Optimizing an Oracle® Database with Dell

Optimizing an Oracle® Database with
Dell™ Compellent™ Automated Tiered
Storage
Implementation Best Practices of Dell™ Compellent™ Data Progression
with Oracle® OLTP workloads
Wendy Chen and Tom Dau
Database Solutions Engineering
Dell Product Group
October 2012
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
This document is for informational purposes only and may contain typographical errors and
technical inaccuracies. The content is provided as is, without express or implied warranties of any
kind.
© 2012 Dell Inc. All rights reserved. Dell and its affiliates cannot be responsible for errors or omissions
in typography or photography. Dell, the Dell logo, Compellent, PowerConnect, and PowerEdge are
trademarks of Dell Inc. Intel and Xeon are registered trademarks of Intel Corporation in the U.S. and
other countries. Oracle is a registered trademark of the Oracle Corporation. Brocade is a registered
trademark of Brocade Communications Systems, Inc. Microsoft, Windows, and Windows Server are
either trademarks or registered trademarks of Microsoft Corporation in the United States and/or other
countries. Other trademarks and trade names may be used in this document to refer to either the
entities claiming the marks and names or their products. Dell disclaims proprietary interest in the
marks and names of others.
Page ii
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
October 2012| Rev 2.0
Acknowledgements
The authors would like to thank the following people for their contribution during the development and
review of this whitepaper:
Connie Sheldon
Page iii
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Contents
Executive summary ....................................................................................................... 2
Introduction ................................................................................................................ 3
Dell solutions for Oracle database ..................................................................................... 4
Technology overview ..................................................................................................... 5
Dell Compellent Automated Tiered Storage – Data Progression ................................................ 5
Dell 12th Generation PowerEdge Servers ........................................................................... 7
Oracle Databases with Dell Compellent Data Progression ......................................................... 8
Test tools and configurations ........................................................................................... 8
Baseline configuration without Data Progression ............................................................... 10
Enabling Data Progression and simulating read-only workload ............................................... 12
Tiered storage configuration with Data Progression ........................................................... 12
Test Results............................................................................................................... 14
Summary .................................................................................................................. 18
References ................................................................................................................ 18
Figures
Figure 1.
Architecture of the tiered storage infrastructure .................................................... 3
Figure 2.
Dell Compellent Automated Tiered Storage - Data Progression ................................... 6
Figure 3.
Configuration architecture of Oracle RAC database and Dell Compellent storage system ... 9
Figure 4.
Volume distribution of the baseline configuration without Data Progression ................. 11
Figure 5.
Volume distribution of the tiered storage configuration created by Data Progression ...... 13
Figure 6.
Oracle TPC-C Transaction per second (TPS) vs. user load ....................................... 15
Figure 7.
Oracle TPC-C average response time (second) vs. user load ..................................... 16
Figure 8.
Dell Compellent Data Progression reduces transaction response time of Oracle OLTP
workload ............................................................................................................ 17
Figure 9.
Dell Compellent Data Progression improves application throughput in TPS of Oracle OLTP
workload ............................................................................................................ 18
Tables
Table 1.
Hardware and software configurations for Benchmark factory test ............................ 10
Table 2.
Oracle volume placement with Data Progression .................................................. 12
Page 1
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Executive summary
Enterprise data can be categorized based on criteria such as performance needs, access frequencies,
I/O patterns, and business values. With the exponential data growth, the traditional “one size fits all”
data storage strategy of storing all data into a uniform storage media has become costly and less
efficient. To solve this challenge, many IT organizations have adopted the tiered storage solution.
Tiered storage is a data storage environment consisting of multiple storage media types forming
storage tiers. Storage tiers can differ in cost, performance and capacity. Tiered storage solution
enables the alignment of different categories of data to different storage tiers, with the goal of
reducing total storage cost and preserving performance. In this paper, we will examine the automated
tiered storage solution, Data Progression, offered by Dell Compellent storage systems. Data
Progression automatically moves data to the optimum storage tier and/or RAID level based on actual
use and performance needs. This paper describes how users can leverage Data Progression to optimize
Oracle database performance. It provides recommended best practices to implement Oracle databases
with Dell Compellent Data Progression.
Page 2
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Introduction
Today’s storage devices come in many choices ranging from high performance and high cost storage
devices such as the flash memory-based Solid State Drives (SSD), to slower performance and less
expensive storage devices like the mechanical hard disk drives (HDD). While it would be ideal to store
all data in the high speed devices, it would be unnecessary and impractical given the different
categories of data, the high cost of the fast disk drives, and the exploded data growth.



Majority of enterprise data is less transactional and not frequently accessed. Transactional or
frequently changed data represents only a small percentage of the entire data storage. It is
cost effective to store only the frequently accessed data in the high performance storage
devices.
Pricing for the high performance storage devices such as SSDs is still higher than HDDs for
comparable storage capacity. It may be impractical to store all data on SSDs for many
customers.
The exponential data growth rate leads to the huge storage capacity requirement. As a result,
companies need to seek new ways to store data more efficiently and more cost effectively to
control the total storage cost. Storing all data in a single type of storage media cannot meet
the goal of reducing storage cost while preserving application performance at the same time.
The above factors lead to the increasing adoption of the tiered storage architecture (Figure 1).
Figure 1.
Architecture of the tiered storage infrastructure
A tiered storage environment consists of two or more types of storage media with the fast and more
expensive disk devices forming the top tiers, and slower performance and lower cost storage devices
forming the lower tiers. Tiered storage aligns different categories of data to different storage tiers.
For example, mission-critical or frequently accessed data might be stored on the high performing SSD
drives in tier 1 or the SAS drives in tier 2; infrequently used data might be stored to the slower and less
Page 3
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
expensive SATA drives in tier 3. However, assigning data to a particular storage tier may be an ongoing
and complex activity which may take a lot of time and effort from the administrators if it is done
manually. Automation is a must to support the tiered storage infrastructure. An automated storage
tiering solution can meet the right balance between performance and storage cost by dynamically
identifying and moving hot data to higher-performance storage tiers, while moving cold data to slower,
lower-cost storage tiers.
Many types of applications can benefit from the automated storage tiering solution. Among them, an
Oracle® database is a prime candidate.




Database storage space represents a large percentage of the overall data storage. Industry
research indicates that the average databases are growing at a double digit annual growth
rate. The driving forces of the rapid database growth rate can be contributed by the factors
such as the increased business demands, the regulatory and legal requirements resulted longer
data retention period, and the high availability requirements with duplicated copies of data,
etc. Managing and retaining the overall database storage usage imposes a big challenge on IT
organizations. The automated storage tiering solution can help address this challenge.
Storage needs for Oracle databases differ from one database to another. Oracle database I/O
workloads can be classified as Online Transaction Processing (OLTP), or Decision Support
System (DSS), or a mix of the two. OLTP workload is typically used by the transactional
processing applications represented with a small set of active data, which can benefit from
placing on the high-performance storage media. DSS applications are typically used by the
reporting and data analysis applications represented with a large set of older data that can be
stored in the high capacity and low cost storage media to reduce the storage cost.
Oracle databases also differ in terms of Service Level Agreement (SLA). Some databases are
mission critical while others are less important. IT organizations can take advantage of the
automated storage tiering solution to align databases to different storage tiers, in order to
minimize the storage cost while satisfying SLAs and meeting the expected performance level.
Furthermore, within an Oracle database, components have different I/O characteristics. For
example, online redo log files have high I/O demands. Indexes in OLTP type of databases are
typically I/O intensive comparing to other database objects. For DSS type of databases,
partitions of older data are usually less accessed. An automated storage tiering solution can
help place different components of an Oracle database to the most optimal storage tier.
Dell solutions for Oracle database
Dell solutions for Oracle products are designed to simplify operations, improve usability, and provide
cost-effective scalability as your needs grow over time. In addition to providing server and storage
hardware, Dell solutions for Oracle include:




Dell Configurations for Oracle―in-depth testing of Oracle configurations for high-demand
solutions; documentation and tools that help simplify deployment
Integrated Solution Management―standards-based management of Dell solutions for Oracle
that can lower operational costs through integrated hardware and software deployment,
monitoring, and updating
Oracle Licensing—licensing options that can simplify customer purchase
Dell Enterprise Support and Infrastructure Services for Oracle―planning, deployment, and
maintenance of Dell solutions for Oracle database tools
Page 4
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
For more information concerning Dell Solutions for Oracle Database, visit www.dell.com/oracle.
Technology overview
Dell Compellent Automated Tiered Storage – Data Progression
The Dell Compellent Storage Center Storage Area Network (SAN) provides a highly efficient and flexible
foundation for enterprise and the cloud. Dell Compellent storage features an innovative Dell Fluid
Data architecture to put the right data in the right place at the right time. It enables the storage
system to dynamically adapt the changing business environment.
Dell Compellent Storage Center provides a fully virtualized storage platform that includes:




Storage virtualization that abstracts and aggregates all resources across the entire array,
providing a high-performance pool of shared storage.
Thin provisioning and automated tiered storage to deliver optimum disk utilization and
intelligent data movement.
Space-efficient snapshots and thin replication for continuous data protection without wasted
capacity.
Built-in automation and unified storage management to streamline storage provisioning,
management, migration, monitoring and reporting.
Dell Compellent pioneered the automated storage tiering solution in its Storage Center products in
2004. The Data Progression feature (Figure 2) delivers sub-LUN automated storage tiering to move
data dynamically, intelligently, and efficiently among multiple storage tiers and RAID levels. Data
Progression is a licensed feature that leverages cost and performance differences between storage
tiers, allowing the maximum use of lower cost and higher capacity SATA or SAS (7.2K RPM) drives for
stored data, while maintaining performance oriented SSD drives, Fibre Channel or SAS (15K RPM) drives
for frequently-accessed data.
Page 5
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Figure 2.
Dell Compellent Automated Tiered Storage - Data Progression
Unlike many newcomers to the storage tiering market, Dell Compellent Data Progression is built right
into the virtualized storage platform. It realizes the full potential of automated tiered storage.






True Virtualization. Dell Compellent engineered its Fluid Data architecture from the ground
up to include Data Progression. Fluid Data was architected to provide truly virtualized
storage that spans all disks in the storage environment. Fluid Data provides an ideal
foundation for tiered storage. With Fluid Data, there are no constraints on where data can
reside, as data is not confined to conventional disk group. Data can be moved without
limitation to a tier with a particular drive type or performance level, or to a particular RAID
level within a storage tier.
Build-in Feature. Dell Compellent Data Progression is built right into the storage platform;
it does not require additional hardware or server-side agents to operate.
Fine Granularity. Dell Compellent Fluid Data facilitates a highly granular approach for Data
Progression to migrate data in small sizes of 512 KB, 2 MB or 4MB. The granular approach
optimizes tiering by moving less data in the backend and placing data with a greater level of
precision.
Real Time Intelligence. Dell Compellent captures real-time use characteristics of each data
block at 512 byte level. This information provides the intelligence to determine whether
and when blocks of data should be moved from one storage tier, or one RAID level, to
another.
Snapshot Integration. Dell Compellent Data Progression is tightly integrated with the Dell
Compellent snapshot feature – Data Instant Replay. In the Dell Compellent architecture,
new data is written by default to tier 1, RAID 10 storage to provide the best write
performance. Replays move to a lower storage tier with RAID 5 or 6 protections. When new
data needs to be written to an existing block that has been migrated to a lower tier, writes
are redirected to the tier 1, RAID 10 storage to guarantee superior write performance.
True Automation. Dell Compellent Data Progression provides fully automated sub-LUN
tiering. Administrators use policy-based profiles to drive placement and movement of data.
Page 6
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Data migration occurs automatically at a set time or on demand. The migration process does
not affect data availability or application performance.
Dell 12th Generation PowerEdge Servers
Dell 12th Generation PowerEdge servers are the newest addition to the PowerEdge server family. Dell
12th Generation PowerEdge servers can help improve IT experience with these notable features:












Manage anywhere anytime with agent-free server management
Reduce maintenance time with auto-update for replacement parts
Control cooling costs with better power monitoring and control
Tailor network to applications with fabric flexibility
Access information quicker via SATA, SAS, SSD and PCI Express Flash drives
Data protection using best-in-class RAID
Protect data at rest with malware resistance and faster encryption
Keep data safer with firmware signing and encrypted credentials
Accelerate high-performance computing (HPC) and virtual desktop infrastructure (VDI)
through integrated graphics processing unit (GPU) technology
Deliver more throughput with major I/O performance enhancements
Better application performance with dual internal RAID controller options for PowerEdgeT620
and R720
No compromise on performance with greater memory density and capacity along with Intel
Xeon E5 processors
Dell PowerEdge 12th Generation servers feature the Intel® Xeon® E5-2600 series processors based on
the Sandy Bridge-EP architecture which deliver more computations per second. Intel Xeon E5-2600
series processors provide up to 8 physical cores or 16 logical cores through hyper-threading, and up to
20 MB cache. Intel Xeon E5-2600 series processors also include features such as the new Intel advanced
vector extensions, and the optimized turbo boost technology.
Dell PowerEdge 12th Generation servers include the express flash PCIe solid state drives to deliver
better internal storage performance by connecting directly to the processor via PCIe bus. These PCIe
solid state drives have up to 3x performance of standard SAS SSDs and 1000x performance of 15K SAS
hard drives.
The PowerEdge R720 (12th generation) servers were used in the test configuration for this paper. The
R720 server is a two-socket, 2U rack server emphasizing performance and scalability. R720 servers are
designed for mid-to-large-size data centers, and are ideal for use as a virtualization or database server.
Some highlighted features of the R720 include:







Large memory footprint 24 DIMMs (768GB)
Dual SD cards for redundant hypervisor
CacheCade RAID enhancement to boost I/O
Internal storage capacity up to 16 x 2.5 HDD or 8 x 3.5 HDD
Maximum of four optional PCIe flash SSD drives
Redundant power supply units (PSU)
Hot plug and swappable PSU, HDDs, and fans
Page 7
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Oracle Databases with Dell Compellent Data Progression
Dell Compellent Data Progression is configured by assigning Storage Profiles to volumes. Storage
Profiles define the RAID level and storage tiers on which data blocks of a volume can be stored. If Data
Progression is licensed, data can be migrated between RAID levels within a tier and between tiers. If
Data Progression is not licensed and a system uses RAID 10 and RAID 5, data can be migrated up or
down within a tier but cannot be migrated between tiers.
Dell Compellent Data Progression is tightly integrated with the Dell Compellent Data Instant Replay.
Data Instant Replay creates space-efficient snapshots of data volumes. To utilize Data Progression
effectively, it is recommended to take Data Instant Replay of the volumes on a regular basis. New data
is written by default to tier 1, RAID 10 storage to provide the best write performance. Replays move to
a lower tier with RAID 5 or 6 protection levels during the next migration cycle. Over time, infrequently
accessed data blocks move to a lower storage tier and RAID level. Moving read-only data from RAID 10
to RAID 5 or RAID 6 maintains the same read performance while freeing up storage space on the higher
tier or RAID level.
Elements of an Oracle database have unique I/O characteristics. The following general guidelines can
be followed in configuring OLTP type of Oracle databases with Dell Compellent Data Progression with
SSD drives.




Applications with the most random data requirements can gain greatest benefit from SSD drives
over hard disk drives. Typical Oracle OLTP workloads are dominated with small and random
I/O’s. Therefore Oracle OLTP type of database is a suitable candidate to deploy with SSD
drives as the top tier in a multi-tier storage infrastructure.
Within Oracle datafiles, database objects have different IO access patterns. For example,
indexes of OLTP databases typically are more IO intensive than other database objects.
Compellent Data Progression is able to identify the frequently accessed data blocks and
automatically move these hot data blocks to the top SSD tier, while identifying and moving less
accessed data blocks to lower storage tiers for the best performance / cost optimization.
Oracle online redo log is an IO intensive component of an Oracle database. However, the
online redo log IO operations are primarily consisted of sequential writes, which benefit little
from SSD drives. The 15K RPM SAS or Fibre Channel drives can well serve the IO requirements
of online redo logs. Therefore, it is recommended to place online redo log files on the 15K RPM
SAS or Fibre Channel storage tier only.
Oracle archived redo log file, which is usually stored in the Flash Recovery Area, can be stored
on SATA or lower cost Fibre Channel drives with slower rotational speed. By using Compellent
Data Progression, one can control the writes to Flash Recovery Area directed to the 15K RPM
SAS or Fibre Channel storage tier, and the replays of Flash Recovery Area stored on the SATA or
lower cost Fibre Channel storage tier.
To measure the performance benefit of Dell Compellent Data Progression for Oracle databases, Dell
engineers conducted a series of benchmark stress tests with OLTP database in various tiering
configurations. The test tools, configurations and results are detailed below.
Test tools and configurations
Quest Software’s Benchmark Factory TPC-C was used in the tests. Benchmark Factory TPC-C is a loadgenerating utility that simulates OLTP users and transactions on a database for a given number of
Page 8
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
users. The database configuration used in the tests was a 2-node Oracle 11g R2 (11.2.0.3) Real
Application Cluster (RAC). The database schema was populated by Benchmark Factory. The TPC-C test
runs conducted simulated loads from 100 to 3,000 concurrent users in an increment of 100. Test
outputs include metrics such as the average transaction response time and transaction per second
(TPS).
The test configuration consists of the following components:







PowerEdge R720 servers running Oracle Enterprise Linux 5 Update 7 x86_64, and Oracle 11g R2
Real Application Cluster (RAC) database Enterprise Edition (EE) version 11.2.0.3.0
Redundant Dell PowerConnect™ Gigabit Ethernet switches for Oracle cluster interconnect
private network
Server-storage interconnect using redundant Brocade® 5300 Fibre Channel switches
Redundant Dell Compellent Series 40 controllers
Dell Compellent SBOD Fibre Channel enclosure populated with 8 SSD drives for tier 1 storage
Dell Compellent 2.5’’ SAS enclosure populated with 24 15K RPM SAS drives for tier 2 storage
Dell Compellent 3.5’’ SAS enclosure populated with 12 7K RPM SAS drives for tier 3 storage
An architecture overview of the test configuration is shown in Figure 3.
Figure 3.
Configuration architecture of Oracle RAC database and Dell Compellent
storage system
The hardware and software details of the test configuration are summarized in Table 1.
Page 9
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Table 1.
Hardware and software configurations for Benchmark factory test
HARDWARE AND SOFTWARE CONFIGURATIONS FOR BENCHMARK FACTORY TEST
Two PowerEdge R720 servers, each with:
• Two Intel Xeon 8-core 2.20 GHz CPU’s
• 64 GB of RAM
• 4 Gb QLogic dual-port QLE2462 HBA for SAN traffic
External Storage
Two Dell Compellent Series 40 controllers connected with
• 7 plus 1 spare 136GB SSD disks in a Dell Compellent SBOD Fibre
Channel enclosure
• 23 plus 1 spare 136GB 15K RPM SAS disks in a Dell Compellent 2.5’’
SAS enclosure
• 11 plus 1 spare 932GB 7K RPM SAS disks in a Dell Compellent 3.5’’
SAS enclosure
• Firmware: Storage Center 6.1.2
Volume Configuration
Four volumes for database files; Four volumes for online redo log files; Four
volume for Flash Recovery area
OS
Oracle Enterprise Linux 5 Update 7
Storage Network
Two Brocade 5300 Fibre Channel switches
Test Software
Quest Benchmark Factory 6.1.1 with Oracle 64 bit 11.2.0.3 EE RAC
Server
The Oracle database in the test configuration was implemented using Oracle Automatic Storage
Management (ASM). Three ASM disk groups were created to host the Oracle data with a total of 12
storage volumes:



DATA disk group consists of four 100GB volumes to host datafiles.
FRA disk group consists of four 100GB volumes to host the Flash Recovery Area which stores the
archived redo log files.
REDO disk group consists of four 25GB volumes to host the online redo log files.
Three storage tiers were configured in the Dell Compellent backend:



Tier 1 – 7 plus 1 spare 136GB SSD drives.
Tier 2 – 23 plus 1 spare 136GB 15K RPM SAS drives.
Tier 3 – 11 plus 1 spare 932GB 7K RPM SAS drives.
To study the performance advantage of Data Progression, we compared between a baseline
configuration without storage tiering and the configuration after Data Progression rearranged data
placement to the three storage tiers. The test details are outlined in the sections below.
Baseline configuration without Data Progression
In the baseline configuration, all 12 Oracle storage volumes reside in the tier 2 15K RPM SAS drives with
RAID 10. Data Instant Replay is not configured on the volumes. The baseline configuration represents
the configuration where Data Progression is not implemented. All read and write IOs to the Oracle
volumes are directed to the 15 K RPM SAS drives configured with RAID 10.
Figure 4 shows the volume distribution of one of the DATA volumes and one of the FRA volumes in the
baseline configuration. All 12 Oracle volumes in the baseline configuration have similar type of volume
Page 10
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
distribution. As you can see, all data pages reside on RAID 10 of tier 2 storage, and there’re no replay
pages.
Figure 4.
Volume distribution of the baseline configuration without Data Progression
Benchmark Factory TPC-C test, which consists of approximately 70% of read IOs and 30% of write IOs,
was conducted on the baseline configuration. The TPC-C test runs simulated workloads from 100 to
Page 11
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
2,200 concurrent users in an increment of 100. Test metrics include transaction per second (TPS) and
average transaction response time in seconds. Test results are discussed in the Test Results section
below.
Enabling Data Progression and simulating read-only workload
Following the TPC-C test of the baseline configuration, Data Progression and Data Instant Replay are
enabled on the Oracle storage volumes except the REDO volumes. As discussed in the previous section
Oracle Database with Dell Compellent Data Progression, Oracle online redo log files benefit most from
placing on the 15K RPM SAS or Fibre Channel storage tier. Therefore, for our test, the 4 REDO volumes
remain on the 15K RPM SAS drives with RAID 10 configuration. Data Instant Replay is not enabled on
the 4 REDO volumes. The 4 DATA volumes and the 4 FRA volumes are grouped together in a
consistency group with a scheduled daily Data Instance Replay.
The Data Progression storage profile assigned to the DATA volumes and the FRA volumes are shown in
Table 2. In this configuration, the DATA volumes span on all 3 tiers of the storage. The FRA volumes
span on tier 2 and tier 3 storages.
Table 2.
Oracle volume placement with Data Progression
Tier 1 SSD
Tier 2 15K RPM SAS
Tier 3 7K RPM SAS
RAID 10
RAID 5-5
RAID 10
Writeable Data
DATA disk group volumes
Replay Data
DATA disk group volumes
DATA disk group volumes,
FRA disk group volumes
RAID 5-9
RAID 10-DM
RAID 6-10
DATA disk group volumes
DATA disk group volumes,
FRA disk group volumes
Dell Compellent storage continuously captures user characteristics of each data block including
creation time, location, access frequency, and data type. This data enables Data Progression to
determine what data pages should be moved to a different storage tier or a different RAID level within
an existing tier. In order to generate sufficient IO activities to enable Data Progression data
movement, a read-only Oracle TPC-C workload was conducted on the test environment during a five
day period. The purpose of the read-only TPC-C workload includes:



Read-only stress load doesn’t make any data updates. The original data set is preserved for the
post Data Progression TPC-C performance test, which can enable an apple-to-apple comparison
with the baseline configuration.
Read-only stress load can generate sufficient IO activities on the storage array and in turns creates
access frequency of data blocks.
The mostly read and write database objects represented by the read-only TPC-C workload and the
typical TPCC workload (30% writes and 70% reads) are almost identical. This enables the read-only
TPCC workload to generate the same type of data access frequency as the typical TPC-C workload,
so Data Progression can move the correct data set for the post Data Progression TPC-C test.
Tiered storage configuration with Data Progression
At the end of the read-only stress load, the Oracle database volumes which were assigned with Data
Progression storage profiles have shown significant data redistributions among storage tiers and RAID
Page 12
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
levels. Figure 5 shows the new volume distribution of the same database volumes illustrated in Figure
4.
Figure 5.
Progression
Volume distribution of the tiered storage configuration created by Data
Page 13
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
The volume distribution of the DATA volume shows that Data Progression effectively redistributes its
data pages to span on all three tiers. Data pages that are moved to tier 1 represent the most active
data pages that can benefit from the high performing SSD drives in the tier 1 storage. Data pages that
are moved to tier 3 represent less active data pages. Moving less frequently accessed data pages to
lower cost disk drives in tier 3 helps reduce the storage space in higher tiers and lower the total cost of
storage. Similarly, the volume distribution of the FRA volume shows that Data Progression redistributes
its data pages to tier 2 and tier 3 as defined in the storage profile.
Benchmark Factory TPC-C test, which consists of approximately 70% of read IOs and 30% of write IOs,
was conducted again on the tiered storage configuration after the redistributions of data pages by Data
Progression. The test simulated workloads from 100 to 3,000 concurrent users in an increment of 100.
The test result is compared with the baseline test result in the following section.
Test Results
Figure 6 shows the comparison of TPS between the baseline configuration and the tiered storage
configuration as user load increases. Figure 7 shows the comparison of average response time (second)
of each configuration as user load increases.
Page 14
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Figure 6.
Oracle TPC-C Transaction per second (TPS) vs. user load
Oracle TPC-C Transaction per Second (TPS) Comparison
160
140
120
TPS
100
Baseline
configuration
80
60
Tiered storage
configuration
40
20
0
100 300 500 700 900 1100130015001700190021002300250027002900
User Load
Page 15
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Figure 7.
4
Oracle TPC-C average response time (second) vs. user load
Oracle TPC-C Response Time Comparison
Avg. Response Time (s)
3
Baseline
configuration
2
Tiered storage
configuration
1
0
100 300 500 700 900 1100 1300 1500 1700 1900 2100 2300 2500 2700 2900
User Load
The blue lines in Figures 6 and 7 represent the results from the baseline configuration, in which no
storage tiering is implemented and all Oracle volumes reside on the 15K RPM SAS disk drives. The red
lines in these Figures represent the results from the tiered storage configuration, in which Data
Progression automatically moves data pages among the three storage tiers according to the access
frequency.
In a typical Oracle OLTP environment, 1 second application response time is generally acceptable.
Therefore, when analyzing the test results, an average transaction response time of 1 second was
chosen as the Service Level Agreement (SLA). All data beyond 1 second average transaction response
time was discarded from the analysis.
As illustrated in Figure 6, application throughput in TPS improves by implementing Dell Compellent
Data Progression as compared to a traditional configuration of placing all Oracle volumes on one type
of storage media. Figure 6 shows that for OLTP workload, moving active data set to the tier 1 SSD
drives can achieve higher throughput.
Figure 6 shows that Data Progression improves the Oracle database throughput by supporting higher
concurrent user loads while meeting the SLA at the same time. This is in line with the results shown in
Figure 5.
The baseline configuration without Data Progression supports 1600 concurrent users when it meets the
SLA of around 1 second response time. At the 1600 concurrent user load, the response time of the
baseline configuration is 0.819 second; the response time of the tiered configuration is 0.469 second.
Page 16
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
As illustrated in Figure 8, using Data Progression to move active data sets in OLTP applications from SAS
disks to SSD disks can reduce transaction response time by 43 percent.
Figure
Dell Compellent
Data Progression
transaction
response
Figure 1.
8.
Dell Compellent
Data Progression
reducesreduces
transaction
response
time oftime
of
Oracle
OLTP
workload
Oracle OLTP workload
Response Time Reduction
0.9
0.819
0.8
43%
Response Time (s)
0.7
0.6
0.469
0.5
0.4
0.3
0.2
0.1
0
Baseline configuration
Tiered storage configuration
While meeting the SLA of around 1 second response time, the baseline configuration is able to deliver
TPS of 80.96; the tiered configuration delivers TPS of 115.96. Based on these data, we can conclude
that Dell Compellent Data Progression improves throughput in TPS of OLTP applications by 43 percent
as compared to the baseline configuration, by moving active data set to tier 1 SSD drives. This is
illustrated in the Figure 9 below.
Page 17
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
Figure 9.
Dell Compellent Data Progression improves application throughput in TPS of
Oracle OLTP workload
TPS Improvement
140
43%
115.96
120
100
80.96
TPS
80
60
40
20
0
Baseline configuration
Tiered storage configuration
From this study of configuring Oracle databases with Dell Compellent Data Progression, we may
conclude the following:


Dell Compellent automated storage tiering Data Progression reduces transaction response time
by 43 percent as compared to a non-tiered configuration in OLTP workload by moving active
data sets to the top tier SSD disk drives.
Dell Compellent automated storage tiering Data Progression improves TPS by 43 percent as
compared to a non-tiered configuration in OLTP workload by moving active data sets to the top
tier SSD disk drives.
Summary
The Dell Compellent sub-LUN tiering solution Data Progression can cost-effectively address the
explosive data growth of Oracle databases, while delivering more superior performance for OLTP
applications. This paper demonstrates these benefits through lab testing. It also discusses the best
practices for deploying automated storage tiering in an OLTP environment using Data Progression.
To learn more about Dell Oracle solutions, visit www.dell.com/oracle or contact your Dell
representative for up-to-date information on Dell servers, storage, and services for Oracle solutions.
References
1. Oracle Best Practices on Compellent Storage Center, a Dell technical white paper.
Page 18
Optimizing an Oracle Database with Dell Compellent Automated Tiered Storage
2. The architectural advantages of Dell Compellent automated tiered storage, a Dell technical white
paper.
http://content.dell.com/us/en/gen/d/business~solutions~power~en/documents~ps2q11cse20110332-comp2.pdf.aspx
3. Storage Center 6.1 System Manager User Guide.
Page 19