaps among the reasons why there is
no period data type in SQL till today is because of the costs
involved: For e.g, if a period datatype was added to SQL,
then it would have to also be added to the stored
procedure language, to all database APIs such as JDBC,
ODBC, and .NET, as well as to the surrounding
technologies such as ETL products, replication solutions,
and others [22].
As a means of representing periods, SQL:2011 adds period
definitions as metadata to tables: A period definition is a
named table component which identifies a pair of columns
that capture the period start and the period end time. At
any point in time, these period definitions can be created
or removed using enhanced CREATE and ALTER table
statements present in this new standard [23]. SQL: 2011
represents time periods using the closed-open model and
ensures that the start time is always less than the end
time. This constraint is enforced as soon as the keyword
PERIOD is declared in a table definition.
SQL:2011 introduces new terminology for some pre-
existing time dimensions from TSQL2 and SQL/Temporal:
Valid time tables are referred to as application-time period
tables and transaction time tables are now called system-
versioned tables [23]. Bitemporal tables can also be
represented in this new standard and are referred to as
system-versioned application-time period tables. Whereas
new syntactic expressions have been introduced for
queries in system-versioned tables and updates in
application-time period tables, it is the opposite case for
updates in system-versioned tables and queries in
application-time period tables. Another important feature
which has been introduced is the possibility to specify
primary key and unique constraints to ensure that no two
rows with the same key value have overlapping periods.
New period predicates have also been introduced in
SQL:2011 which help to formulate queries for conditions
involving periods. These predicates are functionally
similar to Allen operators but produce contrasting results
in many cases. Period predicates existed as far back as in
SQL:92, which had the OVERLAPS predicate. Both TSQL2
and SQL/Temporal also introduced the CONTAINS,
OVERLAPS, MEETS, PRECEDES and SUCCEEDS operators.
The above mentioned predicates have made their way into
SQL:2011, along with other new period predicates.
Notable among the new SQL period predicates are the
IMMEDIATELY PRECEDES and IMMEDIATELY SUCCEEDS
predicates. Considerable amount of success has been
achieved in the area of temporal support for databases but
a lot still needs to be done. Although the SQL standard has
undergone major revisions since it was established in
1989, only a few versions provide some support for
temporal data. The need to provide temporal support in
database systems became a major concern after SQL-92
was released. Since then, the most recognized projects and
efforts towards temporal support in databases occurred
during TSQL2, SQL/Temporal and finally in the most
current standard SQL:2011.
5.DBMSs’ SUPPORT FOR TEMPORAL DATA
In order to store, manipulate and query time related data
in databases, the underlying database management system
must provide built-in support for temporal extensions.
Over the years, quite a number of attempts have been
made by the database community, including DBMS
vendors and SQL standard committees, to develop
temporal extensions to SQL. However, not much
information is available about the research and efforts of
the above mentioned groups between the years 2002 and
2011 after the SQL/Temporal project was cancelled.
Nonetheless, some indications make it obvious that efforts
were actually made; For e.g., DBMS vendors began
adopting temporal features in their DBMS products, and
also concepts proposed by researchers have been
incorporated in the new standard, SQL:2011, to provide
improved temporal support. The historical and current
state of temporal support in DBMSs and the contribution
of some major DBMS vendors towards the current state of
temporal support in databases cannot be overemphasized.

One of the legacy systems which provided some form of
temporal support was the Oracle 9i, a database
management system developed by the Oracle Corporation
in 2002. This database system had the Flashback [26]
feature which permits users to query data as it existed in a
previous state. Oracle 9i has a special utility package,
DBMS_Flashback to handle flashback queries. To activate
the flashback feature, users first have to reconfigure their
databases to grant the appropriate privileges. After which
the user has to decide on one of two approaches which can
be used to implement flashback queries [27]. The first
approach is a time based approach, where users have to
explicitly specify time values from past time. The other
approach is to use a SYSTEM CHANGE NUMBER to identify
the point to go back to [27]. Each of these approaches
employs the “AS OF” clause. For example, the conventional
SQL SELECT statement is modified as: Select * from
employee AS OF TIMESTAMP<specify timestamp> or
Select * from employee AS OF SCN <include scn number
here> [28]. Oracle has maintained this query feature over
the years and its DBMS system, Oracle 12c still supports
flashback [29].
Other major DBMS vendors which have provided
substantive support for temporal data include IBM and
Teradata. IBM DB2 database management systems
implement almost all the temporal features in SQL:2011
using slightly different syntax and terminology [30].
Business time, system time and bitemporal tables in IBM
DB2 are named as application time period tables, system
versioned tables and system versioned application time
period tables respectively in SQL:2011 [30]. Slight
differences exist between the SQL:2011 specification and
the implementation of the temporal features in IBM DB2.
In IBM DB2 version 10.5’s implementation of temporal
concepts, a user does not have the possibility to define
their own names for periods when creating business time
tables: it is predefined as BUSINESS_TIME. This
implementation contrast with the convention for naming
application time period tables in SQL:2011, where period
names in create table statements are user-defined.
Also, the syntax for creating system time tables in IBM
DB2 10.5 differs from system versioned tables in
SQL:2011. In the former, three timestamped columns are
used to store system time values: two timestamped
columns store information about the start and end points
of the system time and one column stores information
about transaction start time [30]. The transaction start
time column tracks when a transaction first executes a
statement which changes the table's data. Users are
permitted to either manually enter or instruct the system
to automatically generate the values for the three
timestamped columns mentioned above. In the latter
instance, the keyword GENERATED ALWAYS is used; This
keyword also exists in SQL:2011 and serves the same
purpose. To create business time tables in IBM DB2 10.5,
one has to include two columns to represent the start and
end time periods, and a PERIOD BUSINESS TIME clause
[30]; This is similar to how application time period tables
are specified in SQL:2011. Bitemporal tables in IBM DB2
10.5 includes specifying columns for both system time and
business time; the same concept in SQL:2011.
In Teradata, the temporal features are based on the TSQL2
specification [31]. Temporal query processing in Teradata
is implemented by using functional query rewrites or by
implementing a native temporal support in the underlying
database engine. The latter approach processes temporal
queries by compiling and rewriting them into generic,
non-temporal operations [31]. Although some members of
the Teradata Company state that the rewrite approach is
generally simpler to implement, they also admit that it
adds a structural complexity to the original query, which
can pose a potential challenge to query optimization [31].
The other way to implement temporal queries in Teradata
is to implement temporal database operations such as
scans, projections and joins directly in the DBMS internals
[31].
The concepts valid time and transaction time are also
“employed” by Teradata. Teradata also supports
bitemporal tables, but at most one system time and at
most one application time is allowed per table [31] as
opposed to the specification in SQL:2011, where two
values represent the application time. Temporal data
types supported in Teradata include the “usual” DATE,
TIME, TIMESTAMP and INTERVAL as well as the “much
anticipated” PERIOD datatype. Teradata supports the
PERIOD data type since V13.0; a PERIOD column can be
any date or timestamp type although the beginning and
end must have the same type, and the format of the date or
timestamp also needs to be specified in the create table
statements. The PERIOD data type uses the closed/open
concept, where the start of the period is included but the
end is excluded. Perhaps Teradata is a “step ahead” with
regards to introducing the period data type, which still
does not exist in SQL:2011. In table 1, we present an
overview of the current state of temporal support for
databases by three of the leading DBMS vendors.
Aside implementing their own temporal features, it is also
known [30] that some of these DBMS vendors such as IBM,
worked with the ANSI and ISO SQL standard committees
to incorporate the new temporal extensions into the latest
SQL:2011 standard. It is interesting to note that, IBM DB2
was the first database vendor, which fully supported the
temporal features of SQL:2011. Defining and using tables
in Teradata follows the TSQL2 specification, thereby
raising the problem of incompatibility with SQL:2011 [21].

Table -1: Current state of temporal support provided
by DBMS vendors
DBMSs
Temporal
concepts in
SQL
IBM DB2 Oracle Teradata
DATE, TIME,
TIMESTAMP
data types
INTERVAL
data type
PERIOD data
type
System
versioned
tables
ST table
(3 columns
for ST)
TT table (1
column for
TT)
TT table (1
column for
TT)
Application
time period
tables
BT (2
columns for
BT )
VT table (2
columns for
VT)
VT (1
column for
VT)
System
versioned
application
time period
tables
Bitemporal
(5 columns
for both ST
and BT)
Bitemporal
(3 columns
for both VT
and ST)
Bitemporal
(2 columns
for both VT
and ST)
ST –System time BT- Business time
TT- Transaction Time VT – Valid Time
 - available X - unavailable
6. CONCLUSION
SQL support for temporal features in databases has
undergone reformation over the years. Some of the
features had to be dropped while others are still available
in the current SQL standard. In figure 2, we look at the
trend of developments with regards to some temporal
support introduced in the various SQL standards and
projects over the years. The figure does not cover all the
temporal concepts in the various projects but just a few of
the major ones. It can be observed from the figure that the
data types INTERVAL and PERIOD have been dropped in
the most recent version of SQL, SQL:2011. Also, the main
idea behind the concepts “valid time”, “transaction time”
and “bitemporal” have been maintained over the years;
However, each of the standards and projects shown in the
diagram referred to the concepts by different names. The
difference between the representations of the concepts in
various standards does not only lie in the naming, but also
the syntax for creating, storing and manipulating data in
the respective tables. Looking at the diagram, it is
observable that the newer projects improve on the
concepts from the preceding ones by either changing
syntax or introducing new concepts.
Temporal extensions of SQL have been developed and
improved over the years and the level of difficulty
associated with managing data related to time in
databases is on a lower level now. In spite of this
achievement, there is still more which can be done,
especially in the area of surveying new forms of temporal
support for databases and working towards the inclusion
of new temporal extensions in the subsequent SQL
standards to be developed.
One major “problem” which has still not been addressed is
the absence of a PERIOD data type in SQL. Proposals for
PERIOD data type could not go beyond the TSQL2,
SQL/Temporal era. Since SQL:2011 did not support this
data type, it remains unclear what the future holds with
respect to it. Many people anticipate that the introduction
of PERIOD data type into SQL will be costly because a lot of
the technologies which “work” with SQL would have to be
modified to adapt to this data type [23]. As at the time of
our research, we found virtually no information about
when a new SQL standard will be released and whether or
not it will include support released and whether or not it
will include support for the PERIOD data type. Aside the
PERIOD data type, there are other features which are still
not supported in SQL and perhaps some of these features
may be introduced in the next standard. These features
include but are not limited to: support for multiple
application time periods per table and improved support
for period joins [23]. A period join is performed by joining
a row from one table with a row from another table such
that their application time or system time periods satisfy a
condition such as overlap. Using the OVERLAPS operator
in SQL:2011, inner joins can be performed but additional
support is still needed in order to perform outer joins [23].
Another feature which could also be introduced is
coalescing, to bring together tuples which have identical
attribute values and with timestamps, which are adjacent
in time or share some time periods in common [27].
Introducing temporal extensions into SQL has been
beneficial to DBMS vendors and application developers.
DBMS vendors have and are still making efforts to adapt
their systems to support temporal data. Temporal support
by DBMSs ensures consistent handling of time-related
events and reducing query complexity in temporal
databases. A study presented in [21], evaluated the
performance of DBMS products with respect to their
architectural support for temporal databases and
performance when executing temporal queries. The
results showed that the support for temporal data is still
in its infancy as all the DBMS they considered store their
data in regular, statically partitioned tables and rely on






x x

standard indexes as well as query rewrites for their
operations, leading to performance overheads [21].
As the SQL standard is revised over the years, it is likely
that we will observe a continued improved support for
handling temporal data in databases. The field of temporal
database research is still active and there even exists an
international center, the TimeCenter, which supports
temporal database applications on traditional and
emerging DBMS technologies [27].
Some members of this Center include very renowned
people who have made significant contributions to the
field of temporal databases. Temporal database research
has surely paid off and although we are not certain of the
timing, the database community still awaits the
development of next SQL standard after SQL:2011, with
the hope that it will provide improved support for
temporal databases.
Figure 2: Some temporal features of the various SQL
standards and projects. Abbreviations: PP- period
predicates, BT - bitemporal table, TT - transaction time
VT – Valid time
REFERENCES
[1]Snodgrass, Richard. "TSQL: A Design Approach." White
paper, University of Arizona, Department of Computer
Science, Tucson ,1992
[2]Snodgrass, Richard Thomas, et al. "TSQL2 language
specification." Sigmod Record 23.1, 1994, pp 65-86
[3]Eisenberg A.; Melton J., SQL: 1999, formerly known as
SQL3, ACM SIGMOD Record Volume 28 Issue 1, March
1999, New York, USA, pp 131 – 138
[4]Snodgrass, R et al., "Transitioning temporal support in
TSQL2 to SQL3." Temporal Databases: Research and
Practice, Springer Berlin Heidelberg, 1998, pp 150-
194
[5]Zemke, Fred. "What’s new in SQL: 2011." ACM SIGMOD
Record 41.1 ,2012, pp 67-73
[6]Chamberlin, Donald D; Boyce, Raymond F."SEQUEL: A
Structured English Query
Language" (PDF). Proceedings of the 1974 ACM
SIGFIDET Workshop on Data Description, Access and
Control (Association for Computing Machinery), 1974
pp 249–264
[7]Snodgrass Richard T. and Ilsoo A, "Temporal
Databases," IEEE Computer 19(9), September, 1986,
pp. 35–42
[8]Snodgrass, R. "The temporal query language TQuel."
ACM Transactions on Database Systems (TODS) 12,
no. 2,1987, pp 247-298
[9]Dyreson, C., Fabio G,, Wolfgang K., Nick K., Nikos L.,
Yannis M., Angelo M. et al. "A consensus glossary of
temporal database concepts." ACM SIGMOD Record
23, no. 1, 1994, pp 52-64
[10]Chamberlin, Donald D. "Early history of SQL." Annals
of the History of Computing, IEEE 34, no. 4 , 2012, pp
78-82
[11]“The SQL-92 Standard”, Second Informal Review
Draft, ISO/IEC 9075:1992, Database Language SQL,
Digital Equipment Corporation, Maynard,
Massachusetts, July 30, 1992
[12]Snodgrass, Richard T. "Temporal databases", In IEEE
computer, 1986
[13]Representing time in SQL, Temporal Information
Systems lecture notes by Prof. Dr. Rainer Manthey,
Department of Computer Science, University of Bonn,
Germany, Summer semester 2014
[14]ANSI/ISO/IEC International Standard (ISO/IEC 9075-
2:1999) Database Language SQL — Part 2: Foundation
(SQL/Foundation), Part 2, September 1999

[15]Van Der L., Rick F., SQL for MySQL Developers: a
comprehensive tutorial and reference, Pearson
Education, 2007
[16]Y2K bug. Encyclopædia Britannica Online
[17]Melton, J, Alan R. S., SQL: 1999: Understanding
relational language components, Morgan Kaufmann,
2001
[18]Darwen, H, Date C.J. "An Overview and Analysis of
Proposals Based on the TSQL2 Approach.", 2005
[19]Eisenberg, A., Jim M., Krishna K., Jan-Eike M., and Fred
Z. "SQL: 2003 has been published." ACM SIGMOD
Record 33, no. 1 ,2004, pp 119-126
[20]ISO/IEC 9075-2:2011, Information technology—
Database languages SQL Part 2: Foundation
(SQL/Foundation), 2011
[21]Kaufmann, Martin, et al .,Benchmarking Bitemporal
Database Systems: Ready for the future or stuck in the
past, Proc. of the 17th International Conference on
Extending Database Technology (EDBT), March 24-28
2014, Athens, Greece
[22]SQL:2011, International Organization for
Standardization (ISO) webshop
http://www.iso.org/iso/iso_catalogue/catalogue_tc/c
atalogue_detail.htm?csnumber=53681
[23]Kulkarni, K. and Michels, Jan-E.. , Temporal features in
SQL:2011, ACM SIGMOD Record, 2012, Vols. 41.3, pp
34-43
[24]ISO/IEC 9075-2:2011, Information Technology-
Database languages-SQL-Part 2. Foundation
(SQL/Foundation), 2011 as cited in Kulkarni K.,
Michels J.E, Temporal features in SQL:2011, ACM
SIGMOD Record, 41.3, 2012
[25]Baumunk, C., “Bitemporaldata.com: An overview of
temporal features in SQL:2011”, Slides on temporal
features of SQL:2011, May 22, 2015
[26]Adrian Billington ,Flashback query in oracle 9i, Oracle
Developer.net, November 2002,
[27]Petković, D., Temporal Data in Enterprise Database
Systems, The 7th International Conference on
Information Technology, 2015, pp 276-282
[28]Irfan Haq, Recovery made simple, Oracle Flashback
Query ,Oracle FAQ online, October 2004
[29]Using Oracle Flashback Technology, Database
development , Oracle Database Online Documentation
12c Release 1 (12.1) / Database Administration
[30]Saracco, Cynthia M., Nicola, M. and Ghandi, L. , A
matter of time: Temporal data management in DB2 10,
IBM, April 3, 2012
[31]Al-Kateb, Mohammed, et al., Temporal query
processing in Teradata, EDBT '13: Proceedings of the
16th International Conference on Extending Database
Technology, ACM, March 2013, pp 573-578

SQL and Temporal Database Research: Unified Review and Future Directions

More Related Content

Similar to SQL and Temporal Database Research: Unified Review and Future Directions (20)

More from IRJET Journal (20)

Recently uploaded (20)

SQL and Temporal Database Research: Unified Review and Future Directions