Time Savings and Ease of Deployment Comparison Study Oracle

Time Savings and Ease of Deployment Comparison Study Oracle
Time Savings and Ease of Deployment
Comparison Study
Oracle Database Appliance
vs.
Microsoft SQL Server
January 2012
625 Avenue of the Americas, 2nd Floor
New York, NY 10011
www.orcinternational.com
Table of Contents
Executive Summary ...................................................................................................................................... 3
Study Background ......................................................................................................................................... 5
Testing Methodology ..................................................................................................................................... 6
Comparison Tests ......................................................................................................................................... 7
Test: Installation and Initial Deployment ................................................................................................... 7
Test: Maintenance .................................................................................................................................... 9
Test: Support .......................................................................................................................................... 11
Conclusion................................................................................................................................................... 13
Tables
Table 1 - Summary of Testing Results .......................................................................................................... 3
Table 2 - Testing: Installation and Initial Deployment Tasks ........................................................................ 7
Table 3 - Testing: Maintenance Tasks .......................................................................................................... 9
Table 4 - Testing: Support Tasks ................................................................................................................ 11
2
Executive Summary
As the amount and importance of corporate data grows, companies of all sizes are finding that they
increasingly need to deploy high-availability database solutions to support their business-critical
applications. High DBA workloads have made the ease with which these systems can be installed,
administered and supported an important consideration in their selection.
The purpose of this study was to conduct comparison testing between Oracle Database Appliance and
Microsoft SQL Server in order to determine the productivity differences between these two solutions. A
significant portion of the cost of owning a DBMS is the cost in time and dollars of performing daily
management. This study focused on Oracle’s answer to these issues with the recent addition to its family
of engineered systems, Oracle Database Appliance. This two-node clustered database server seeks to
reduce, and even remove, the primary cost factors of installation and initial deployment, maintenance,
and support. ORC had a unique opportunity to evaluate this product in its early rollout stages and
measured the amount of time savings this appliance gives back to an administration team, while still
maintaining high availability within a database infrastructure.
ORC’s research determined that a DBA would save 835 hours in the first year of system implementation
using Oracle Database Appliance rather than SQL Server, and 669 hours in each subsequent year of the
system’s life. Given these dramatic productivity savings, Oracle Database Appliance is a choice that
makes sense for any organization looking to increase the availability of its databases. Table 1 below
summarizes the report’s findings, and highlights how much more effective a DBA is when operating
Oracle Database Appliance rather than SQL Server. As the table below shows, the time savings
provided by Oracle Database Appliance were particularly striking in the critical area of ongoing system
maintenance.
Table 1 - Summary of Testing Results
Task
Description
A. Installation and Initial Deployment (Hours)
1 Sizing (Choose Hardware type, CPU, IO, Memory,
etc.)
2 Order (Hardware, Storage, Internal Networking,
etc.)
3 Research and Best Practices
4 Deployment
5 Test Unique Configuration
6 Resolve Interoperability Issues
B. Maintenance (Hours)
7 Verify Patch Compatibility
8 Run Database Appliance Manager Patch
9 Patch and Verify Firmware
10 Patch and Verify OS
11 Patch and Verify RDBMS
12 Test Unique Configuration
C. Support (Hours)
13 Run Database Appliance Manager Diagnostics and
Support
Time (Mins)
Oracle
SQL Server
4.0
194.5
60
2400
60
1200
0
120
0
0
19.8
0
100
0
0
0
0
12.0
2400
867
2400
2400
592.8
60
0
120
286
98
2400
76.0
60
0
3
14
15
16
Call Vendor Support to Verify Environment
Local Log Files for Each Component / Vendor
Send Log Files to Each Vendor
0
0
0
240
120
0
This study did not focus on day-to-day manageability of the two database platforms. A detailed
comparison of the database administration capabilities of Oracle Database and SQL Server can be found
in the companion ORC white paper, “Database Manageability and Productivity Cost Comparison Study:
Oracle Database 11g Release 2 vs. SQL Server 2008 Release 2”. That report found that Oracle
Database’s superior manageability offered customers increased productivity and cost savings over SQL
Server.
While the version of SQL Server evaluated in this study was SQL Server 2008 Enterprise Edition Release
2, ORC had an opportunity to evaluate SQL Server 2012 Community Technology Preview 3. It’s the
determination of ORC that while SQL Server 2012 does have significant improvements in the area of high
availability with the introduction of AlwaysOn Availability Groups, it still requires tremendous effort and
knowledge to configure, implement and support it. It is therefore ORC’s professional opinion that all the
conclusions presented in this document remain relevant when comparing Oracle Database Appliance and
SQL Server 2012.
4
Study Background
This in-depth study compared the ease of deployment and management capabilities of Oracle Database
Appliance and Microsoft SQL Server, employing pre-defined tasks and steps that DBAs perform daily or
regularly on a high availability platform. The study focused on three key areas:



Installation and Initial Deployment
Maintenance
Support
For this study, the tools provided out of the box with each platform were the only ones used to complete
assigned tasks, and no third-party tools were used. In the few instances where a tool was not available,
the task was completed using a SQL command rather than a GUI-based tool.
Oracle Database Appliance hardware consists of a single 4U rack mountable chassis containing two
Oracle Linux server nodes, each with two 6-core Intel Xeon processors and 96 GB of memory. The two
server nodes are connected via an internal redundant gigabit Ethernet (GbE) interconnect for cluster
communication, and each provides both 1 GbE and 10 GbE external networking connectivity. The
appliance contains 12 TB of raw storage that’s triple-mirrored, offering 4 TB of usable database storage.
The appliance is designed with mission-critical requirements in mind, with hot-swappable and redundant
components.
To help customers easily deploy and manage their databases, Oracle Database Appliance includes
Oracle Database Appliance Manager. This software provides one-button automation to provision, patch,
and diagnose Oracle Database Appliance.
5
Testing Methodology
Several DBAs in multiple disciplines and environments were interviewed to determine the weighting of the
testing subject areas. The tasks to be tested were chosen by a team of database experts and identified
as the most time-consuming areas for a DBA. The categories and tasks evaluated were:



Installation and Initial Deployment
o
Sizing
o
Ordering of Components
o
Research
o
Deployment
o
Testing the Unique Configuration
o
Resolving Interoperability Issues
Maintenance
o
Verifying Patch Compatibility
o
Running Oracle Database Appliance Manager to Apply Patches
o
Patching and Verifying Firmware
o
Patching and Verifying the OS
o
Patching and Verifying the RDBMS
o
Testing the Unique Configuration
Support
o
Running Oracle Database Appliance Manager for Diagnostics and Support
o
Communicating with Vendors to Verify the Environment
o
Locating Log Files for Each Component and Vendor
o
Sending Logs with Details to Each Vendor
Several of these tasks included assumptions based on feedback from interviewed DBAs. These
assumptions will be noted throughout the report.
Every task was broken into steps, each of which required a user action to continue the workflow. Timing
methods were applied to track the time and number of steps required to complete each task. Time was
measured from the start of a given task, when the mouse began to move to complete the task. Whenever
possible, defaults were selected in all test cases, and at the first sign of a “Finish”, “OK” or “Submit”
button, that button was selected. For tasks that could run in the background, time was measured when
the job or process was submitted to the DBMS. The job was tracked as a subtask and the run time was
recorded independent of the task that required the DBA’s attention. Tasks were tracked individually for
time and step count, so results were available at both the detail and aggregate level.
6
Comparison Tests
Test: Installation and Initial Deployment
The tasks covered in this test including everything from evaluating requirements to the actual installation
of the system. The amount of time it took to complete each of the tasks in this category is shown in Table
2 below:
Table 2 - Test: Installation and Initial Deployment Tasks
A. Installation and Initial Deployment
Time (Minutes)
Oracle
SQL Srv
60
2400
60
1200
0
2400
120
867
4
Sizing (Choosing hardware, CPU, I/O,
Memory)
Ordering (Hardware, Storage, Internal
Networking)
Research and Best Practices (OS, Storage,
Grid Infrastructure, Database)
Deployment
5
Testing the Unique Configuration
0
2400
6
Resolving Interoperability Issues
0
2400
4.0
194.5
1
2
3
Total (Hours)
We’ll now look at each of these tasks in detail.
Sizing
The sizing task took into account the selection of hardware, CPU, memory, I/O and other various
components to ensure that the system could perform as expected and handle the required load.
Oracle Database Appliance comes in standard configurations, with all required hardware already preconfigured for a fail-over solution. When defining the configuration of a comparable SQL Server-based
solution, it was determined that approximately a week was needed to define an appropriate configuration,
communicate the available options to decision-makers and make a selection.
Ordering
Since Oracle Database Appliance comes in pre-determined configurations, ordering is straightforward.
For this reason, it was determined that the decision for choosing a configuration and ordering it would
take approximately an hour.
In the case of assembling a SQL Server-based high-availability solution, ORC concluded that
coordinating the ordering of hardware, software, disk, and networking from different vendors would take
approximately twenty hours. In this scenario, time must be attributed to the coordination of various
vendor deliveries to ensure the effective use of an administrator’s time in the data center. Careful
attention must also be paid to the review of confirmation letters or emails to ensure the order was placed
correctly, since a single component being ordered or keyed in incorrectly could be costly to the overall
project.
7
Research and Best Practices
A major cost factor in planning for a high availability system is conducting research. It can take hours of
research to determine the best hardware available and identify known configuration issues. ORC
determined that an analyst or administrator assigned to design a required high-availability configuration
would spend forty hours performing this research. This figure assumes that a company has the required
expertise in-house, and does not require the support of outside consultants, which would significantly
increase costs and time.
Oracle Database Appliance removes the need for this research since Oracle has already assembled the
best possible components for a high availability solution. With Oracle Database Appliance, there’s no
need to conduct any research to determine what components work with various operating systems or
database platforms, since that’s already been designed into the system.
Deployment
Deployment was a task that ORC was able to capture timings against. During the deployment of SQL
Server, ORC made several assumptions. We assumed it would take two hours to unpack, rack, and hook
up all cables for two database servers. We also assumed it would take four hours to unpack, rack, hook
up all cables and configure storage for a clustered environment. We timed the installation of the
operating system (Windows Server Enterprise Edition) on both the primary and secondary nodes, the
configuration of the storage at the operating system level, installation of SQL Server, and bringing all
nodes up to the latest patch level. Overall, the SQL Server deployment took 14.5 hours to complete.
Conversely, we were able to fully deploy Oracle Database Appliance in just under two hours. The
required tasks for our administrators were to unpack, rack and plug cables into the appliance, and run
Oracle Database Appliance Manager.
Testing the Unique Configuration
Before a high availability solution can be deployed in production, many testing scenarios have to be
analyzed. Administrators and business stakeholders must have a high level of comfort prior to allowing a
new solution to be operationally deployed. Ensuring that there are no issues can take anywhere from 40
to 120 hours. ORC determined that for the purposes of this test, 40 hours would be used. At ORC we
realize that this is a conservative estimate, especially when one considers the personnel with various
roles and responsibilities (e.g. DBAs, storage experts, operations, management) that must be involved in
any testing scenarios.
The configuration of Oracle Database Appliance has already been tested and approved by Oracle, so
there’s no need for an organization to test anything in relation to a unique configuration. However, ORC
recognizes that best practices usually dictate fully testing a product after deployment to ensure its proper
operation.
Resolving Interoperability Issues
Failures between the various components of multi-vendor solutions will invariably occur. Discovery and
resolution of these, sometimes unique, issues can be quite costly. The employment of technology
specialists within an organization to discover and solve these issues is the primary cost in this area. We
again made the assumption that 40 hours are required to locate, diagnose, and resolve these issues. As
previously stated, it’s important to remember that this estimate is again conservative, especially when one
considers that many employees will be involved in this task.
Oracle Database Appliance removes this pitfall altogether. Because of the nature of having an appliance
solution, interoperability issues no longer exist, since all components have been tested prior to the
complete system being delivered to the customer.
8
Test: Maintenance
While installation is an important task, ongoing maintenance is where DBAs spend the majority of their
time. In our testing we assumed that there would be only a single issue of maintenance per month. We
realize this is probably too ideal a situation, considering that vendors do not coordinate the release of
their various patches with concern to the operating system, database, or firmware. The maintenance
tasks that we evaluated are listed in the table below:
Table 3 - Test: Maintenance Tasks
B. Maintenance Tasks
7
Time (Minutes)
Oracle
SQL Srv
9
Verifying Patch Compatibility
Running Oracle Database Appliance
Manager to Apply Patches
Patching and Verifying Firmware
10
Patching and Verifying OS
0
286
11
Patching and Verifying RDBMS
0
98
12
Testing the Unique Configuration
0
2400
19.8
592.8
8
Total (Hours)
0
60
100
0
0
120
Note: Maintenance multiplier of 12 applied for single maintenance instance per
month
Verifying Patch Compatibility
Before installing any patch within an existing high availability infrastructure, it’s imperative that
administrators evaluate the patch’s compatibility with the current configuration. Upgrading an operating
system or piece of firmware could have a disastrous impact on another component, leading to node, or
even entire system failure. Performing the proper research on each patch is therefore very important.
For example, the Windows Automatic Updates patching system is not cluster-aware, and cannot
recognize that a SQL Server database is part of a clustered environment. While this will change in the
upcoming Windows Server 8 operating system, for most production environments today a cluster-aware
Windows Automatic Update is not available; therefore Windows Automatic Update has to be disabled on
any production servers.
Validation of patches is not only complex for the operating system or attached storage, but also for the
firmware and SQL Server. Each patch needs to be analyzed and confirmed. We assumed that during
each patch cycle an analyst or administrator would need to spend one hour validating the patch’s
compatibility with the unique configuration. Assuming only twelve of these per year is conservative,
considering the rate at which patches are produced for Microsoft products.
Since all patches are analyzed and tested prior to shipping based on pre-defined Oracle Database
Appliance configurations, patches never need to be verified or validated for a unique configuration.
However, as in the case of deployments, ORC recognizes that best practices often dictate fully testing a
product after patching. Tools such as Oracle Real Application Testing can be used to perform this type of
testing.
Running Oracle Database Appliance Manager to Apply Patches
One software program that we found very useful for patching was Oracle Database Appliance Manager.
Oracle Database Appliance Manager consolidates operating system, firmware, and RDBMS patching into
a single command. This utility looks at the current configuration and automatically brings it up to the
9
proper patch level. Rather than babysitting the application of all patches, executing this command allows
the DBA to walk away and leave this process unattended. By using Oracle Database Appliance
Manager, an exhaustive patch that updated disks, controller, operating system and Oracle Database took
only 1 hour and 40 minutes to complete.
Microsoft SQL Server does not have a similar patching manager.
problematic, resource intensive and time consuming.
Patching is therefore potentially
Patching and Verifying Firmware
The process of validating firmware patches is just as time intensive as verifying the compatibility of a
given patch. Since firmware patches are not cluster-aware, special consideration must be taken into
account before performing any firmware patch upgrade. We therefore assumed two hours for this
process in a Windows Server environment. This includes making any configuration changes that are
needed in the cluster, and actually applying and validating the patch.
With Oracle Database Appliance, firmware patching is easily accomplished through the Oracle Database
Appliance Manager.
Patch and Verifying the OS
In order to patch the Windows Server environment, we ran the Automatic Update process in order to
bring it to the latest patch level, selecting all “Important” options available, and selecting all “Optional”
updates that were related to the operating system. The process, including configuration, download time,
installation, reboot, and validation (on a 100MB dedicated internet trunk at a Tier 1 facility) took a total of
286 minutes to complete. This time accounted for the work required for both nodes.
With Oracle Database Appliance, operating system patching is taken care of by the Oracle Database
Appliance Manager.
Patching and Verifying the RDBMS
For this test, we applied Service Pack 1 to an existing SQL Server 2008 EE installation. This patch took
49 minutes to apply to each node. Unlike the operating system, SQL Server service packs are cluster
aware, therefore making it easier to apply a patch to SQL Server than Windows Server.
Still, it’s worth noting that it took longer to apply the SQL Server patch than it did to patch the entire
Oracle Database Appliance.
Testing the Unique Configuration
After applying a set of patches, it’s best practice to test the configuration that’s been deployed. For most
organizations this includes enlisting technologists from the Quality Assurance, Operations and DBA
departments. Additional management personnel or business stakeholders that might need to sign off on
the patch for release into production may also need to be included. We again chose a conservative figure
of forty hours for performing this task in a Windows Server environment running SQL Server.
With Oracle Database Appliance, all configurations have already been tested prior to being delivered to
the customer. Oracle Database Appliance has already gone through rigorous testing by Oracle to ensure
there are no compatibility issues between its various components.
10
Test: Support
Backup and recovery are vitally important to a DBMS platform and not surprisingly, both solutions were
able to perform these tasks. However, Oracle Database Appliance showed significant productivity gains
over Microsoft SQL Server. The table below outlines the tasks performed:
Table 4 - Test: Support Tasks
C. Support Tasks
13
14
15
16
Running Oracle Database Appliance
Manager for Diagnostics and Support
Communicating with Vendors to Verify
the Environment
Locating Log Files for Each Component
and Vendor
Sending Logs with Details to Each
Vendor
Total (Hours)
Time (Minutes)
Oracle
SQL Srv
60
0
0
240
0
120
0
20
12.0
76.0
Note: Support multiplier of 12 applied for single support instance per month
Running Oracle Database Appliance Manager for Diagnostics and Support
Oracle Database Appliance Manager continued to prove its value in this phase of testing. In under an
hour, it was able to validate the environment and gather all necessary log files for Oracle Support. Its
ability to analyze the entire environment in regards to firmware, hardware, operating system and RDBMS
showed the long-term value of this tool in resolving support issues.
In contrast to Oracle Database Appliance, Microsoft does not have a similar support tool available for
Windows Server and SQL Server.
Communication with Vendors to Verify the Environment
A multi-vendor high availability solution like one based on SQL Server makes resolution of support issues
more difficult. Coordinating communication between all of these vendors can be an arduous task,
especially when one considers that various department heads may also have to be included for these
conversations. Each vendor may also require a customer to supply different documentation regarding
their environment. Given the time necessary for communication and assembling documentation, we
made an assumption that four hours were needed per every support occurrence.
Since Oracle Database Appliance is a pre-configured solution, Oracle Support already knows every detail
of its configuration.
Locating Log Files for Each Component and Vendor
While the log files for Windows Server and SQL Server are in commonly known locations, log files for
firmware and other components may not be so easy to locate. Assembling these and deciding what
exactly needs to be sent to the vendor can be time-consuming and difficult. If a customer sends too
much information it could slow the issue resolution process, and if they send too little, a customer may
have to start the problem resolution process over again.
11
With Oracle Database Appliance, all log file locations are known to Oracle Database Appliance Manager
and log files can therefore be easily located and assembled for shipping to Oracle Support.
Sending Logs with Details to Each Vendor
ORC assumed in this study that for each support issue, emails containing the various log files necessary
for problem resolution would need to be sent to each vendor. We assumed 20 minutes would be needed
for drafting, attaching, reviewing and sending an email to a vendor. This length of time also includes
gathering input from various technologists or department heads to ensure that the email correctly explains
the issue.
In contrast to SQL Server, Oracle Database Appliance Manager automatically takes care of sending log
files to Oracle Support, after it has run diagnostics and verified the environment.
12
Conclusion
Based on its research, ORC has concluded that Oracle Database Appliance provides significant time
savings over Microsoft SQL Server when performing DBMS deployment, maintenance and support tasks.
ORC found that Oracle Database Appliance is able to achieve these time savings because its preconfigured nature and unique manageability tools greatly simplify many time-consuming DBA tasks.
Below is a summary of ORC’s findings in three important areas DBAs have responsibility for:

Installation and Deployment – In performing installation tasks, ORC found that since Oracle Database
Appliance comes with all required hardware already pre-configured for a fail-over solution,
deployment took just under 2 hours. In contrast, the deployment of a comparable SQL Server-based
solution took 14.5 hours.

Maintenance – From a maintenance perspective, ORC found that Oracle Database Appliance
Manager’s ability to determine the current configuration and automatically bring it up to the proper
patch level gave it a very significant time savings advantage over SQL Server.

Support – Oracle Database Appliance Manager was able to validate an environment and gather all
necessary log files for Oracle Support in under an hour, while assembling and communicating this
information for a SQL Server environment was a time-consuming process.
Oracle Database Appliance therefore saves customers significant time compared to SQL Server by
greatly simplifying previously time-consuming DBA tasks. It offers high availability and reliability to
customers for their mission-critical business applications, while at the same time providing exceptional
ease of use.
13
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement