Enhancing Data Staging as a Mechanism for Fast Data Access

International Journal of Computer Applications Technology and Research
Volume 4– Issue 8, 584 - 591, 2015, ISSN: 2319–8656
www.ijcat.com 584
Enhancing Data Staging as a Mechanism for Fast Data
Access
Reagan Muriithi Gatimu
School of Computer Science
and Information Technology
Jomo Kenyatta University of
Agriculture and Technology
(JKUAT)
Nairobi, Kenya
Wilson Cheruiyot
(JKUAT)
Nairobi, Kenya
Michael Kimwele
(JKUAT)
Nairobi, Kenya
Abstract: Most organizations rely on data in their daily transactions and operations. This data is retrieved from different source
systems in a distributed network hence it comes in varying data types and formats. The source data is prepared and cleaned by
subjecting it to algorithms and functions before transferring it to the target systems which takes more time. Moreover, there is pressure
from data users within the data warehouse for data to be availed quickly for them to make appropriate decisions and forecasts. This has
not been the case due to immense data explosion in millions of transactions resulting from business processes of the organizations. The
current legacy systems cannot handle large data levels due to processing capabilities and customizations. This approach has failed
because there lacks clear procedures to decide which data to collect or exempt. It is with this concern that performance degradation
should be addressed because organizations invest a lot of resources to establish a functioning data warehouse. Data staging is a
technological innovation within data warehouses where data manipulations are carried out before transfer to target systems. It carries
out data integration by harmonizing the staging functions, cleansing, verification, and archiving source data. Deterministic
Prioritization Approach will be employed to enhance data staging, and to clearly prove this change Experiment design is needed to test
scenarios in the study. Previous studies in this field have mainly focused in the data warehouses processes as a whole but less to the
specifics of data staging area.
Keywords: Data Staging; Source System; Deterministic Prioritization; Data Warehouse
1. INTRODUCTION
The growing number of business transactions in any
enterprise is directly proportional to growth of data size. This
data comes from variant source systems and applications and
needs to be organized in a workable state so that it remains
relevant and meaningful to the users. Technological
development has led to the rise of Data Warehouse (DW).
(Inmon, 2002) defines a data warehouse as “collection of
integrated, subject-oriented databases designated to support
the decision making process”. Both (Kimball and Inmon,
2002) agree that a DW has to be integrated, subject-oriented,
nonvolatile and time variant. This concept of time-variance is
so crucial and ultimate concern and sets the basis for this
research since it focuses on improved data access. The
foundations of a DW as explained by (Zineb, Esteban, Jose-
Norberto, Juan, 2011) encompass integration of multiple
different data sources. This allows the provision of complete
and correct view of the enterprise operational data which is
synthesized into a set of strategic indicators and measures that
the users of the data can associate with.
DW has business intelligence implemented in three major
processes used to prepare data to match user’s needs. They are
commonly referred to as ETL processes namely; Extraction,
Transformation and Loading. Extraction process retrieves data
as is from source systems before subjecting it to any
manipulations. Transformation process also referred to as
transportation phase is the operational base and the most
intriguing of all. Business rules and functions are some of the
operations applied to the extracted data. Loading process
involves moving the desired data as determined by the users
to the DW. It’s important to note that the discussed flow of
data is not as simple and smooth as it sounds and this is as a
result of impeding performance issues raised by the following
observed bottlenecks.
(El-Wessimy et al, 2013) shows the relevance of DW in
decision making in today’s environment. “The best decisions
are made when all the relevant data is taken into
consideration. Today, the biggest challenge in any
organization is to achieve better performance with least cost,
and to make better decisions than competitors. That is why
data warehouses are widely used within the largest and most
complex businesses in the world.”
In this paper, SQL Server Integration Services tool is used to
experimentally show the impact of prioritizing the data from
sources as per the confidence levels and the distinctiveness of
the data. The units of measure also focus on general
performance of the whole ETL process after enhancement of
the data staging area.
2. RELATED WORK
2.1 Extraction Stage
This is the initial stage of data migration to a data warehouse.
(Kimball et al., 1998) informs that the extraction process
consists of two phases, initial extraction, and changed data
extraction. In the initial extraction, data from the different
operational sources to be loaded into the DW is captured for
the first time. This process is done only one time after
building the DW to populate it with a huge amount of data
from source systems. The next phase involves incremental
extraction also referred to as changed data capture (CDC).
(Stephen, 2013) informs that “The staging tables usually get
populated by some outside source, by either pulling or

www.ijcat.com 585
pushing the data from the source systems. This process is
usually an insert only process and therefore does not rely on
statistics for its successful execution.”
2.2 Transformation Stage
Once the data is extracted to the staging area, there are
numerous potential transformations, such as cleansing the data
(correcting misspellings, resolving domain conflicts, dealing
with missing elements, or parsing into standard formats),
combining data from multiple sources, reduplicating data, and
assigning warehouse keys. These transformations are all
precursors to loading the data into the data warehouse
presentation area. Unfortunately, there is still considerable
industry consternation about whether the data that supports or
results from this process should be instantiated in physical
normalized structures prior to loading into the presentation
area for querying and reporting.
(Erhard and Hong, 2000) elaborate on activities within
transformation phase towards clean data. These include data
analysis that focus on meta-data and due to fewer integrity
rules it cannot guarantee sufficient data quality of a source.
Two approaches have been put across to assist in data analysis
i.e. data profiling and data mining. Data profiling focuses on
the instance analysis of individual attributes. It derives
information such as the data type, length, value range, discrete
values and their frequency, variance, uniqueness, occurrence
of null values, typical string pattern providing an exact view
of various quality aspects of the attribute. Data mining helps
discover specific data patterns in large data sets, e.g.,
relationships holding between several attributes.
2.3 Loading Stage
Loading in the data warehouse environment usually takes the
form of presenting the quality-assured dimensional tables to
the bulk loading facilities of each data mart. The target data
mart must then index the newly arrived data for query
performance. When each data mart has been freshly loaded,
indexed, supplied with appropriate aggregates, and further
quality assured, the user community is notified that the new
data has been published.
When it comes to moving data to DW (Stephen, 2013)
informs “The biggest question for the staging area is –
how do we keep the statistics up-to-date such that the
statistics for a particular daily load are always available and
reasonably accurate. This is actually more difficult than it
sounds. If the partitions would only be analyzed in the first
quarter of the month each night, going to every other night
and eventually each week because of the 10% stale
setting. This obviously leaves us with a problem…. In order to
have the statistics available for the latest day which is loaded,
the statistics would have to be gathered after the staging tables
have been loaded but before the ETL process starts.
2.4 Data Staging
Data staging emerges as a new technological development
with an attempt to handle the low performance issues noted
above. Its location within the ETL process differs as (Kimball
and Ross, 2002) state that data staging is available in the
extraction and transformation phases of ETL framework. In
some current systems a data stage exists as a location that
interconnects Online Transaction Processing systems
(OLTPs) to the Online Analytical processing systems
(OLAPs).
2.5 Reasons for Data Staging Enhancement
Although data staging is not a new technology since it has
been researched before, the focus has been shifted to designs
and development of data staging frameworks. Little attention
has been given to its operability and its significant role in
speeding the ETL process.
In a production environment especially a busy organization
that deals with large transactions in its daily operations, data
flow to DW and storage repositories becomes an issue. Some
may not be experiencing the performance problems initially
but when their data levels grow they start getting intermittent
performance. This should be handled early to have a
maintained work path. It still remains a challenge on selection
algorithms that can pre-determine the data needed at the target
system. The proposed solution in data staging which forms the
enhancement is to work on the pre-determining and
prioritization mechanisms on the data to load.
(Aksoy et al, 2001) introduces a more workable approach to
data staging concerns. They based their work on broadcast
scheduling and data staging. According to them the key
design considered for development of large scale on-demand
broadcast server was the scheduling algorithm selection useful
for selecting of items to be broadcast. However, this solution
is based on assumptions that data will be available before
hand which is not true due to its dynamism.
3. PROPOSED APPROACH IN DATA
STAGING
Considering the improvements already observed from great
works of other authors it’s important to appreciate their efforts
in finding gaps in existing systems. Relating to the recurring
problems within the data staging, the researcher identifies the
following dependent and independent variables that assist in
choosing deterministic prioritization approach as a probable
solution.
3.1 Dependent and Independent Variables
Figure 1. Dependent and Independent variables
Legacy systems use stacking, where a job is pushed into the
pool of tasks and popped out of the stack sequentially when
its time of execution arrives according to the queue structure.
The order of execution is not highly dependent on pre-
arranged structure but on the procedural mode which degrades
performance.
The cache offer pre-fetching capability where data storage
occurs temporarily for the most used data. The scheduled
procedure looks firstly in the cache memory before checking
on secondary storage locations to minimize the search time.
The limitation lies on the cache size and amount of data to be
maintained in cache at a time.
To achieve concurrent operation there needs to be selective
algorithms to decide the priority of jobs from the source
systems to the data staging area. With Opportunistic
scheduling there is high probability of improving speed of
data retrieval and access.
Remote connection to source systems affects the speed of
retrieval and query execution is delayed by the time-lapse for

www.ijcat.com 586
distributed systems. This impact on the nature of ordering
results from query execution and thus optimization should be
introduced to work with stored procedures and cache facility.
The data characteristics are defined by type and formats since
it comes from disparate systems. Destination requirements
must be matched before data is moved to the target systems.
Poor data manipulation functions result in longer time
processing the data slowing down the systems. The functions
for manipulating flat files are different from the ones for
relational tables and databases due to underlying data formats.
3.2 Deterministic Prioritization (DP)
After thorough considerations of the above variables and the
research gaps identified, Deterministic Prioritization approach
is put forward as a solution to the data access problem. The
implementation of deterministic prioritization in the data
staging area expounds the relationship of the other ETL
processes. This approach will tend to manipulate data
immediately it is collected and availed to the staging area.
Less activity is experienced in the extraction phase but the
actual data work area is within the staging area. With this
approach appropriate data selection is coordinated to filter out
unwanted “dirty” data and assigning priority to the important
“clean” data that is moved to data warehouse or data marts.
The fact remains that previous activities within staging area
are important and hence the approach aims to improve on the
order of execution to avoid redundancy and repetition of
tasks. This concept is illustrated in the following diagram that
shows the relationship among the ETL processes of a data
warehouse by applying the DP rules.
Figure 2. Proposed Conceptual Framework
This implementation is in two sections i.e. deterministic
section and prioritization section as discussed below.
3.3 Deterministic Section
According to Macmillan dictionary, the term deterministic
means “using or believing in the idea that everything is caused
by another event or action and so you are not free to choose
what you do“. The deterministic approach is relying on data
selections based on the confidence levels or behavior of data
accesses in relation to previous selections. The key
characteristic of determinism is that, the output remains the
same for the same inputs. When applied to data staging, the
selection of column values from a table is based on the
distinctiveness of the data stored and frequency of previous
accesses. The most distinct values (fewer NULLS) the higher
the confidence levels to be loaded to the next stage in ETL.
The data columns with high nullity are negated from the
selection hence reducing the amount of load to the next stage.
3.4 Prioritization Section
According to Macmillan dictionary, the term prioritize means
“to decide in what order you should do things, based on how
important or urgent they are”.
The main goal of implementing prioritization is to reduce the
amount of active data being manipulated at a time by focusing
on the minimum and meaningful details. Once the data to be
loaded has been determined, then the order at which
processing occurs is vital to reduce delays in data handling.
The data selection from the raw tables needs to be prioritized
by altering the query execution plan to give more priority on
the data columns with high confidence levels. To implement
this clustered indexes are introduced based on these distinct
columns. These indexes alter the normal execution plan for
the queries hence improving performance and efficiency. This
execution plan is explained by (Grant, 2009) that “an
execution plan is the result of query optimizer’s attempt to
calculate the most efficient way to implement the request
represented by the T-SQL query...”
The query executer has an engine that optimizes query
execution on its own and by altering the data selection time is
reduced. Prioritization affects the logical aspects enforced by
the business rules to maintain the dimensional model and
giving way for faster way of retrieving data.
3.5 Included Improvements
Creating a new stable staging framework that is freely
available to everyone and run across different hardware
platforms (cross-platform) and supporting concurrent
processing. This will magnify the core benefits of having
intelligent data warehouses that are supportive to the top-level
management systems majorly due to development of staging
area.
(Stephen, 2013) elaborates their approach in Oracle
environment.” Most ETL applications use a staging area to
stage source system data before loading it into the warehouse
or marts. When implemented within an oracle environment a
partitioning strategy is usually employed such that data that is
not required any longer can be removed from the tables with
the minimum amount of effort.”
The proposed framework will have forecasting and
prioritization mechanisms to decide which data is necessary
before transfer begins hence saving on network services and
bandwidth.
Current systems such as HANA databases have high
processing capability which is meaningless if there is no
proper scheduling of resources. This can result to lots of
losses of resources not being manned properly. Hardware is
static while data is dynamic and at some point the available
hardware would not be sufficient to handle the data affecting
on performance. Eventually, Scheduling plays a vital role in

www.ijcat.com 587
the performance implications of any system. It represents the
effect being sought is measurable to make comparison.
3.6 Prioritization by indexing specific
columns
The newly distinct derived columns that were added to each
staging table (external columns; hence do not affect the data
from the source in any way to ensure consistency and
integrity), are used to create indexes as shown below.
Figure 3. Index Creation Sample
The created index named “IX_newStaging_Customer” is
prioritizing the derived unique column named
“STCustomerID” on the staging table named
“newStaging_Customer”. The created user defined index hints
the order of execution of the Data Definition Language (DDL)
queries submitted to the server hence overriding the server’s
query execution plan. The created index is also not affected in
future in case of recreations of the source data from extraction
stage since its an external derived column. Prioritization by
distinct columns enhances the efficiency and performance of
the query execution plan by the server during query search.
This results in optimized selection costs while maintaining the
quality of data to the data warehouse. The measure of
improved efficiency is shown below for a selection query.
Figure 4. New Execution plan for Customer Table
3.7 Test scenario preparation in SSIS tool
The ETL processes of a data warehouse are demonstrated
using the SSIS tool. The first scenario setup is for the current
situation before enhancement and the second scenario setup is
for the new situation after enhancement. The following is a
demonstration of the scenario used in the experiments in run
mode.
Figure 5. Test Scenario Setup.
4. RESULTS
The experiments are performed and collecting results of time
variables in a Ms Excel file. This file is generated
automatically from the scripts written in Visual Studio 2008
and C# programming language when the scenario is run. The
experiment is run for fifteen times and for each cycle it
records the time change for the different variables to the file.
Finally, the comparison is made based on these results from
both scenarios to note the impact of the change introduced as
shown below.
Table 1. Before Enhancement ETL Processes results
TestRunNum
ber
prevExtraction
Time
prevStagingT
ime
prevLoading
Time
1 63368.28 68717.21 92851.52
2 97940.12 79815.68 132989.3
3 52449.29 45874.92 62384.48
4 54358.59 87338.59 49553.42
5 49600.26 50350.14 45086.72
6 48285.48 54800.62 45312.21
7 52569.22 48670 48858.79
8 53581.86 51044.78 66801.6
9 88992.69 96697.22 60973.71
10 72884.88 56306.1 52252.09
11 58839.45 54556.57 59913
12 69740.99 58111.31 53562.38
13 59401.77 50840.68 51099.61
14 65250.34 55357.5 52560.51
15 56115.72 71712.82 58035.14

www.ijcat.com 588
Table2. After Enhancement ETL Processes results
TestRunN
umber
newExtractio
nTime
newStagin
gTime
newLoadin
gTime
newTotal
Time
1 62261.37 50691.27 4824.541 117777.2
2 59639.95 53591.02 4299.265 117530.2
3 55766.46 54468.57 3652.78 113887.8
4 54012.87 53544.93 3552.135 111109.9
5 61811.67 54836.81 3340.999 119989.5
6 59963.43 57082.69 3374.887 120421
7 63116.5 69659.48 4147.719 136923.7
8 61813.26 61603.78 3455.976 126873
9 69365.94 54521.06 3623.27 127510.3
10 56561.51 52100.49 5690.811 114352.8
11 54106.91 59241.29 3701.506 117049.7
12 59291.2 86575.51 3493.712 213390.8
13 66104.49 89692.62 4130.86 159928
14 72926.03 56987.29 5984.634 135898
15 57962.95 50812.86 3926.081 112701.9
5. DISCUSSION
The following discussion is based on the comparison of the
results obtained from the tests. Each of the ETL stage is
compared separately for the two situations and graphically
shown in the following figures to have a clear distinction of
the two situations.
5.1 Comparison of Extraction stage
The following is an illustration of the individual comparison
per stage of the ETL processes to bring out a clear view of the
improvement made. Explanation for each comparison follows
for every illustration. The negative sign indicates that it is in
the reducing direction thus showing the enhancement has
taken place by reducing the particular running time per stage.
Table 3. Extraction stage results analysis
Test prevExtractionTime newExtractionTime
Average 62891.92931 60980.30203
Change -1911.627273
Change % -3.039543061
Figure 6. Extraction stage comparison
The above illustrations show the time taken for extraction
using Deterministic Prioritization approach has reduced by
3.04% compared to the previous extraction time.
5.2 Comparison of Staging stage
Table 4. Staging stage results analysis
Test prevStagingTime newStagingTime
Average 62012.94289 60360.64505
Change -1652.29784
Change % -2.664440297

www.ijcat.com 589
Figure 7. Staging Stage Comparison
The above illustrations show the time taken in staging area
using Deterministic Prioritization approach has reduced by
2.66% compared to the previous extraction time.
5.3 Comparison of Loading stage
Table 5. Loading stage results analysis
Test prevLoadingTimenewLoadingTime
Average 62148.96424 4079.94504
Change -58069.0192
Change % -93.43521635
Figure 8. Loading Stage Comparison
The above illustrations show the time taken in loading stage
using Deterministic Prioritization approach has immensely
reduced by 93.44% compared to the previous loading time.
This is a great change and is attributed to the fact that the
number of operations talking place at this stage being minimal
and less traffic to the destination targets of the data i.e. the
data warehouses. However with increased traffic of data
especially over the network, the loading time percentage
change may further reduce. The loading process has benefited
a lot from changes in previous stages.
6. SUMMARY OF FINDINGS
There has been a significant change for every stage of the
ETL processes after implementation of deterministic
prioritization approach. Moreover, the subsequent impact of
this improvement has been depicted at the loading stage
where the rate of transfer improved by a large margin of
93.44%. This shows that any change occurring in the staging
area will definitely affect the entire process flow. However, if
more time is taken in proper design of the data staging area,
then there are high expectations for further improvements on
performance beyond the ones achieved by the researcher.
Based on the limitation of resources and the specifications of
the workstation used to run the tests, the performance levels
are considered acceptable for an organization with fast
growing data sets. However, the setup is limited by the
resources at run time but with increased disk space as
provided by the owners as data size grows, then the total data
access time in the data warehouse will be highly improved.
The choice of data prioritization mechanism using indexes for
columns is supported by (Cecilia and Mihai, 2011), where
they state the use of Indexes on database queries improves the
performance of the whole system. Clustered indexes perform
better than nonclustered indexes when the expected returned
records are many and should be set for the most unique
column of a table. This proposition goes in line with the
research with the use of clustered indexes in the staging area
mainly due to the need to fetch and process large data.
(Costel G et al, 2014) did a research on query execution and
optimization in the MSSQL Server and put across the missing
of indexes as a contributor to low performance of query
execution. They inform that when a table misses indexes, the
search engine has to parse through the entire table step by step
to find the searched value. The resources spent on this process
are enormous and considerably increases time to execute the
queries.
(Grant, F. 2012) explains about execution plan management
done by query optimizer. The database relational engine
performs logical reads within the cache memory while the
storage engine performs physical reads directly from disk.
Improvements are highly realized mostly for data
manipulation language statements since the engine needs to
parse the query for correctness. The SQL server generates
statistics against the indexes and sends them to the optimizer
to determine the execution plan.
(El-Wessimy et al, 2013). Similarly did an enhancement in the
data warehouse staging area by using different techniques
(FIFO, MC,RR time and record rotation) targeting the loading
phase. The tests ran captured the time taken to transfer data in
each stage of the ETL process and suggest the most suitable
technique. They did a comparison amongst all techniques and
noted that FIFO performed better for less data set while
Record Limit Based Round Robin was best for large data sets.
Their research supported further reduction of overall time
taken to deliver data from source to destination. The
uniqueness of this study is the ability to handle large data sets
from the beginning as well as newer inclusions of data from

www.ijcat.com 590
the sources without any readjustments of the system structure
setup.
7. CONCLUSION
The adoption of Deterministic Prioritization approach in the
staging area has shown promising results and the users at the
presentation level of the data warehouse are rest assured of
fast data access and retrievals. They can timely make decisive
conclusions and reports based on current data that is made
available in a timely fashion similar to real time systems.
They will also benefit to wide range of data availed since the
ETL processes aim to denormalize the data comprehensively
before its delivered to users. This forms an association that is
deterministic in nature for future priority loads. The
researcher was keen to avoid compromising data quality for
high performance gains and this resulted in a more balanced
system setup where the data still meet the qualitative
assurance defined by the users’ requirements.
8. RECOMENDATIONS AND FUTURE
WORK
In view of the results and findings of the experiments
undertaken in this research, the researcher recommends the
incorporation of the deterministic prioritization approach in
the design and development phase of the data staging
frameworks. This is so because it is cross-platform to all
database management systems that support the SQL language
in the market today. The change gives room for further
customization since it only happens at the design and before
query execution. Notably for well formed queries the
performance will be even better.
The data staging area is an area which has not been researched
exhaustively and the impact of high resources should be
considered as a next check on performance gain over cost.
The setup experiments are carried based on same format of
data sources and further studies should be carried out on
variant data sources and using different staging framework
other than the SSIS tool used here. The scenario in this test is
demonstrated in a local workstation and it would be essential
to note the performance levels in a distributed system with
both sources and targets widely separated by networks.
9. REFERENCES
[1] Abbasi, H., Wolf, M., Eisenhauer, G., Klasky, S.,
Schwan, K., & Zheng, F. (2010). Datastager: scalable
data staging services for petascale applications. Cluster
Computing, 13(3), 277-290.
[2] Akkaoui, Z. E., Munoz, E. Z. J.-N., and Trujillo .J. A.
(2011). Model-Driven Framework for ETL Process
Development. In Proceedings of the international
workshop on Data Warehousing and OLAP. pp. 45–52
Glasgow, Scotland, UK.
[3] Aksoy, D., Franklin, M. J., & Zdonik, S. (2001). Data
staging for on-demand broadcast. In VLDB (Vol. 1, pp.
571-580).
[4] B´ezivin, J. (2005). On the unification power of
models.Software and System Modeling, 4(2):171–188
[5] Cecilia, C., Mihai, G. (2011). Increasing Database
Performance using Indexes. Database Systems Journal
vol. II, no. 2/2011. Economic Informatics Department,
Academy of Economic Studies Bucharest, Romania.
[6] Costel, G.C., Marius, M. L., Valentina, L., Octavian, T.
P.(2014). Query Optimization Techniques in Microsoft
SQL Server. Database Systems Journal vol. V, no.
2/2014. University of Economic Studies, Bucharest,
Romania.
[7] Da Silva, M.S.,Times, V.C., Kwakye, M.M. (2012).
Journal of Information and Data Management.3 (3).
[8] Deterministic. (2009-2015). In Macmillan Dictionary.
Macmillan Publishers Limited: Accessed from:
www.macmillandictionary.com on 20th
July 2015.
[9] Eckerson, W., & White, C. (2003). Evaluating ETL and
data integration platforms. Seattle: The DW Institute.
[10] El-Wessimy, M., Mokhtar, H. M., & Hegazy, O. (2013).
ENHANCEMENT TECHNIQUES FOR DATA
WAREHOUSE STAGING AREA. International Journal
of Data Mining & Knowledge Management Process,
3(6).
[11] Erhard, R., and Hong, H.D. (2000). Data Cleaning:
Problems and Current Approaches. Journal IEEE Data
Eng. Bull.23 (4), 3-13.
[12] Grant, F.(2009).The Art of High Performance SQL
Code: SQL Server Execution Plans. Simple-Talk
Publishing. ISBN 978-1-906434-02-1
[13] Grant, F. (2012). SQL Server Execution Plans. Second
Ed. ISBN: 978-1-906434-92-2. Simple Talk
Publishing.
[14] El-Wessimy, M., Mokhtar, H. M., & Hegazy, O. (2013).
ENHANCEMENT TECHNIQUES FOR DATA
WAREHOUSE STAGING AREA. International Journal
of Data Mining & Knowledge Management Process,
3(6).
[15] Firestone, J. M. (1998). Dimensional modeling and ER
modeling in the data warehouse. White Paper No, Eight
June, 22.
[16] Flinn, J., Sinnamohideen, S., Tolia, N., &
Satyanarayanan, M. (2003). Data Staging on Untrusted
Surrogates. In FAST (Vol. 3, pp. 15-28).
[17] Inmon, W. H. (2002). Building the Data Warehouse.
Wiley.
[18] Inmon, W. H. (2005). Building the data warehouse. John
wiley & sons.
[19] Kimball, R., & Caserta, J. (2004). The data warehouse
ETL toolkit. John Wiley & Sons.
[20] Kimball, R., Reeves, L., Ross, M., & Thornthwaite, W.
(2008). The Data Warehouse Lifecycle Toolkit, 2nd ed.
Practical Techniques for Building Data Warehouse and
Business Intelligence Systems.
[21] Muller, P. A., Studer, P., Fondement, F., & Bézivin, J.
(2005). Platform independent Web application modeling
and development with Netsilon. Software & Systems
Modeling, 4(4), 424-442.
[22] Per-Åke, L., Cipri, C., Campbell, F., Eric, N. H.,
Mostafa, M., Michal, N., Vassilis, P., Susan, L. P.,
Srikumar, R., Remus, R., Mayukh,
S.(2013).Enhancements to SQL Server Column Stores.
ACM 978-1 -4503-2037-5/13/06. New York, USA.
[23] Prioritize. (2009-2015). In Macmillan Dictionary.
Macmillan Publishers Limited: Accessed from:
www.macmillandictionary.com on 20th
July 2015.
[24] Ralph, K. and Margy, R. (2002). The Data
WarehouseToolkit: The Complete Guide to Dimensional
Modeling. Second Edition. Published by John Wiley and
Sons, Inc. Canada.

www.ijcat.com 591
[25] Russom, P. (2012). BI Experts: Big Data and Your Data
Warehouse's Data Staging Area. TDWI Best Practices
Report, Fourth Quarter. Retrieved from
http://tdwi.org/articles/2012/07/10/big-data-staging-
area.aspx
[26] SAP AG. (2002). Business Information Warehouse –
Data Staging Retrieved from
http://scn.sap.com/docs/DOC-8100.
[27] Stephen, B. (2013). Staging, Statistics & Common
Sense: Oracle Statistics Maintenance
[28] Strategy in an ETL environment Retrieved from
http://www.seethehippo.com/
[29] Vassiliadis, P. (2009). A survey of Extract–transform–
Load technology. International Journal of Data
Warehousing and Mining (IJDWM), 5(3), 1-27.
[30] Zineb, A., Esteban, Z., Jose-Norberto.M., Juan, T.
(2011). A Model-Driven Framework for ETL Process
Development. In DOLAP 11 Proceedings of the ACM
14th international workshop on Data Warehousing and
OLAP Pages 45-52. ACM New York, NY, USA.
ISBN: 978-1-4503-0963-9

Enhancing Data Staging as a Mechanism for Fast Data Access

More Related Content

What's hot (14)

Viewers also liked (18)

Similar to Enhancing Data Staging as a Mechanism for Fast Data Access (20)

More from Editor IJCATR (20)

Recently uploaded (20)

Enhancing Data Staging as a Mechanism for Fast Data Access