cloud database management system
TRANSCRIPT
Cloud Database Management System
IS508E group NO.6 project
Group NO.6 members:
PENG Yu
KALAI Kumaraguru
KUTTIKKAT VENUGOPAL Sreehari
Contents
General business case ............................................................................................................... 1
Introduction and problems ....................................................................................................... 1
The existing technology ............................................................................................................ 2
Challenges of Implementation: ................................................................................................. 9
Reference: ............................................................................................................................... 10
1
General business case
Thanks to the successful management, the business of the company expands very
fast. However, the database management system cannot withstand the quickly and
greatly increased work load, the break-down frequency increases, which make the
decision-making efficiency and customer experience drop. The company tries to turn
around this bad trend and poses two solutions: one is to update the present DBMS
including hardware, software and human resource; the other one is to make use of
SQL Azure from Microsoft to set up cloud environment and transfer the DBMS there.
After cost analysis, we find that to reach the same efficiency standard, the cost of
first solution is 9 times of the second solution, and the second solution has merits
such as capability, flexibility and elasticity, so we suggest implementing the second
solution: Cloud DBMS, SQL Azure.
Introduction and problems
We are an express company in North America. We transport the parcels for
customers. We have good performance and with the blooming online shopping, our
business expands very fast. Recently, a large online retailer just outsources its express
service to us. Our daily express order increases by 2 times more than before. Our IT
department, where there are three employees. They are in charge of maintaining the
database management system (DBMS) and the website providing order reacceptance
and parcel tracing service. Because of the expansion of business and the outsourcing
contract, the requests for placing orders and tracing parcels on the website soar 4
times than before. The present web server cannot support such huge data stream,
sometimes it crashes, and this influences the clients experience and the business
terribly. In front of these greatly increased clients, parcels and tracking information
needs, our present database management system is not responsive and powerful
enough to meet the searching, ordering, processing and indexing requirement.
Now there are two solutions to these problems. One solution is to buy software,
update the hardware and recruit more employers to strengthen the present DBMS;
however, this is another problem brought by this solution. The online shopping is
seasonal fluctuating, which means that in official holidays, such as the national day,
thanks giving day, charismas day, New Year day and so on, the online shopping
activities are much more intensive than usual. Correspondently, the express orders
and requests for parcel tracing also rise greatly. Here is the problem, how should we
equip the present database management system, to meet the peak amount of
requests or to meet the usual amount of requests. If we choose to meet the peak
amount of requests, the investment will sure much bigger than that of meeting the
normal amount of requests, and when the holiday passes, the capacity of system will
2
be left unused, this is a waste of resources and investment; if we choose to meet the
usual amount of requests, the limited capacity of system during the holidays will
cause bad customer experience, which will bring the loss of orders and clients,
moreover, the profit of the orders in holidays takes big proportion of the annual
profit, and this makes the result worse.
The other solution is to make use of the cloud computing technology, which means
to set up DBMS on the cloud computing platform, and move all the present data to
cloud DBMS. The cloud DBMS service is scalable and flexible, it can adjust the
capability according to our needs, and then we do not have to face the dilemma
system configuration problems.
After talking with the responsible of IT department, we make a consensus that the
cloud DBMS is a better choice than the in-house DBMS. Not only the cloud DBMS will
give more powerful performance, it requires less labor support and it simplifies the
future investment and improvement process for the DBMS. Then we will present the
cloud DBMS technology and make a comparison between cloud DBMS and in-house
DBMS. We will also show the advantage and disadvantage of cloud DBMS, the cost
and time to implement. An analysis of the cloud service market will report for the
sake of choosing a proper cloud service supplier. The challenges in the
implementation will not be neglected.
The existing technology
DBMS
A DBMS is the software that is installed on a physical server and stores and manages
databases. It can run databases from multiple software. For example, it could
manage the databases for the accounting system, the CRM and the ERP. Each system
would then be on a separate database in the DBMS. The main objective of a DBMS is
to provide error-free data and to keep a good level of performance. Since databases
need live information, the modern DBMS are very powerful in managing a lot of
users and processing huge amount of data.
Cloud DBMS
Cloud database management system (CDBMS) is a distributed database that provides
computing as a service rather than a product. It is the distribution of resources,
software, and information between multiple devices through a network which is
commonly the internet. It is expected that this number will grow significantly in the
future.
3
Benefits of cloud database over traditional database
Lower upfront costs
Cloud computing significantly reduces capital expenditure of a company to deploy
DBMS application. The part of the upfront costs reduced mainly includes expenses
for hardware, user licenses, and implementation, excluding user training and
customization.
Lower operating costs
Cloud-based DBMS reduces operating costs for energy, maintenance, configuring,
upgrades, and other IT staff costs and efforts. This benefit is generally considered as
a growing significance for SMEs using traditional DBMS.
Rapid implementation
Rapid implementation is generally agreed to be among the best advantages of
cloud-based DBMS. It could also help to facilitate cloud services providers in reducing
the time to deliver new products in certain types of businesses.
Scalability
Resource pooling and rapid resource elasticity of cloud-based DBMS make the
infrastructure capacity highly elastic. That in turn enables faster time to market, high
level of strategic flexibility and improved competitiveness. This feature is reported as
a possible advantage that is particularly relevant for SMEs in competing with large
rivals.
Focus on core competencies
DBMS cloud and other business applications allow focusing resources that would be
used to maintain an IT department in other key areas of the business. Thus it reduces
pressure of internal IT department which enables it to focus on servicing core
competencies.
Access to advanced technology
Cloud-based applications often enable access to specialized technology and
advanced computing resources that otherwise would not have been accessible to
SMEs.
Rapid updates & upgrades
Cloud-based DBMSs usually get faster updates and new functionality than traditional
DBMSs
Improved accessibility, mobility, and usability
Unlike Traditional Database system, Cloud based database applications could be
easily accessed through any internet connected device, therefore it is accessible
anywhere any time.
4
Improved system availability and disaster recovery
In many cases SaaS providers ensure measures such as backup routines, fallback and
recovery procedures, conditioned power etc. of higher quality than most SMEs do
in-house.
Disadvantages (shortcomings) of traditional database system
High cost
Installing new database system may require investment in hardware and software.
The DBMS requires more main memory and disk storage. Moreover, DBMS is quite
expensive. Therefore, a company needs to consider the overhead cost of
implementing a new database system.
Training new personnel
When an organization plans to adopt a database system, it may need to recruit or
hire a specialized data administration group, which can coordinate with different user
groups for designing views, establishing recovery procedures, fine tuning data
structures to meet the organization requirements. Hiring such professionals is
expensive.
Explicit backup and recovery
A shared corporate database must be accurate and available at all times. Therefore, a
system using online updating requires explicit backup and recovery procedures.
System failure
When a computer system containing the database fails, all users have to wait until
the system is functional again. Moreover, if DBMS or application program fails a
permanent damage may occur to the database.
Cost, time, feasibility estimates and justification of investmen
This part will present the cost, time and feasibility of in-house DBMS and cloud DBMS,
and compare them to make the justification of the final choice, cloud DBMS.
In order to decide the cost of strengthening the in-house database management
system (in-house DBMS), there are several elements need to be concerned.
1. The infrastructure cost. For the servers, we will choose 2 HP ProLiant BL460c
Gen8 E5-2640v2 2.0GHz 8-core 2P 48GB-R P220i/512 FBWC servers to avoid
system breaking down, each one cost 4314$, so the total cost is 8628$; For the
hard drive, we choose Seagate expansion 4TB, 3 pieces, 154$ for each, so the
total cost is 462$; For the new recruited 3 IT employees, we will prepare new
5
desk top computers for each one; We choose DELL Optiplex 3020, each costs
519$, the total cost is 1557$. The infrastructure cost for DBMS is 10647$. When
we have these hardware, they will cost 20% of the initial price as the
maintenance fee, which is 2129.4$/year.
2. The cost of new database management system (DBMS). There are several
mainstream DBMS on the market, IBM’s DB2, Microsoft’s SQL Server and Oracle’s
Oracle. There is no need to compare the technical difference among those three
DBMS, because all these main engines have been around for many years and
have undergone constant development. If any of these were truly incompetent in
normal usage, it would long ago have been weeded out. That's another way of
saying that all of these engines are highly competent; more than capable of
running 90 to 95% of the database applications that we are likely to throw at
them. So, we will assume to choose Microsoft’s SQL Server, because its relatively
lower price and it’s more compatible with our present operating system,
Windows. We take SQL Server 2012 standard edition. There are two licensing
models, per processor licensing and server/CAL licensing. As our IT department is
small and they are the only persons in the company who need access to
manipulate the DBMS, and including the new recruited 3 employees, there are 6
employees in IT department, so we choose server/CAL model. What’s more, each
server running SQL Server requires a license, so 8 Licenses are needed, each
license cost 898$, the total cost is 7184$. The license need to be renewed each
year to get the new features, the renewal fee is 25% of the license price, so each
year we have to pay 1796$ to renew the license annually.
3. Human resource cost. As it is referred above, the present three employees are
not capable of maintaining the DBMS, not even saying after the new equipment
arrives. More IT employees are needed to strengthen the efficiency of the
department. Three more software development engineers are planned to be
recruited; the national average salary of this post is 63000$ per year, and we take
this average figure, the human resource cost will be 189000$ per year.
4. Time consumed of implementation and integration of new software, new
hardware and new staff. The lead time of software and hardware can be 1 week,
and to install, adjust and test them may take 3 weeks, the recruitment process
can last for 1 month. Time also needed for the new employees to get familiar
with the whole information system of the company, it will take 3-4 weeks in an
urgent situation. If we begin the preparation activities, such as ordering software,
hardware and recruiting IT engineers, at the same time, the whole process will
take 2 months.
Here is a summary of in-house DBMS solution:
6
Hardware cost DBMS software
cost
Human
resource cost
Time cost
In-house
DBMS
10647$ first buy,
2129.4$/year
after
7184$ first time,
1796$/year after
189000$/year 2 months
For the cloud database management systems management (cloud DBMS), we will
compare the same elements that we do in the in-house DBMS.
1. Hardware cost. Because we choose cloud DBMS service, which means to choose
a cloud service provider, we would occur a monthly payment; we would not need
to purchase the hardware itself nor would we need to purchase the windows
servers operating system. And we would also avoid the extra 20% maintenance
fee of hardware and software discussed previously, because all the activities
related to this item are performed by the cloud service provider;
2. Cloud DBMS cost. Here we have two choices, Amazon RDS and SQL Azure. Both
of them are among the biggest cloud DBMS service providers in the world. As
both SQL Azure and SQL Server are products of Windows, they are compatible
very well. Recently, Amazon RDS also announced the support to SQL Server. Then
we will compare the monthly pricing of Amazon RDS and SQL Azure respectively.
SQL Azure pricing:
Service Price Unit Cost
SQL Server License(SQL Azure) 0 0 0
10GB instance 45.96 per month 1 45.96
Storage 0.1 per month per GB 10 1
Tx inbound(ingress) 0 0 0
Tx outbound(egress) 0.1 Per GB per month, first5GB free 10 0
Total Estimated Database Cost 46.96
10 Web Role Instances(Web) 0.12 per hour 7200 864
2 Web Role Instances 0.24 per hour 1440 345.6
Tx outbound to Internet 0.12 per GB per month, first 5GB free 1000 119.4
Total Estimated Web & Middle-Tier Servers 1329
Reporting Services 0.16 per GB 720 115.2
Total Estimated Reporting Services Cost 115.2
Total Monthly Cost(SQL Azure) 1491.16
Amazon RDS pricing:
7
Service Price Unit Cost
SQL Server License 0 0 0
On-Demand SQL Server RDS
Small Instance 0.65 per hour 720 468
Storage 0.1 per month per GB 10 1
I/O Rate 0.1 per million requests 1 0.1
Tx inbound(ingress) 0 10 0
Tx outbound(egress) 0.12 Per GB per month, 1st GB free 10 1.08
Total Estimated Monthly Database Cost 470.18
10 EC2 Small On-Demand Instances(Web) 0.115 per hour 7200 828
2 EC2 Medium On-Demand Instances (Middle-Tier) 0.23 per hour 1440 331.2
Tx outbound to Internet 0.12 per GB per month, 1st GB free 1000 119.88
Total Estimated Monthly Web & Middle-Tier Servers 1279.08
SQL Server 2012 License (express edition) 0 0 0
1 EC2 Small On-Demand Instance 0.115 per hour 720 82.8
Tx outbound to Internet 0.12 per GB 1 0.12
Total Estimated Monthly Reporting Services Cost 82.92
Total Monthly Cost(Amazon RDS) 1832.18
To compare the pricing in the tables above, SQL Azure offers a less expensive
service with a price advantage of about 18%. The total yearly cost of SQL Azure is
17898.92$. However, we still need to buy the license of SQL Server. In order to
deploy in cloud environment, the enterprise edition is needed, the price is
6874$ with the license renewal fee 1718.5$. It is interesting to notice that the
major difference of price is relate to the database platform price, which is
monthly database price, as shown in the graph below. Because both SQL Azure
and SQL Server are products of Microsoft, Microsoft is able to offer a great price
break. There is also no doubt that SQL Azure will be compatible with SQL Server.
From the economic and function aspect, SQL Azure is a better choice than
Amazon RDS.
8
3. Human resource cost. Because we do not buy the servers or exterior storage
device, there is no need to recruit extra IT engineers to test and maintain this
equipment. After the implementation of SQL Azure, the efficiency of IT
department will be hugely increased and the work load will not be as heavy as
before, because the cloud service provider will be in charge of the maintenance
of DBMS. So there is no human resource cost for this solution;
4. Time cost. As there is no hardware to buy, to install or to test, the
implementation time will largely decrease. When we get SQL Server and SQL
Azure, only installation, configuration and data migration are needed, these
activities will be finished in one week, and it will take another week for the
present IT employees to get used to the new DBMS, so the total time cost is two
weeks.
Here is the summary of cloud DBMS solution:
Hardware cost DBMS software
and cloud cost
Human
resource cost
Time cost
Cloud
DBMS
0$ 24772.92$ first time,
19617.42$/year
after
0$ 2 weeks
If we merge the summary tables of the two solutions, it is easy to tell which solution
is better:
Hardware cost DBMS software
cost
Human
resource cost
Time cost
In-house
DBMS
10647$ first buy,
2129.4$/year
7184$ first time,
1796$/year after
189000$/year 2 months
0
200
400
600
800
1000
1200
1400
Monthlydatabse cost
Monthlyservers cost
Monthlyreporting cost
Amazon RDS
SQL Azure
9
after
Cloud
DBMS
0$ 24772.92$ first time,
19617.42$/year
after
0$ 2 weeks
To reach a more practical and continuous vision, we make three-year cost analysis as
following:
Year 1 Year 2 Year 3 Total
In-house
DBMS
206831$ 192925.4$ 192925.4$ 592681.8$
Cloud DBMS 24772.92$ 19617.42$ 19617.42$ 64007.76$
The three-year total cost of In-house DBMS is 9 times of that of Cloud DBMS.
Although the human resource cost takes a big proportion, it is evident that the
company has to recruit more IT engineers when the devices increase, the company
has no other choice but to do that.
Challenges of Implementation:
The speed of data transfer is a barrier to the performance of the cloud database.
This affects the performance of the cloud database. The requests sent to the
database are very fast, but the time taken to save data from data center depends on
the speed of the internet. The solution to this challenge is that to have faster speed
cables, but that cost very high and the purpose of having a cloud database will
unwanted.
The business workload and the query work load. In query workload, it depends on
the number of queries, and it is not known how many users will be there who will be
creating queries to the database. When In the business workload, we can get an
estimate about the time that will be required while on the other pointer, we cannot
estimate about the time of query workload.
Challenges of big data. About previous years ago, most of the data composed by
organization consisted of transaction data that could appropriate into rows and
columns of relational database management system. Later then, there has been an
explosion of data from web traffic control, email messages and social media as well
as machine generated data from sensor. These data may be unstructured or semis
structured and are not suitable for relational database products that organize data in
the form of rows and columns. So now one solution is that we use the word big data
to define these dataset with capacities so huge that they are outside the ability of
typical DBMS to store, analyze and capture.
10
Privacy and security. Privacy has been the most important issue when it comes to
cloud database software as a service. The cloud computing is a more advanced in
terms of the availability to the users and hackers who like to interruption into the
system. The privacy in the cloud database is the very important thing that saves the
record of the customers of the companies. The companies cannot afford to outflow
out the information that is stored in their database. If there is encryption of data in
database, then it is quite easy to store in a secure way.
Reference:
1. Aljaz Zrnec, Dejan Lavbic, ‘Comparison of cloud vs. Tape backup performance and
cost with Oracle Database’, JIOS, VOL. 35, NO.1 (2011);
2. Yvette E. Gelogo, Sunguk Lee, ‘Database Management System as a Cloud Service’,
International Journal of Future Generation Communication and Networking, Vol.
5, No. 2, June, 2012;
3. Daniel J. Abadi, ‘Data Management in the Cloud: Limitations and Opportunities’,
IEEE Computer Society Technical Committee on Data Engineering, 2009;
4. Mike Chapple, ‘Microsoft SQL Server 2012: Choosing the Correct Edition’,
http://databases.about.com/od/sqlserver/a/sql_server_2012_licensing.htm;
5. Herve Roggero, ‘Sample Pricing Comparison: On-Premis vs. Private Hosting vs.
Cloud Computing’,
http://geekswithblogs.net/hroggero/archive/2013/02/25/sample-pricing-compar
ison-on-premise-vs.-private-hosting-vs.-cloud-computing.aspx;
6. HP, ‘Servers’,
http://h71016.www7.hp.com/ctoBases.asp?oi=E9CED&BEID=19701&SBLID=&Pro
ductLineId=431&FamilyId=3530&LowBaseId=38904&LowPrice=%242%2C057%2E
00;
7. John Treadway, ‘Amazon RDS vs. SQL Azure: The Birth of the DBMS Utility‘,
http://cloudbzz.com/2009/10/27/amazon-rds-vs-sql-azure-the-birth-of-the-dbms
-utility/;
8. Paul DeGroot, ‘Why Your Microsoft Enterprise Agreement Renewal Costs May
Explode Over The Next Three Years’,
http://70.32.90.244/blog/microsoft-enterprise-agreement/microsoft-enterprise-
agreement-negotiations/why-your-microsoft-enterprise-agreement-renewal-cost
s-may-explode-over-the-next-three-years/;