Database, Open Source

SQL Server 2012 – Cheaper than Open Source Database options

On a recent project we were tasked to review and recommend a database technology platform for a new Data Warehouse environment. This quickly turned into a database evaluation not just for the new Data Warehouse, but also for all future applications. The client was currently on Sybase, but was open to the most cost-effective option. The client also had experience with Open Source technologies in the past for both server operating systems (linux) and reporting solutions. (JasperSoft)

The criteria we specified had cost and performance as two important factors. Since we had limited experience with any Open Source databases in a Data Warehouse environment, we felt it would be useful to evaluate the options from a cost, functionality, and performance point of view.

This post is a recap of that process and of the somewhat surprising recommendation.

The Candidates

Since the client already was using Sybase, it was a no-brainer to include that one. It addition to Sybase, we included SQL Server and Oracle from the commercial DBMS’s due to their market share and existing use at the client. DB2 was excluded due to DB2 not currently being used at the client and the perceived higher cost of implementation of DB2. (Rightly or wrongly) It was felt that either Oracle or SQL Server could represent the commercial DBMS’s.

On the Open Source side, the DBMS’s MySQL and PostgreSQL were selected. MySQL was selected due to its large adoption in the Open Source community and recognition for performance. PostgreSQL was selected also for its recognition in the Open Source community for robustness and performance. MySQL and PostgreSQL were the only two Open Source DBMS’s as it was felt they were the clear leaders in the Open Source DBMS’s.

After performing detailed performance tests between MySQL and PostgreSQL, PostgreSQL was chosen as the Open Source DBMS option to compare against commercial DBMS’s. This was due to the fact that MySQL had significantly poorer performance when using the InnoDB engine that guaranteed ACID compliance and provided referential integrity. MySQL has an interesting piece of functionality to allow you to use different database engines that provide different functionality and types of processing. Unfortunately the engine used in many of the published benchmarks do not provide the functionality required by standard enterprise DBMS solutions.

The Requirements

The current Data Warehouse requirements were reviewed and it was determined that the vast majority of the reporting requirements were operational reports delivered in text format. This is probably not unlike many companies out there that are early in the use of Data Warehouses. This is not to preclude the use of more analytical functions in the future, but that was not the use of the Data Warehouse currently.

The current Data Warehouse also stored data in the 10’s of gigabytes. So although this was a considerable amount of data, it also did not require advanced Data Warehouse appliances or Big Data solutions. Traditional Data Warehouse architecture and designs would be sufficient for now and well into the future.

The requirement was that the Open Source solutions must be supported. For example, this meant that we would require support from EnterpriseDB if we chose PostgreSQL. I propose this is pretty typical for larger companies evaluating Open Source.

The Architecture

Due to the current requirements and expected future requirements, the recommended Data Warehouse solution architecture was designed to incorporate both an Operational Data Store and a Star Schema Data Warehouse. The Operational Data Store would be used to generate pure operational reports and would be loaded at least daily. The Star Schema Data Warehouse would be used to generate more analytical reports and would be loaded on a weekly basis.

The Evaluation

The recommended configuration for the evaluation was for two servers each with two Quad core CPUs. This was due to the current workload and the expected increased workload in the future. We did evaluate higher levels of configuration to ensure the cost was linear for future growth.

This was not a small Data Warehouse and is probably typical of most enterprises as an entry into Data Warehouse technology.

We separated our cost evaluation into two sections:

1)       Initial Cost

2)      Ongoing Cost

Initial Cost

Initial cost incorporated the server and DBMS license fees for both development and production environments and also two human factors:

Training – What is the cost of the initial training required for the administrators.

Efficiency – What is the expected efficiency for the administrators in the first three months.

The results can be seen below:


As you can see, the supported PostgreSQL option was significantly more affordable as far as initial cost.

Annual Cost

The Annual cost incorporated the following two factors:

License Support costs – These costs are the support and maintenance costs for both the Production and Development environments.

Estimated Annual Maintenance Costs – It is estimated that there will be additional ongoing effort to maintain the different DBMS solutions as compared to the current Sybase solution. This is due to the different complexities of the DBMS solution, the available tools for the DBMS solution, and the available support in the DBMS user communities.

  Estimated Annual Maintenance Rating Rationale
Sybase ASE

0%

No additional maintenance effort is required due to the fact that Sybase is currently used
Microsoft SQL Server 2012 Enterprise

5%

No additional maintenance effort is required due to the fact that Microsoft has an excellent set of tools and resources and the environment is similar to Sybase which is currently used

Baseline 5% additional cost has been added to account for the fact that a second DBMS must now be learned and supported.

Oracle 11

20%

Additional maintenance effort of 15% is estimated due to the complexity and toolset for Oracle.

Baseline 5% additional cost has been added to account for the fact that a second DBMS must now be learned and supported.

PostgreSQL – supported

10%

Additional maintenance effort of 5% is estimated due to the simplicity of the PostgreSQL solution and the relatively limited toolset and resources.

Baseline 5% additional cost has been added to account for the fact that a second DBMS must now be learned and supported.

The results can be seen below:

As you can see, the annual ongoing costs end up being in favour of SQL Server.

The 10 year Total Cost of Ownership (TCO) is:

DBMS Cost
Sybase ASE

$1,173,057.20

Microsoft 2012 Enterprise

$1,019,909.00

Oracle

$4,040,910.00

PostgreSQL – supported

$1,074,025.00

Although SQL Server has an initial license cost, the ongoing support costs are less than supported PostgreSQL in almost all categories.

Conclusion

Over a ten-year period, the TCO for SQL Server is less than a supported version of PostgreSQL. This also includes the server licensing costs. As an added benefit, our solution also required a Reporting Solution and an Extract, Transform, and Load solution – SQL Server provides both of these solution bundled with the DBMS for no additional cost.

From a cost perspective, SQL Server is the clear winner.

For our situation, although cost was important, it was not the overriding factor. In fact, cost accounted for only 10% of the entire weight. The factors and their weights are listed below:

Evaluation Criteria

Weight

MBC Required Functionality

350

Cost

100

Future Commercial Viability

100

High Availability

50

Scalability

50

Product Track record/Future stability

50

Technical Functionality

350

Architectural Standards and Connectivity

50

DBMS standard functionality

100

Data Warehouse standard functionality

150

Security/Encryption

50

DBMS Management Functionality

50

Ease of Maintenance/Management

30

Database Configuration

20

People and Future Flexibility

50

Future Migration/Flexibility

20

Standard Technology/People Availability

30

Performance Functionality

200

Bulk Load Performance

20

Bulk Update Performance

20

Bulk Delete Performance

10

Basic Query Performance

20

Cross Product Query Performance

20

Sub Query Performance

20

Parallel Query Performance

20

Analytical Query Performance

20

Indexed Update Performance

20

Indexed Delete Performance

20

Non-Indexed Update Performance

10

Total

1000

After all the factors were evaluated, SQL Server finished 196 points ahead of PostgreSQL.

SQL Server was selected and I was enlightened. Although people assume Open Source DBMS’s are the most inexpensive options, SQL Server is the leader once you include other factors. In fact, SQL Server was the leader in both cost and functionality.

SQL Server Integration Services (SQL Server’s ETL solution) and SQL Server Reporting Services (SQL Server’s Reporting solution) were NOT included in this comparison. If they were, the results would have been even more pronounced.

I encourage you to do your own evaluations. You may be surprised at the results. I know I was.

Please let me know if anyone is interested in specific evaluation criteria. I’d be happy to share our rationale and SQL performance scripts.

About Terry Bunio

Terry Bunio has worked for Protegra for 14+ years because of the professionalism, people, and culture. Terry started as a software developer and found his technical calling in Data Architecture. Terry has helped to create Enterprise Operational Data Stores and Data Warehouses for the Financial and Insurance industries. Along the way Terry discovered that he enjoys helping to build teams, grow client trust and encourage individual career growth, completing project deliverables, and helping to guide solutions. It seems that some people like to call that Project Management. As a practical Data Modeller and Project Manager, Terry is known to challenge assumptions and strive to strike the balance between the theoretical and real world approaches for both Data Modelling and Agile. Terry considers himself a born again agilist as Agile implemented according to the Lean Principles has made him once again enjoy Software Development and believe in what can be accomplished. Terry is a fan of Agile implemented according to the Lean Principles, the Green Bay Packers, Winnipeg Jets, Operational Data Stores, 4th Normal Form, and asking why

Discussion

2 thoughts on “SQL Server 2012 – Cheaper than Open Source Database options

  1. One of the links is broken, but I am curious as to the breakdown of the costs, specifically the PostGres and Sql server.

    Posted by RealPage CO Tom Ballinger | September 22, 2016, 2:12 pm

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: