What
is IBM DB2 UDB's Data Links?
What
is the problem domain that Data Links addresses?
What
sort of applications is Data Links appropriate for?
How
does an application designer use Data Links?
What
flavors of file reference integrity does Data Links provide?
What
flavors of file reference security access does Data Links provide?
What
distributed system capabilities does Data Links support?
How
is Data Links Replication integrated with DB2 UDB database replication?
Does
Data Links complicate file administration tasks?
How
does Data Links relate to DB2 UDB's object-relational capabilities
of BLOBs, User Defined Types (UDT) and User Defined Functions
(UDF)?
Is
there not performance degradation when accessing files managed
by Data Links?
Are
there any customer references, white papers and articles on
Data Links?
What
IBM products incorporate Data Links technology?
Is
Data Links an IBM proprietary technology?
Are
there any Data Links similar products in the marketplace?
Will
I get locked in to IBM DB2 UDB if I use Data Links in my applications?
What
operating system platforms is Data Links currently available
on?
When
did Data Links become available?
What
future enhancements are planned for Data Links?
How
much does Data Links cost?
Who
can I contact for more information about Data Links?
What
is IBM DB2 UDB's Data Links?
IBM DB2 UDB's
Data Links technology extends the mission-critical relational DBMS
capabilities of referential integrity, value-based security, transactional
consistency and coordinated backup and recovery to file systems.
Data Links
manages files that reside outside the database as though they are
logically within the database. It guarantees referential integrity
to these external files, provides enhanced access control to them,
and supports automatic and coordinated backup and restore capability
within transactional environments that are crucial for data management.
Data Links simplifies and reduces system administration costs and
complexities, by providing a single administration point for file
and database data. Data Links technology thus enables a powerful
symbiotic relationship between IBM DB2 UDB and file systems.
Back to questions
What
is the problem domain that Data Links addresses?
Estimates vary
that between 70% and 95% of data resides in file systems rather
than databases, and that the Internet has exacerbated this divide
in favor of file systems. Current e-business trends such as e-commerce,
supply chain management, and customer relationship management, require
applications that can integrate content from various DBMS and file
system sources, including unstructured data. Examples of such unstructured
data are audio, video, and images traditionally stored in an assortment
of file formats. Organizations are being challenged to support existing
& emerging applications that utilize less robust file systems,
while integrating the information in these file systems into the
new applications of e-business.
Data Links
enables organizations to use their existing & emerging file
system based applications by leaving them untouched, while enabling
the same files to be integrated with database systems to meet the
rigorous integrity, security and transactional demands of e-business.
It does this by allowing the files to reside outside the database,
while extending the umbrella of mission-critical DBMS capabilities
over the files as if it were logically within the database.
Back to questions
What
sort of applications is Data Links appropriate for?
Data Links
is a candidate for any application which involves processing information
from multiple heterogeneous sources that include databases and file
systems, where it is required that this information be consistent
between the different sources, secure, accurate, and timely. Besides
the e-commerce, customer relationship management, and supply chain
management e-business applications mentioned earlier, Data Links
can be used effectively in automotive and health insurance applications
to integrate files such as X-rays, ECG results, vehicle damage pictures;
engineering & manufacturing applications involving design documents
and plans; configuration and asset management application, such
as, web assets (web pages, server-side programs and templates);
and the management of large volumes of scientific data residing
in file systems.
Back to questions
How
does an application designer use Data Links?
An application
designer can use Data Links to maintain a reference to an existing
operating system file in a column of a DB2 table. This reference
is stored via a new standard SQL data type called DATALINK, using
a URL (Uniform Resource Locator) syntax. Other columns of this table
would generally maintain meta data about the DATALINKed files. Typically,
the application programmer would insert rows in this table with
meta data about the file, and its file reference (URL syntax) in
the DATALINK column. The application would then typically use an
SQL query against this table on the meta data columns to locate
the file(s) of interest, retrieve the file reference in the DATALINK
column, and then use that URL to directly access the file using
the native APIs of the file system or through a browser.
For the application to update or delete to a linked file, it must
first unlink the file from the DB2 UDB table before the operation
is allowed to proceed. Multiple DATALINK columns can be defined
for a single DB2 UDB table.
Back to questions
What
flavors of file reference integrity does Data Links provide?
Data Links
allows an application designer the option of choosing from a range
of file reference integrity options depending upon the requirements
of the application. One may choose to enforce referential integrity
or not for a file reference. If referential integrity enforcement
is chosen, one may choose to either allow the files to be updated
while linked, or made read-only while linked. Special considerations
apply depending upon the exact option chosen.
Back to questions
What
flavors of file reference security access does Data Links provide?
At a broad
level, Data Links allows an application designer the option of permitting
file access to either
- All users
who have database access privileges to the particular rows in
the DB2 UDB table which contains the DATALINK column. In this
case, any user requiring access to a file, must first access the
database to obtain a Data Links access control token before attempting
to access the linked file. This model is appropriate for applications
that would like the database to exclusively control all access
to DATALINKed files
or
- All users
who have access privileges to the file granted to them via the
operating system mechanism. The file can be accessed with or without
requiring any database access. This model is appropriate for applications
that would like to maintain existing file system access modes
to DATALINKed files without any database intervention or interaction.
Back to questions
What
distributed system capabilities does Data Links support?
Data Links
has been designed to support a distributed computing environment,
with capabilities that include the following:
- A DATALINK
column in a DB2 UDB table can reference one or more file systems
spread over one or more file system servers associated with different
operating systems such as Unix and Windows NT
- A single
DataLink Manager can be associated with DATALINK columns in one
or more DB2 UDB databases
- A DATALINK
column can reference files residing in Transarc's distributed
file system DCE-DFS
- Bi-directional
coordinated replication of DATALINKed files is supported in an
atomic, automatic and consistent way in conjunction with DB2 UDB's
database replication capabilities -- see following question for
more details.
Back to questions
How
is Data Links Replication integrated with DB2 UDB database replication?
DB2 UDB Version
7 has been enhanced to replicate the meta data associated with DATALINKed
columns, and the files referenced by the DATALINKed columns to a
target site in an atomic, automatic and consistent manner. When
an application updates the DATALINKed column of a row, and/or the
non-DATALINKed columns in the same row, then the DB2 UDB Version
7 coordinated replication mechanism ensures that the changes to
the non-DATALINKed columns, and the file associated with the DATALINKed
are propagated to the target site, and applied within a single unit-of-work.
Back to questions
Does
Data Links complicate file administration tasks?
No. On
the contrary, Data Links simplifies many of the tasks associated
with managing files that are logically integrated with database
information. For instance, Data Links ensures that a backup of a
database is coordinated with a backup of all the files referenced
in that database i.e. a point-in-time image is captured for the
combination of database and file data. In the same vein, a restore
of the database from a backup results in Data Links automatically
restoring the corresponding content of the files, thereby guaranteeing
consistency. Contrast this with other systems where the onus of
keeping the database and file systems synchronized is left to
the database administrator and prone to error -- thus potentially
jeopardizing the integrity of the information and the application.
Back to questions
How
does Data Links relate to DB2 UDB's object-relational capabilities
of Blobs, User Defined Types (UDT) and User Defined Functions (UDF)?
Data Links
treats information residing in file systems as though they were
logically within the database, leaving existing applications untouched.
DB2 UDB's object-relational capabilities permit an application designer
to either write UDFs that process files either stored within Blobs,
or are referenced by a DATALINK column or a UDT. If files are stored
within BLOBS, and are also currently used in existing applications,
then the potential exists for synchronization problems between the
BLOB and native file information. Therefore, while both features
support file processing via the database, they address different
application requirements, and are complementary in nature.
DB2 UDB is
unique in the industry in offering the customer these two choices
to best serve the custom demands of their application.
Back to questions
Is
there not performance degradation when accessing the contents of files managed by
Data Links?
No. Data
Links does not add any overhead to the file READ and WRITE
operations. As described earlier, Data Links ensures referential
integrity of file references from the database by intercepting file
operations such as OPEN, DELETE and RENAME that could compromise
the integrity of the file reference. File READ and WRITE operations
do not compromise referential integrity and are therefore
obscured from Data Links line of sight . Also, given that the number
of file READ and WRITE operations generally outnumber the occurrences of
OPEN, DELETE or RENAME on a file, the performance degradation if
any is minimal to marginal.
Back to questions Are
there any customer references, white papers and articles on Data
Links?
Data
Links technology has been received enthusiastically in the
engineering design and manufacturing community to manage their
voluminous engineering drawings. Dassault Systèmes
has implemented Data Links in their ENOVIA system, which is
used widely by customers, such as, Boeing, Daimler Chrysler,
and Peugot-Citroen. Please refer to the http://www.almaden.ibm.com/projects/datalinks.shtml
web site for more detailed information on Data Links technology
and usage.
Back to questions
What
IBM products incorporate Data Links technology?
Data Links
technology is currently supported by DB2 UDB's Text Extenders and
Websphere Application Server's Enterprise Java Beans' container
managed persistence. Plans are underway to include Data Links technology
in IBM Content Manager, Net.Commerce, and other upcoming products.
Back to questions
Is
Data Links an IBM proprietary technology?
No. While
IBM invented Data Links technology in the early 1990s, the specification
of the DATALINK data type has been accepted by both ANSI and ISO
in SQL Part 9: Management of External Data (SQL/MED). ISO has progressed
SQL/MED to final committee draft (FCD) in November 1999.
If things go according to schedule, it should become a standard
in late 2000.
Back
to questions
Are
there any Data Links similar products in the marketplace?
To our knowledge,
we are unique in the industry in offering this technology to our
customers. It is our understanding however, that our competitors
have projects underway to fill this void in their products.
Back
to questions
Will
I get locked in to IBM DB2 UDB if I use Data Links in my applications?
No. As
indicated earlier, the DATALINK specification is in ISO FCD status
and is expected to become a standard in the near future. Application
Designers who choose to implement DB2 UDB's Data Links, and then
require to port this application to other RDBMS platforms may do
so by emulating the Data Links API using the RDBMS' object-relational
capabilities of user defined data types and user-defined functions.
Of course, in such cases, the application would not obtain the benefits
delivered by DB2 UDB's enforcement of referential integrity, access
control, transaction consistency and coordinated backup and recovery
-- in other words, these capabilities would either have to be written
in the user-defined functions or sacrificed until the RDBMS stepped
up with these capabilities.
Back
to questions
What
operating system platforms is Data Links currently available on?
Data Links
technology is currently implemented in IBM's DB2 UDB for AIX, DB2
UDB for NT and DB2 UDB for AS/400. Solaris support is planned for
late 2000, with other platforms being added on a demand basis. File
systems supported include AIX JFS, AIX DFS, Windows NTFS and Solaris'
UFS.
Back
to questions
When
did Data Links become available?
Data Links
initially became available on the IBM AIX platform in DB2 UDB Version
5.2 in 1998
Back
to questions
What
future enhancements are planned for Data Links?
- We will
continue to add new platform and file system support on a demand
basis.
- Functionally,
we are looking at adding file-update-in-place, file transactional
semantics, logical file object support, interfacing hierarchical storage
management systems, and scaling the number of file servers that
can be connected to a database server.
- Other research
involves the development of cutting-edge applications of the Data
Links technology to enable efficient and robust intranet asset
management.
Back
to questions
How
much does Data Links cost?
Data
Links is a separately priced feature of IBM's DB2 UDB product
for the Windows NT and UNIX platforms. Details of current
pricing can be obtained at http://www.ibm.com/software/data/db2/udb/
Back
to questions
| 
