WHITE PAPER HP Integrity Servers Running Oracle Database on

WHITE P APER
HP Integrity Servers Running Oracle Database on Linux
Sponsored by: HP
Jean S. Bozman
Carl W. Olofson
Al Gillen
May 2009
Global Headquarters: 5 Speen Street Framingham, MA 01701 USA
P.508.872.8200
F.508.935.4015
www.idc.com
EXECUTIVE SUMMARY
Linux servers are increasingly being used to support Oracle Database workloads on a
worldwide basis. Oracle Corp. is sustaining this trend by paying strong attention to
supporting the Linux operating environment and open source software stacks with its
Oracle Database products. Oracle develops Oracle Database products on Linux and
Windows platforms — and then ships production code on Linux servers early in the
rollout of those products. Given the importance of Oracle Database workloads on Linux
servers, HP is supporting these workloads with its HP Integrity servers, which are based
on Intel Itanium processors and which run Red Hat Enterprise Linux (RHEL) and Novell
SUSE Linux Enterprise Server (SLES). HP Integrity systems support scalable workloads
with their large cache size, support for large memory and I/O, and a range of hardware
and software features that ensure high availability for database workloads and the
applications that access those databases. All of these attributes are important to missioncritical Oracle Database workloads supporting important business processes and
ensuring business continuity.
HP and Oracle have long been partners in supporting Oracle Databases on HP
server systems. HP Integrity servers support technical and commercial workloads,
and these workloads can be "scaled up" on a single server system or clustered
together to support scale-out deployments. This support for customer choice is typical
of HP's approach to run a wide spectrum of enterprise workloads on its systems. The
combination of HP Integrity servers and Oracle Database provides a platform for
flexible deployments supporting business agility, even as it supports business
continuity for ongoing business operations.
INTRODUCTION
Today's datacenter bears little resemblance to the datacenter of ten, or even five,
years ago. Taking the place of midrange and high-end enterprise servers are rows of
rack-optimized and blade server systems, a process that began during the economic
downturn of 2001–2003, when volume servers — that is, servers priced less than
$25,000 — became the chief growth drivers for the worldwide server market in terms
of both revenue and unit shipments.
As a result, many enterprises have deployed dozens, hundreds, or even thousands of
volume servers in their datacenters, which in turn has led to increased operational
costs associated with items such as IT staff time and power and cooling. To reduce
these operational costs, many IT organizations are now consolidating workloads
onto fewer volume servers, onto bladed server chassis, or onto scalable midrange
enterprise servers. Combined with virtualization and automated provisioning, server
consolidation is helping enterprises not only to reduce operational costs but also to
manage their servers more efficiently.
Emergence of Linux
Linux servers are playing a strong role in the IT transformation process. The
worldwide Linux server marketplace grew significantly from $2.9 billion in 2003 to
$7.3 billion in 2008, which represents a five-year compound annual growth rate
(CAGR) of 20%. Within the overall Linux server market, Linux/Itanium server revenue
accounted for a total of $2.4 billion from 2003 through the end of 2008.
As the deployment of enterprise workloads on these Linux servers increases, so does
the need to provide scaling mechanisms for these workloads as well as the need to
protect them with reliability, availability, and serviceability (RAS) features; security;
and advanced management software. IDC expects this trend to bring more growth to
the Linux/Itanium segment of the worldwide Linux server market because the
attributes of Itanium were designed to support enterprise workloads.
In the early days of Linux adoption, many of the workloads Linux supported were in the
IT infrastructure and Web infrastructure categories. Today, a wide range of enterprise
workloads, including business processing (e.g., OLTP and line-of-business [LOB]
applications, including ERP), decision support (e.g., database-oriented workloads and
business intelligence [BI]), and collaborative workloads (e.g., email and groupware), are
running on Linux servers. This adoption has occurred because the Linux software
ecosystem is maturing, providing a wider range of ISV applications in the
enterprise/commercial market space. As a result, business continuity (including highavailability components and disaster recovery planning) and ongoing support for
business processes are both critical features that must be incorporated into enterprise
Linux server environments.
As the process of IT transformation continues, organizations must find ways to "map"
the workloads within their changing infrastructure that need the most capacity,
reliability, availability, and security. This mapping process reveals which servers will
be the most scalable resources for Linux enterprise workloads — and which will need
the highest degree of availability for important business applications and data.
DAT AB ASE WORKLO ADS ON LINUX
In a multi-tier computing infrastructure, databases hold a special place. Often located
in the data tier (as opposed to the application tier or the client access tier), databases
represent the points in the network where updates can be applied from the application
and client access tiers, and an outage at the data tier can bring the entire system to a
halt. IDC's Workloads research confirms that databases are being used to store
business transaction data for business processing workloads such as OLTP, decision
support for ERP and CRM, and business intelligence. Therefore, provisions for
availability, security, and optimization of performance are all important considerations
for the support of database workloads, especially for those that support end-to-end
applications that span the enterprise.
2
#218463
©2009 IDC
Due to their ubiquity, database workloads are already important in sites where Linux
is chosen as the preferred operating system; however, not all Linux platforms are
inherently scalable, nor are they all outfitted with the hardware-based RAS features
and high-availability software that are desirable to ensure continued access to
information stored in databases. Therefore, customers must carefully consider the
needs of the application workloads that they plan to deploy on Linux servers — and
take steps to ensure that all reliability and availability characteristics will be present
when mission-critical workloads are deployed.
Figure 1 shows the revenue generated worldwide by a range of workloads for Linux
servers in 2006 and 2007, as reported by IDC's annual Workloads study of more than
1,000 IT managers. This customer-based data, capturing how these IT managers
used their installed servers, shows that business processing (e.g., OLTP and ERP)
remains the largest workload deployed on Linux servers. However, in 2007, decision
support (e.g., data analysis/data mining and data warehouse/data mart) increased in
importance year over year for Linux servers, surpassing IT infrastructure (e.g.,
network protocol support, file/print) as the second largest workload deployed on Linux
servers. This underscores the increasing role of Linux on Itanium servers in
supporting enterprise workloads. These trends in the data are evident across
company size and across vertical market segments. Some of these workloads are
highly scalable, supporting four to eight processor sockets or more on selected large
servers. Others are smaller in size, but provide important "persistent data-store
support" for Web-enabled workloads.
FIGURE 1
Worldwide Revenue of Itanium-Based Servers Running Linux
by Workload Type, 2006 and 2007
Business processing
Decision support
Collaborative
Application development
IT infrastructure
Web infrastructure
Technical
Other
0
20
40
60
80
100
120
($M)
2006
2007
Source: IDC's Server Workloads research, 2007 and 2008
©2009 IDC
#218463
3
Oracle Database on Linux
Oracle's database and middleware products such as Oracle Database and Oracle
WebLogic Application Server are available across a range of Linux distributions,
including Red Hat Enterprise Linux, Novell SUSE Linux Enterprise Server, and
others. Aided by this support, the Oracle technology platform can run a range of Linux
applications and workloads across the enterprise, often in multi-tier deployments.
IDC notes that Linux accounts for a large and growing percentage of all Oracle
Database deployments. Oracle reports that Linux is its fastest-growing platform, with
nearly 30% of its database products shipped on Linux platforms. Further, IDC data
shows that Oracle accounted for 67% of the Linux RDBMS market in 2007 and that
Oracle's presence on Linux platforms grew 30% from the prior year.
In fact, Oracle develops new products on Linux and Windows platforms and then
ports them from Linux to other operating environments, underscoring the importance
of the Linux computing world to Oracle. For example, more than 9,000 developers
develop key Oracle products such as Oracle Database, WebLogic Application Server,
and E-Business Suite using Linux and Windows as their base development platforms.
In addition, Oracle is providing database tools and software development tools for the
Linux operating environment with Oracle Database.
Drivers for Adoption of Oracle Database on Linux
A number of factors lead customers to choose to use Oracle Database products on
Linux. Some customers have run Oracle on a variety of server platforms in the past,
and they are now moving some of their database workloads to Linux platforms.
Others have deployed Linux in the enterprise, and they have been running Oracle
Databases for some time. Following is a list of drivers for the deployment of Oracle
Database products on Linux server platforms:
Open source policy. Some organizations have a policy to deploy platforms
supporting open source programming within IT organizations. This policy is
driven by the ability to reduce development costs and the ability to add code to
open source software stacks in a collaborative style, without waiting for the next
round of commercial releases.
Systems cost reduction. Many users choose Linux because it enables them to
use a Unix-like environment to run non-critical workloads on low-cost server
systems that are less expensive than midrange or high-end Unix/RISC systems
that have often been deployed to run mission-critical workloads with high levels
of availability and reliability. By leveraging Oracle Database's built-in availability
features, scale-out deployments of Oracle Databases on Linux servers can take
on some of the workloads that formerly ran only on scalable Unix servers.
Standardization of IT infrastructure. Many enterprises have complex,
heterogeneous infrastructures brought about by successive waves of IT
purchasing decisions as well as by corporate mergers and acquisitions activity.
Introducing Linux servers provides these enterprises with the opportunity to
standardize large portions of their IT infrastructures, reducing the complexity and
cost of server management.
4
#218463
©2009 IDC
Linux flexibility. Very sophisticated users, such as online service providers and
large IT organizations, like the ability to modify and optimize the Linux kernel as
part of the open source model for software use. This is especially useful when
tuning and optimizing performance of Linux-based workloads running on Linux
distributions and associated middleware (e.g., the Linux software stack). Most
users, however, rely upon off-the-shelf Linux distributions supported by such
providers as HP or a commercial Linux distribution vendor rather than modifying
the operating system code.
License cost reduction. Consolidating on Linux provides the opportunity to
reduce license costs for Oracle Database deployments, with fewer copies, or
instances, of Oracle Database running on fewer, more scalable servers.
Hardware Choices for Oracle Databases Running on Linux
Customers have a wide range of hardware platforms from which to choose when they
run their Oracle Database on a Linux software stack (e.g., Linux, Linux applications,
and open source software), including x86, EPIC (Itanium-based), RISC, and even
mainframe platforms. The fact that Linux runs on this wide range of platforms is one
of the attributes that makes it attractive to IT managers within large organizations.
IDC supply-side shipment data shows that Linux Itanium servers are being shipped in
a variety of form factors, by a variety of vendors, on a worldwide basis. This
demonstrates their utility to customers and their expanding presence in the
datacenter, in IT infrastructure, and in corporate networks.
Customer evaluations of servers to support database workloads typically center on a
number of characteristics, including:
Performance and capacity scaling. This includes such characteristics
as processor capacity, memory capacity, and I/O capacity for a given server,
along with that server's ability to support both scale-out and scale-up
deployments, as needed, to support the customer workload. Servers with larger
cache sizes can support larger data sets than earlier generations of servers, and
these large cache sizes are important in supporting the fully 64-bit capabilities
of the Oracle Database.
Server support for large amounts of I/O. Database workloads are often I/Ointensive, which is why strong support for I/O throughput helps make database
processing faster and more efficient, reducing overall processing times for given
database tasks.
High-availability features. This includes hardware RAS features and software
clustering and availability software (CLAS) to address customer availability
requirements and to meet business SLAs. RAS features are often provided by
processor error-correcting code (ECC), chipsets that regulate system bandwidth
and throughput, and support for on-chip security.
Total cost of solution. This includes the cost of hardware acquisition, software
acquisition, deployment, and ongoing operation and maintenance of the entire
solution.
©2009 IDC
#218463
5
Scale Up and Scale Out
It is in the nature of databases to grow over time, and the use of scalable servers
ensures that there will be additional capacity for future database growth — and
ensures that no other workloads will interfere with the performance of the hosted
database engine.
Customers need to decide whether they will "scale up" their Oracle Database image
(or instance), meaning to deploy additional compute resources to support one Oracle
Database image in a single server cabinet, or "scale out," meaning to deploy
additional server resources by connecting additional Linux/Integrity servers in an
enterprise cluster using products such as Oracle RAC.
IDC research shows that most large organizations deploy a combination of scale-up
and scale-out server configurations, with IT departments making different decisions
based in part on varying IT skill sets and preferences. IDC research also shows that
many companies are likely to deploy scale-up configurations in critical areas of their
corporate networks while using scale-out configurations for divisible workloads such
as Web hosting or email. Enterprises with tiered server architectures tend to favor
scalable servers in the data tier and modular servers in the Web-serving tier.
However, IDC research found that as clustering and scale-out technologies have
gained functionality in recent years, customers increasingly have the option to deploy
workloads in scale-up or scale-out configurations.
Furthermore, scalable servers themselves can be clustered together, whether in
central-site datacenters or within business units. In this way, the individual server
nodes within scale-out clusters can be scalable SMP servers. Therefore, clustering
and scalable servers are not incompatible; rather, scalable servers represent an
important building block of IT infrastructure that can be deployed in a variety of ways
throughout the datacenter and the corporate network.
Many end users choose to scale up to provide improved application performance by
adding capacity within the same instance of the server operating system. These
customers prefer to deploy databases on servers that are dedicated to running the
database workload — scaling the workload by taking advantage of the on-board
resources of scalable servers. By investing in Integrity servers with more than two
processor sockets, customers have the option of populating additional sockets, as
needed, when the database grows and demand for processing power increases.
Other customers choose to deploy scale-out architectures linking multiple volume
servers or midrange servers into a single Linux cluster using Oracle RAC to connect
server nodes that can host Oracle Database instances across a cluster, allowing
IT sites to add server nodes to the cluster as computing demands grow. These
customers choose to house and to manage those systems inside a configuration of
rack-optimized servers or blade servers, using Oracle tools to automate management
of the entire RAC deployment.
6
#218463
©2009 IDC
Growing Server Resources to Meet Changing Application Requirements
Whether a scale-up or scale-out scenario is used, the size of the hardware server
itself and the amount of memory and resources that can be applied to the database
workload can be scaled up as needed so that individual server "nodes" can be
scalable. This allows customers to consolidate smaller Oracle Databases that have
been running in their environment onto a single, more scalable Oracle Database on a
more scalable Integrity server node.
Scale-up and scale-out approaches provide customers with options for deploying
Oracle Databases on Linux servers. In scale-out scenarios, multiple rack-optimized
servers may be clustered together or multiple Itanium-based blades may be linked
into a scale-out database resource. In scale-up scenarios, more scalable servers can
host large corporate databases on a single, dedicated machine, giving it full access to
all the computing resources on that server system.
Key Features of HP Integrity Servers
A Portfolio of Servers to Satisfy a Range of Workload Requirements
HP's line of Integrity servers includes models in the volume, midrange enterprise, and
high-end enterprise server classes. Volume servers include the rx2660 and rx6600
models, which are often deployed to support Oracle Databases, and Integrity blade
servers shipping within the HP BladeSystem chassis support Oracle Database,
especially for Web-enabled workloads.
For larger, more scalable workloads, midrange enterprise and high-end enterprise HP
Integrity server models can be deployed, supporting larger numbers of end users and
larger single-system-image (SSI) databases with as much as 100TB or more of data
stored within those databases. Midrange enterprise models include eight-socket
servers each with dual-core Intel Itanium processors (the rx7600 series) and
16 processor sockets (the rx8600 series), while high-end enterprise servers such as
the Superdome server models have 16 to 64 sockets, each supporting dual-core
Itanium processors. Because each socket can house a dual-core Itanium Series 9000
processor, the total number of processor cores can range up to 128 cores in top-ofthe-line Superdome models — and 2TB of directly addressable memory for a large
Oracle Database.
Support for Enterprise-Scale Workloads
One of the top drivers for selection of scalable servers is the ability of these servers to
support large database instances (also known as single system images) required for
enterprise-scale workloads. This capability allows end users, including those
accessing data analysis applications, to access a single instance of corporate data.
Because all scalable servers now support 64-bit addressing and include multiple
terabytes (TB) of attached storage, these systems can work with very large data sets
and can support searches of extremely large databases.
Pay per Use for HP Integrity Servers Running Linux
HP announced a pay-per-use model for Linux on Integrity servers in August 2007.
This model allows customers to pay for server capacity based on actual usage,
©2009 IDC
#218463
7
allowing them to "dial up and dial down" the usage and associated payments as
business needs require. Usage is determined using HP Utility Pricing Solutions (UPS)
metering software. Support for pay per use is gaining in importance as organizations
try to match IT infrastructure more closely to changing business needs. So,
seasonality, time of day, time of quarter, and peak demands associated with
specific industries can all lead to a surge in demand, resulting in peaks and valleys
of usage. HP's support for pay per use allows customers to pay for computing as
they use it and to reduce capital expenditure costs when that additional capacity is no
longer needed.
HP virtualization technology allows capacity to be added through the use of Integrity
Virtual Machines (Integrity VMs). In scalable servers, capacity is built into the system
itself and can be activated as needed. In the case of HP Integrity servers supporting
Linux, the Integrity VM feature of HP's Virtual Server Environment (VSE) is another
option that can be used to supply additional capacity to Linux instances or to add
more Linux instances to the system.
Electrical Isolation
Multiple electrically isolated partitions can be used to consolidate workloads, including
workloads running under separate operating systems. To address the need for
scalable and highly available servers to support enterprise workloads, HP offers a
range of Integrity servers that feature dual-core Intel Itanium processors and scale
from two-socket servers to 64-socket systems. In addition, midrange and high-end
Integrity servers use a cellular hardware design that improves scalability while
allowing for hardware partitioning and the electrical isolation of workloads. Each cell
contains four sockets, maintaining electrical isolation between cells, and multiple cells
are combined to build a scalable midrange or high-end Integrity system.
All HP Integrity servers support multiple operating systems, including HP-UX 11i,
Microsoft Windows Server, Linux distributions, and HP OpenVMS. This allows
IT managers to gather mission-critical workloads that had been running on dedicated
servers across the network, consolidating them on a scalable server that provides
higher RAS and security than the individual server nodes that formerly supported the
same workloads. Customers have a choice of when and where to deploy Linux on
Integrity servers within their IT infrastructure. The reasons that users are increasingly
selecting Linux to support database workloads can vary from site to site, but certainly
familiarity with Linux programming and Linux system administration models is often a
key factor in determining a move to Linux platforms for database workloads, as is
adoption of Linux and open source software stacks.
RAS Features for Reliability and Availability
Importantly, RAS features — hardware features that ensure reliability and availability
— are built into both the Integrity server platform and the Itanium processor, creating
an optimized platform to support scalable deployments of Oracle Database running
on Linux, with access to multiple terabytes (TB) of data. HP provides an array of
server technologies that deliver RAS, scalability, and security in scale-up computers.
These technologies include the following:
8
#218463
©2009 IDC
HP Integrity chipsets, including the zx2 and sx2000 chipsets. Housed on the
motherboards along with the Itanium processors, HP's innovative zx2 and
sx2000 chipsets support and protect database workloads. Both chipsets provide
additional functionality to ensure data integrity. Features include double-chip
spare memory protection, ECC and parity protection, a hardware memory
scrubber, and PCI bus fault isolation.
Explicitly Parallel Instruction Computing (EPIC) architecture supports
scaling of workloads. The Itanium processor's EPIC architecture provides high
levels of instruction parallelism, large caches, and massive processor execution
resources. It expedites scaling for database workloads that are expanding to
support more users. For SQL-based query workloads for relational databases,
additional users can be supported by additional threads that are processed in
parallel on Itanium processors.
RAS features for high availability. Itanium processors were designed to deliver
mainframe-class levels of availability through the incorporation of extensive RAS
capabilities. Itanium processors trap soft errors, such as a transistor state change
that can be triggered by an alpha particle or cosmic radiation passing through the
processor, which can cause incorrect results if ECC is not applied. Itanium
processors also trap hard errors, such as a failed memory bit, line, or device.
When RAS features are used in combination with the HP Integrity server
platform, other layers of high-availability safeguards are provided, including
partition-based isolation of workloads, system-based memory management, and
workload balancing.
Features of Itanium-Based Integrity Servers
That Support Database Workloads
One of the key "scale-up" rationales underlying the use of Itanium servers with Oracle
Databases has to do with mixed workloads rather than dedicated workloads. While
Oracle provides the means of managing dedicated workloads with low-cost x86 server
hardware and "scale-out" approaches, IDC believes that mixed workload approaches
can be effectively supported by Itanium-based servers, such as HP Integrity servers,
which ensure high levels of reliability and availability through the presence of built-in
hardware RAS, error-correcting features, and large cache sizes.
Key design features of the Itanium processor include inherent parallelism in the
processor design, large on-chip caching of large data sets, large amounts of directly
addressable memory, and multi-core capabilities. Itanium features designed to support
enterprise workloads such as scalable databases include the following:
Large processor caches with Intel Cache Safe. The large, low-latency, highbandwidth caches on Itanium processors enable more data to be staged closer to
the CPU for processing, thus accelerating system performance and system
throughput. Intel Cache Safe protects these caches from errors, ensuring the
integrity and high availability of mission-critical data.
©2009 IDC
#218463
9
Enhanced Machine Check Architecture (EMCA). Itanium processors use
EMCA, with well-defined interfaces for error handling at the hardware, firmware,
and operating system levels. EMCA enables efficient coordination throughout the
server, with appropriate handoffs to escalate corrective action as needed and to
initiate reporting functions. EMCA reduces the likelihood of data corruption, and it
virtually eliminates the chances of a server "hang-up" because EMCA functions
can automatically reset the server system in response to an otherwise fatal error.
64-bit addressing combined with support for large memory features. The
Itanium processor supports not only 64-bit operating systems, such as 64-bit
Linux, but also 64-bit workloads. When databases are run in a 64-bit environment,
they can directly address multiple terabytes (TB) of data, which greatly
accelerates database processing compared with database workloads running in
32-bit mode. To support this powerful 64-bit addressing capability, HP is
supporting large memory features in which to store large data sets being
processed by the database management system. HP Integrity servers are
designed to support large memory and I/O capacity. For example, HP's twosocket Integrity rx3600 server supports 192GB, while its four-socket Integrity
rx6600 server supports up to 384GB of memory capacity and its Integrity
Superdome server supports up to 2TB of memory.
Enterprise Scenarios for Deployment
Application workloads can be deployed by using a variety of server deployment
scenarios, ranging from single-tier deployments in which both the database and the
application reside on clustered or standalone servers to tiered deployments in which
the database and the application reside on separate, scalable server arrays.
Multi-tier deployments facilitate a mix-and-match approach to applying workloads to
hardware platforms. Specifically, some customers choose to deploy in a "split-tier"
configuration, in which applications run in the application tier (often on rack-optimized
servers or bladed systems) and scalable databases run in the database tier (often on
clustered or standalone, scalable servers). IDC also notes that Integrity blades are
being deployed to take on workloads that formerly ran on RISC processors, provided
that the workloads can be consolidated onto a single blade.
Scenarios for enterprise Oracle Database products on Linux deployment include the
following:
Scalable midrange enterprise systems. In this case, Oracle Database
products are deployed on midrange enterprise HP Integrity servers, each having
four sockets or more. Because each socket houses a dual-core processor, scaleup in this scenario provides plenty of compute resources for a growing database
workload.
Scale-out deployment of scalable server nodes. Multiple midrange enterprise
HP Integrity servers can be clustered together using Oracle RAC to add more
server nodes to support a scale-out deployment. In some cases, two Oracle RAC
clusters can be linked via a "heartbeat" to achieve a geographic cluster, further
enhancing high availability in case one site experiences disruptions or outages.
10
#218463
©2009 IDC
Mixed HP Integrity/ProLiant servers in multi-tier deployments. Tiered
workloads can use Integrity servers in the database tier (whether scale-up on
scalable Integrity servers or scale-out on volume or blade chassis servers) and
x86-based ProLiant servers in the application tier. This approach enables
enterprises to take advantage of the low cost of the x86-based servers combined
with the scalable performance and high availability of Itanium-based Integrity
servers for the enterprise database, all running a common Linux operating platform.
Blade chassis deployment. Itanium-based Integrity blade servers can be
interspersed with other types of blades within an HP BladeSystem c-Class
enclosure, including both Intel Xeon-based and AMD Opteron-based x86 server
blades in the same chassis. In this scenario, multiple blades running existing
applications can tap into the Integrity blades supporting the Oracle Database —
and the entire group of blades can be managed together to reduce overall
operational costs.
CHALLENGES AND OPPORTUNITIES
Customers benefit greatly from the ability to choose from among different vendors'
offerings when building out their IT infrastructure. This freedom of choice is further
expanded when evaluating systems to support open platforms such as Linux. As a
result, IT organizations can constantly review platform decisions; perform new
evaluations, pilots, and testing; and incorporate new technologies into new waves of
projects across the corporation.
This means that companies such as HP and Oracle need to constantly optimize their
product and services offerings as they compete with other systems vendors and other
database vendors. While the innovation cycle creates challenges for IT vendors, it
provides benefits to customers by driving continuous performance increases while
keeping prices competitive.
All system vendors and database vendors face additional unknowns, including
potential shifts in how technology will be acquired in years to come. For example, in
the future, customers will be able to obtain server processing through a variety of
approaches: acquisition (purchase or lease), the use of hosted services provided by
third-party companies, or pay-as-you-go software services (e.g., software as a
service, or SaaS). Licensing models may also change over time, with new models —
such as per-employee, per-use, or per-site pricing — emerging as alternatives to
today's per-processor or per-server pricing. IDC notes that some customers elect to
negotiate a site license that pays for all of the Oracle Databases running within the
enterprise, on an annual basis. Further, for some sites, consideration of these new
acquisition approaches may also affect, or lengthen, the customer decision-making
process in unpredictable ways.
CONCLUSION
Oracle Database workloads are often deployed on Linux servers, and HP Integrity
servers are robust, scalable platforms on which to deploy software stacks including
the Oracle Database, Linux operating environments, and associated open source
©2009 IDC
#218463
11
software. Using Integrity servers, customers can choose to scale up or scale out, or
both, depending on the size of the database instance and on the performance
characteristics required to support their particular business applications.
The HP Integrity platform has a number of technical features that support scalable
Oracle Database deployments on Linux. These features include built-in security, high
bandwidth, low-latency access to large memory and large amounts of I/O, and Intel
Cache Safe and EMCA, which are designed to protect the workloads themselves
from system outages caused by the failure of hardware or software components.
Looking forward, IDC expects the market for Linux servers to grow, with companies
hosting an increasingly broad range of workloads for both technical and
commercial/enterprise applications on Linux servers. As this happens, HP Integrity
servers running Linux will be considered and evaluated as platforms on which to
deploy demanding Oracle Database workloads. Customers who have selected Oracle
Database products and Linux as their software stack for a given workload or
application will have access to a range of deployment options on HP Integrity
platforms, including shipment in a variety of form factors: standalone, rack-optimized,
and blade servers. Based on their technical specifications and performance
characteristics, HP Integrity servers are well positioned to compete in the growing
Linux server marketplace.
Copyright Notice
External Publication of IDC Information and Data — Any IDC information that is to be
used in advertising, press releases, or promotional materials requires prior written
approval from the appropriate IDC Vice President or Country Manager. A draft of the
proposed document should accompany any such request. IDC reserves the right to
deny approval of external usage for any reason.
Copyright 2009 IDC. Reproduction without written permission is completely forbidden.
12
#218463
©2009 IDC
Download PDF