Datacenter Operations by lqf20778

VIEWS: 0 PAGES: 29

More Info
									An Oracle White Paper
February 2010



Oracle Enterprise Manager Ops Center:
Changing the Economics of Datacenter
Operations
                          Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Introduction ......................................................................................... 1
How Much Is Too Much? .................................................................... 3
External Pressures.............................................................................. 4
Operational Sciences .......................................................................... 6
   Information Technology Best Practices........................................... 6
   Technology Standardizations.......................................................... 7
Routes to Reducing Total Cost of Ownership ..................................... 8
   Monitoring ....................................................................................... 8
   Point-in-Time Provisioning ............................................................ 11
   Software Lifecycle Management ................................................... 15
   Virtualization Management............................................................ 20
Building a Business Case for
Oracle Enterprise Manager Ops Center............................................ 24
Conclusion ........................................................................................ 26
                           Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Introduction

In today’s tight economy, every organization is trying to reduce costs wherever possible.
Although IT is a key contributor to the business of any modern organization, it’s also a
significant drain on resources. Thus, it’s essential that operations find a balance between the
benefits of IT services and the costs associated with running them. An approach focused on
TCO can result in long-term sustainable savings by achieving efficiencies without sacrificing
functionality. When considering TCO in the context of an enterprise datacenter, the following
aspects are normally examined:


•   Capital asset depreciation schedules


•   System administration costs


•   Systems management


•   Asset-per-user or asset-per-administrator ratios


•   Datacenter automation


•   Technology lifecycle


•   Energy consumption


Although these are valid considerations, recent technological advances are shifting the focus
to new technologies that directly impact the cost of running a datacenter. These hardware and
software innovations lower the lifecycle costs of datacenter assets, but they also require a
higher initial investment. Such technologies can be divided into two distinct categories:




                                                                                                                                 1
                         Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




•   Technology standardization and advancements in operational science. Improvements
    in datacenter management and system interoperability standards and the increased support
    for these standards by the different datacenter components and subsystems have enabled
    the IT industry to develop datacenter efficiencies.


•   Intrasystem abstraction, management, and observability. Improvements in observability
    and the virtualization technologies that are integral to the native operating system (OS) and
    hardware have enabled computing, storage, and networking resources to be abstracted.


Technologies and methods that help optimize datacenter asset lifecycle costs have developed
rapidly in recent years. Included are IT best practices that have been standardized through
Information Technology Infrastructure Library (ITIL) and Control Objectives for Information and
Related Technology (COBIT) specifications, as well as interoperability standards that include
standard data models, protocols, and workflows. These methods and standards define the
interfaces and interactions of each datacenter component, help reduce complexity, and enable
the development of technologies to automate many datacenter administration and
maintenance tasks.


At the same time, technologies to abstract datacenter services—such as resource usage
monitoring, system management standards, hardware fault management, and datacenter
automation—are becoming increasingly standard in datacenter systems. The interaction of
these technologies, both within and across systems, is what facilitates automation and, in turn,
lowers the TCO of IT solutions.


The benefits afforded by these technologies, however, come at a cost in resources,
infrastructure, and software licenses. To assess both the full costs and overall benefits of such
datacenter management solutions, this white paper examines the primary inefficiencies found
in typical datacenters, and then presents the strategies commonly used to resolve them. The
document also demonstrates how Oracle Enterprise Manager Ops Center can be used as a
cost-effective means of addressing these inefficiencies.




                                                                                                                               2
                              Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




How Much Is Too Much?
Although many purchasers of IT solutions evaluate the total lifecycle costs of such solutions, they still
give primary consideration to the initial purchase price, ironic because a lower lifecycle cost, that is,
cost from purchase to decommission, normally necessitates a higher initial price point. This is because
the same features that drive down the lifecycle cost⎯for example, energy efficiency, smaller form
factors, improved reliability, better quality, solid-state storage, and remote management
capabilities⎯also increase the purchase cost.
Further complicating matters, budgets are typically allocated for relatively short periods, for example,
quarterly. This means that offsetting a higher purchase price against future savings might require a
decision at a more-senior level, a more-detailed and compelling business case, a longer business
horizon, or a combination of all these factors. An additional consideration is that the budgeted
purchase cost is specific and must be spent, whereas the calculation of lifecycle savings justifying the
solution is inherently less accurate. Thus, the question of how much you would need to save over the
life of the solution to justify a specific initial cost increase is not a simple one to answer (see Figure 1).
An increased initial purchase price makes financial sense if the savings in TCO offset the initial
cost increment.
A TCO analysis is calculated by summing the following factors in terms of present-day funds:
•   Purchase price
•   Maintenance cost
•   Power
•   Installation cost
•   Real estate cost
•   Administration cost
•   Training cost
•   Value of alternative use of funds, or opportunity cost
•   Parts and replacement cost
•   Software lifecycle cost
•   Decommission cost




                                                                                                                               3
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 1. As the investment in features that reduce TCO increases over time, TCO falls.


Whereas individuals find it easy to justify spending more for an efficient car that uses less fuel or a
better-quality appliance that requires less maintenance or is more reliable, most of the effort in buying a
technology solution is dedicated to maximizing its business benefit. As a result, the solution’s TCO is
often ignored, in spite of the incremental burden on the infrastructure this creates. The tendency to
ignore TCO persists even in the face of shrinking IT budgets coupled with increasing demands for IT
services. Ultimately, to lower a solution’s TCO, an enterprise must be willing to examine the full cost
of the solution and consider making a larger initial investment.


External Pressures
Operational costs combined with purchase price normally dominate TCO for most IT solutions.
However, enterprise IT operations can be highly inefficient, adding to the TCO of a solution. 1
Consider the following:
•   Average CPU use is often very low: a 2008 Uptime Institute and McKinsey & Co. report puts it at
    only 6 percent, and anecdotal evidence rarely shows it to be higher than 10 percent. 2
•   Although regulations requiring corporations to reduce their carbon footprints are becoming
    increasingly widespread, the cost of energy to power and cool a server over its lifetime is estimated at
    50 percent of the cost of its purchase. 3




1 Refer to www.baselinemag.com/c/a/Virtualization/The-Growing-Appetite-for-Virtualization.
2 Uptime Institute and McKinsey & Co., Revolutionizing Data Center Energy Efficiency, July 2008,
www.mckinsey.com/clientservice/bto/pointofview/pdf/Revolutionizing_Data_Center_Efficiency.pdf.
3 IDC, John Humphreys, Beyond Consolidation: Adopting Virtualization for Mobility, 2008.




                                                                                                                                    4
                           Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




•   Business services are looking to improve response times and availability by building applications as
    encapsulated, self-contained virtual appliances that can be deployed any number of times on any
    number of servers.
•   To remain competitive, corporations must spend a significant percentage of their revenue on
    IT solutions.
•   Major vendors provide five-year warranties for their servers, which help determine the time the
    servers are deployed in production.
•   A device’s depreciation schedule in the United States is three years.4
These operational constraints, costs, and requirements create a significant overhead, causing
organizations to constantly purchase new equipment and update existing solutions—a situation
exacerbated by interdependencies that affect other software and hardware components whenever
equipment is replaced or software updated. Resolving this issue can require regression testing and, in
some cases, updating other components.
As software platforms develop and technology vendors bundle more features with software platforms,
enterprise IT developers are using these new features to create complex business solutions. However,
although these efforts often generate a tangible business benefit to the enterprise, they often lack the
ability to integrate effectively with management frameworks and are thus difficult to administer and
maintain. In this way, bundled components can increase TCO. The flip side of that equation is that
some of the new bundled technologies, such as hypervisors, create significant opportunities to reduce
operational cost and lower TCO.
Figure 2 illustrates the effect of investing to reduce operational cost. Over time, assets represent less
cost as the enterprise depreciates its hardware on a schedule. Unfortunately, as these costs lessen, the
operational cost of maintaining the equipment in production normally increases. Appropriate
investment based on a complete TCO analysis that takes both trend lines into consideration can
generate a positive ROI.




4   Refer to michigan.gov/documents/Life_Cycle_Boilerplate_Report_86875_7.pdf.




                                                                                                                            5
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 2. As the initial investment in purchased hardware is depreciated, the operational cost increases.




Operational Sciences
A TCO analysis must include the cost of the datacenter’s daily activities. Optimizing operations
through automation lowers the TCO for all the solutions hosted in the datacenter. Given that some
optimizations are bound to offer greater benefits than others, identifying which optimization measures
are most effective in different settings and sectors can help planners identify the best way to implement
optimizations in a specific datacenter. There are two primary sources of information that can be used
to help discover these optimizations: IT best practices and technology standardizations.

Information Technology Best Practices
Over the last decade, the IT industry has expended significant effort in creating best practices, the
result of which is a number of recommendations, concepts, and policies⎯for example, ITIL Version 3,
COBIT, and ISO/IEC 20000. All these guidelines address datacenter operations, IT practices, and
layout practices to effectively run datacenters. By examining the focus of these standards, it’s possible
to determine the concerns of IT operations across different environments.
To have the greatest impact on TCO, resources are applied to the datacenter operation areas where
they are deemed to be the most effective. COBIT and the other IT best practices have documented
these areas as workflows. Figure 3 includes the focus areas of the COBIT specification and illustrates
the activity flow when implementing a best practices analysis and implementation of enterprise IT, as
defined in COBIT. Systems management software has the most significant effect on the Deliver and
Support, Monitor and Evaluate, and Information phases of the workflow—all of which can be
successfully addressed with the right mix of features from off-the-shelf solutions such as Oracle
Enterprise Manager Ops Center.




                                                                                                                                    6
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 3. The flow of activities when implementing a best practices analysis and implementation of enterprise IT, as defined in COBIT



Technology Standardizations
Many vendors are currently making significant investment in standards-compliant systems management
technologies. By adhering to interoperability standards, vendors allow customers to build management
solutions based on interoperable components from different vendors. The history of interoperable
standards-based software frameworks can be traced back to the beginning of networked computing, in
the early 1980s. As IT operations moved away from the centralized computing models typical of
mainframes, they needed to find new technologies, designs, and concepts for distributed computing
networks comprising heterogeneous components. Project Athena—often cited as the first major effort
to address these challenges—generated technologies that supported distributed computing by defining
and implementing communication protocols and data model standards.5
Other initiatives that followed Project Athena and defined technology standards included working
groups such as the Open Software Foundation’s Distributed Computing Environment, in the 1990s,
and, more recently, the Distributed Management Task Force. These working groups support effective
management of IT systems by allowing different organizations to collaborate in developing, validating,
and promoting systems management standards.
The focus of the IT best practices standards and industry working groups, as illustrated by their top-
level concerns and organizational structures, indicate the following as the primary concerns of
datacenter managers in relation to TCO analysis:
•   Monitoring of hardware and software health and performance, security, compliance, and inventory
•   Nonrecurring activities, for example, new OS installations



5Project Athena was a joint project by the Massachusetts Institute of Technology, Digital Equipment
Corporation, and IBM to produce a campuswide distributed computing environment for educational use.




                                                                                                                                        7
                          Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




•   Software maintenance and patching
•   Virtual machine (VM) management and monitoring


Routes to Reducing Total Cost of Ownership
Oracle Enterprise Manager Ops Center addresses each of the areas discussed in the previous section,
as the following subsections will demonstrate.

Monitoring
When evaluating monitoring frameworks, consider the following issues to assess quality:
•   Does the solution use the system’s currently available telemetry in both the hardware and software
    stacks?
•   Do the monitoring and management capabilities create a load on the systems being monitored?
•   Does the system use agentless techniques to minimize administrative overhead?
•   Does the system monitor useful attributes?
•   Does the system expose the most common faults in the hardware stack?
•   Is the information gathered presented in a useful way?
•   Does the agentry use cutting-edge technology and not simple administration scripts and persistent
    storage?
•   Does the solution allow other systems to access the information?
•   Does the solution allow for event information to flow into another product to manage the next step
    in problem resolution?
Oracle Enterprise Manager Ops Center, which takes full advantage of existing hardware and software
to create a minimal incremental network load, includes all data relevant to Oracle’s Sun Fire and Sun
SPARC Enterprise systems. Able to detect a system by communicating with its OS or service
processor, Oracle Enterprise Manager Ops Center provides enough data about the potential faults and
attributes of Sun Fire and Sun SPARC Enterprise systems that operational staff can monitor them with
minimal training, thus reducing the overhead associated with deploying Oracle Enterprise Manager
Ops Center.
Modern servers use built-in service processors to monitor their own power consumption—a feature
that allows enterprises to track IT energy costs and the resulting carbon emissions. Oracle Enterprise
Manager Ops Center makes it possible to monitor energy use on any capable system, and then
aggregate that information relating to groups of servers so that energy use can be associated with a
specific application. Figure 4 illustrates this with an annual forecast of energy cost based on three
weeks of real-time power monitoring. Power usage is sampled every five minutes, and the hourly
average is used to calculate the kWh figure, which is easily mapped to the cost.




                                                                                                                           8
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 4. By leveraging the data collected and exported by Oracle Enterprise Manager Ops Center, it’s easy to estimate the cost of energy for a

server based on a sampling of its power use.


The service processor accumulates hardware status information and makes it available to Oracle
Enterprise Manager Ops Center. Figure 5 illustrates the service processor’s significant role in enabling
remote monitoring and management (called lights-out management, or LOM) in Oracle Enterprise
Manager Ops Center and products from other vendors.




                                                                                                                                            9
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 5. ILOM impacts all aspects of datacenter management.


By taking advantage of the service processor for integrated LOM (ILOM), Oracle Enterprise Manager
Ops Center allows hardware to be managed centrally⎯without deploying agents. However, for
monitoring at the OS level, Oracle Enterprise Manager Ops Center uses a software agent, which can be
automatically installed from the Oracle Enterprise Manager Ops Center controlling tier. Using minimal
resources⎯for example, a mobile phone version of the Java platform, Java Platform Micro
Edition⎯this capability can be achieved without placing a significant load on the server.
The more-subtle difference between the two forms of telemetry collected by Oracle Enterprise
Manager Ops Center is the fact that one (hardware information) is a polling implementation and the
other (OS level) is a push. See Figure 6. In the polling solution, the Oracle Enterprise Manager Ops
Center controlling layer asks the service processors for information. Conversely, in the push design, the
Oracle Enterprise Manager Ops Center agents push information to the controlling layer and pull new
tasks down to execute. This helps ensure that the management solution does not saturate the network.




Figure 6. Oracle Enterprise Manager Ops Center uses several technologies to implement OS provisioning and universal discovery. Lightweight

agents use pull methods with minimal networking overhead to provide an effective automation platform.




                                                                                                                                      10
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Oracle Enterprise Manager Ops Center revolutionizes the enterprise management world by leveraging
architectural implementations mainly found in the internet space. Similar to how users subscribe to
blogs with Really Simple Syndication and the Atom Syndication Format, Oracle Enterprise Manager
Ops Center agents use the same techniques to communicate tasks and exchange information among
their agent, proxy, and controller layers. These subscription-based standards enhance the scalability of
Oracle Enterprise Manager Ops Center. Employing a central console and administration portal to track
all datacenter assets across complex and distributed networks, Oracle Enterprise Manager Ops Center
represents a cost-effective solution in that it eliminates the need for dedicated devices.
One problem with datacenter monitoring solutions is that they can generate an inordinate amount of
meaningless events and false alarms, which can obscure real issues. To address such problems,
datacenter managers must invest in tools to automatically preprocess this data before administrators
can evaluate it. Oracle Enterprise Manager Ops Center eliminates the need for such tools by selecting
the OS performance attributes that are most likely to indicate a real problem on the system, and by
displaying the collected information in easy-to-understand graphs and tables that allow datacenter staff
to set site-specific thresholds on logical groups of systems or individual systems (see Figure 7).




Figure 7. Oracle Enterprise Manager Ops Center’s event interface allows administrators to set site-specific alarm thresholds on groups or

individual systems.



Point-in-Time Provisioning
Oracle Enterprise Manager Ops Center uses point-in-time provisioning mechanisms to handle tasks
initiated by external occurrences, such as installing a new OS release at a given time.
Given that there are several different ways to install an OS, most of the cost related to OS installation
can be attributed to the following factors:
•   Connecting to the server in a way that allows the administrator to provision it remotely
•   Deploying a system that maintains standardized OS profiles together with a mechanism that maps
    profiles to their business function
In the past, serial consoles, terminal servers, remote UPS control devices, and more were required to
provide administrators with remote access to systems before installing an OS. Typically requiring a




                                                                                                                                            11
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




direct physical connection to systems and providing software to interact with more than one system,
such solutions were costly and usually employed proprietary feature sets and interfaces. Today, most
vendors provide remote access through bundled LOM interfaces that support standard protocols.
Oracle Enterprise Manager Ops Center uses these interfaces to support features that require a direct
interface to system hardware for monitoring, discovery, OS installation, and other bare-metal features.

Operating System Provisioning Tasks

The different aspects of OS provisioning supported by Oracle Enterprise Manager Ops Center are
described in the subsections that follow.

Discovery

The IT staff members who physically install servers in racks are not always responsible for installing
the firmware and OS. When such is the case, a semimanual system alerts administrators that a new
server is available to be provisioned for operation. Oracle Enterprise Manager Ops Center automates
this notice by discovering datacenter assets using standard mechanisms and protocols such as secure
shell, Telnet, Sun Service Tags, Intelligent Platform Management Interface, and Simple Network
Management Protocol. In this context, Oracle Enterprise Manager Ops Center also supports a LOM
configuration, allowing administrators to easily use LOM features available in Oracle hardware
(see Figure 8).




Figure 8. Oracle Enterprise Manager Ops Center’s discovery user interface provides a rich set of predefined information for Oracle hardware.



Active Management

Once Oracle Enterprise Manager Ops Center accesses the service processor, it can control the location
identifier, power state, network boot configuration, and firmware version. Each of these active
management features helps improve datacenter operational efficiency. Take, for example, the firmware
version control feature: Many organizations fail to regularly update the firmware on their servers
because it’s a manual process that requires significant administrative resources. Oracle Enterprise




                                                                                                                                         12
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Manager Ops Center changes this scenario by supporting firmware maintenance through firmware
compliance reports, as shown in Figure 9, and by streamlining the firmware update process as follows:
•   An administrator obtains the firmware update and checks it into Oracle Enterprise Manager
    Ops Center.
•   A specific firmware update is associated with a specific platform type.
•   The action required after the firmware has been updated⎯for example, a service processor
    reset or a host reboot⎯is defined.
•   The servers requiring firmware updates are identified.
•   The servers requiring updates are queried to check whether they need a chain of firmware
    versions to be installed.
•   The administrator can then trigger the firmware update for the required servers.




Figure 9. Firmware compliance reports help organizations streamline the firmware update process.


For the benefit of onsite datacenter staff, Oracle Enterprise Manager Ops Center supports remote
active management, for example, remotely activating or deactivating the server indicator light to
identify a server that requires physical maintenance. See Figure 10.




                                                                                                                                 13
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 10. Once an individual server or logical group of servers is selected through the Oracle Enterprise Manager Ops Center interface, an

administrator can perform active management tasks remotely.


Rolling power outages have become a common occurrence affecting datacenters in many locations.
Given a proactive warning about a pending power outage, Oracle Enterprise Manager Ops Center can
gracefully power down entire populations of servers⎯reducing the need for onsite staff and dedicated
remote power solutions and reducing the risk of file system or application corruption.

Bootstrapping Operating System Provisioning

Integrating a number of servers into a datacenter is an effort-intensive and time-consuming task. If, for
example, 20 servers with five network interface controller ports each are installed, this can result in 100
Media Access Control (MAC) addresses that must be tracked and maintained in the Dynamic Host
Configuration Protocol (DHCP) configuration database. Not only can Oracle Enterprise Manager Ops
Center identify which server a MAC address belongs to, but it can also track the specific port and
maintain this information in the DHCP configuration upon request. If the server is tagged with a label
that identifies the rack it is in, Oracle Enterprise Manager Ops Center can be employed to list all the
MAC addresses used by the servers in that rack. This feature further increases the level of automation
in the datacenter and reduces the need for manual management.
Oracle Enterprise Manager Ops Center automation can be used to maintain the topology of the
network. For example, if a server is remotely selected on a specific subnet, Oracle Enterprise Manager
Ops Center is designed to follow its mapping to discover the closest caching proxy server to the
selected server. It can then help ensure that the selected server is accessed through the proxy rather
than the controller layer. This allows for significant savings in network bandwidth because the proxy
local to that machine is used to cache its requests. If other hosts on that subnet are selected in the
future, Oracle Enterprise Manager Ops Center is designed to use the same local caching proxy server.




                                                                                                                                          14
                         Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Once the network is configured, Oracle Enterprise Manager Ops Center can use OS provisioning
technologies such as JumpStart, Kickstart, and AutoYaST. Oracle Enterprise Manager Ops Center
provides administrators with a unified user interface for all of these bootstrapping technologies. This
helps reduce complexity and operator errors and saves administrator time. For example, Oracle
Enterprise Manager Ops Center can create the required file systems, shares, and services. The
administrator can then check in the appropriate software images or Oracle Solaris Flash archives and
decide whether to add finish scripts or JumpStart Enterprise Modules from a centralized source. Oracle
Enterprise Manager Ops Center’s ability to provide administrators with transparent access to multiple
OS provisioning tools makes the choice of OS simply an attribute of the task at hand. Oracle Solaris,
Red Hat Enterprise Linux (RHEL), Oracle Enterprise Linux (OEL), or SUSE Linux Enterprise Server
(SLES) can be selected as part of the task of deploying a business service.
To enable further automation, Oracle Enterprise Manager Ops Center allows administrators to save
configuration choices as profiles and select from a list of saved profiles when needed. An enterprise
can create and maintain standard configurations by requiring administrators to save OS images, OS
profiles, and finish scripts in a centralized repository for Oracle Solaris, RHEL, OEL, and SLES. This
saves administration resources and helps reduce errors and inconsistencies. In addition, Oracle
Enterprise Manager Ops Center can assist in the actual provisioning of the OSs through an intuitive,
Web-based user interface. As a result, issues resulting from inconsistent OS installation are minimized.
Using systems like Oracle Enterprise Manager Ops Center helps ensure commonality and account-
ability, resulting in a solid, reproducible system and a record of who installed what on any given system.

Software Lifecycle Management
No matter how perfect the system is when first activated, unexpected problems can occur, such as
when an end user inadvertently employs the application stack or OS utility in an undesirable manner.
When such problems occur in a distributed network, more than one server can be affected. As a result,
datacenter operations staff members invest significant resources troubleshooting unexpected issues.
One of the most effort-intensive aspects of server maintenance is updating software with the latest
patches. All major OS vendors offer patch updates to fix or prevent issues. The most damaging and
costly potential result of patching is when a specific patch unexpectedly and adversely affects a
particular OS or application configuration. Ideally, administrators should be able to focus on
minimizing this risk.
With the tools available today, there’s no reason for administrators to be spending their time on
patch maintenance tasks. Instead, all the mundane, resource-intensive chores involved in patching—
from downloading the patch to saving the current state of the systems (in case rollback is required),
identifying patch candidates, installing prerequisite patches, and finally installing the patch itself—
should be automated. Oracle Enterprise Manager Ops Center does just that, leaving administrators
with more time to focus on more-important issues. Oracle Enterprise Manager Ops Center can
also protect against preventable unplanned outages caused by patches, as detailed in the
following subsections.




                                                                                                                        15
                         Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Dependency Management

Every installed software component creates a dependency tree, resulting from the different
components it uses. Given that no single utility, application, or OS remains in its original configuration
for long, dependency trees can become very complex. Because a standard operating environment
(SOE) is customized through site-specific configurations, bug fixes, security fixes, and removal of
superfluous or unwanted packages, the SOE becomes an interdependent ecosystem of undocumented
dependencies.
Oracle Enterprise Manager Ops Center provides a software service to help reduce the cost of
application ownership by discovering these dependencies across OS software components from
multiple vendors. The service maintains connections with the Red Hat Network, Novell Network,
Oracle, and SunSolve, which it uses to continuously scan available information on different OS
distributions. It scans patch readme files, package depend files, RPM Package Manager header files, and
other vendor-supplied sources of information. Next, Oracle Enterprise Manager Ops Center examines
the symbols referenced in binary files within the package or RPM Package Manager to discover
dependencies at the library level. With this combined information, the service forms a dependency tree
that flows from the original distribution to the branches introduced when a new package or patch is
released. In addition to discovering these new dependencies, the service maintains a historical record of
all the dependencies that ever existed between the different software components.
The service is provided to subscribers through a secure, lightweight Web service. If the information is
needed without an internet connection, it can be manually downloaded for use in a disconnected
datacenter. Oracle further validates these dependency trees by testing various configurations.
Simply having a constantly updated catalog of all software ever produced by Red Hat, Novell, and
Oracle saves on operational resources; however, the in-depth software dependency analysis also
increases reliability because the dependencies are dynamic and can change over time.

Managing and Applying Knowledge

Knowledge is power, and the more an enterprise knows, the less it has to invest to resolve issues.
Vendors constantly release software updates to fix defects or protect against security risks. Some
vendors report periodically on the most prevalent issues that affect their users, and the U.S.
government funds organizations that track known security-related flaws and fixes. As you might
expect, though, tracking all this information is resource intensive. And the costs only continue to
increase as the enterprise tries to determine what knowledge is applicable across different development,
test, and production environments.
Oracle Enterprise Manager Ops Center can be configured to perform periodic analysis, obtain
knowledge from external sources, check the applicability of this knowledge to individual systems across
the corporation, and trigger patch installation, if necessary—capabilities that can result in significant
benefits and cost savings.
The Oracle Enterprise Manager Ops Center baseline analysis report can report on vendor
recommendations or site-specific custom profiles consisting of files, packages, or patches. In the
example in Figure 11, the Oracle Enterprise Manager Ops Center baseline analysis report found 94




                                                                                                                        16
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




changes recommended by the vendor for Oracle Virtual Desktop Infrastructure. Automating these
tasks is key to reducing TCO.




Figure 11. Oracle Enterprise Manager Ops Center’s baseline analysis report can provide vendor recommendations or site-specific custom profiles.



Support for Deployment

Even if dependencies are well known, a maintenance procedure can still fail on multiple systems. This
is because each OS has native utilities for installing software, and these utilities depend on files stored
locally on each server. Problems can occur in these local repositories; thus, it’s important for the
system to identify any such issues before starting maintenance procedures. Unfortunately, these issues
can be time consuming to troubleshoot, can require assistance from the vendor, and can cause IT staff
to miss difficult-to-schedule maintenance windows.
Oracle Enterprise Manager Ops Center can help reduce the risk of these issues occurring by allowing
administrators to initiate simulated updates before starting a maintenance procedure. The simulation
downloads patches locally to the target machines and calls the native OS utilities to simulate a patch or
package installation. Running a simulation also benefits the update process by delivering patches and
packages to the target systems before entering the maintenance window. In this way, Oracle Enterprise
Manager Ops Center reduces the risk of unplanned downtime and potentially large business impact.
Oracle Enterprise Manager Ops Center also supports Oracle Solaris Live Upgrade, which enables
administrators to create alternate boot environments. They can then install patches on an inactive




                                                                                                                                       17
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




boot environment, reboot the system to run from it, and revert to the previous boot environment,
if necessary.
Oracle Solaris Live Upgrade and Oracle Enterprise Manager Ops Center jointly provide significant
operational savings. Figure 12 illustrates how Oracle Enterprise Manager Ops Center offers all
alternate boot environments available on a system and allows the administrator to choose one during a
patch or package deployment. In addition, the administrator can choose to remain with the currently
active boot environment or immediately boot into a patched alternate boot environment. Oracle
Solaris Live Upgrade enables operational savings by allowing the administrator to roll back changes,
thus reducing downtime. Oracle Enterprise Manager Ops Center amplifies these benefits by making it
easier to use Oracle Solaris Live Upgrade across multiple systems.




Figure 12. Oracle Enterprise Manager Ops Center supports Oracle Solaris Live Upgrade by allowing the administrator to select and patch

alternate boot environments.



Operational Philosophy

Many corporations depend on IT services to support core business functions. For these corporations,
the financial impact of any level of unplanned downtime is immense. As a result, some corporations
tend to define proven software profiles, composed of custom combinations of packages, patch levels,
and configuration files that are maintained under strict version control. Other corporations rely on
vendors. Whatever the case might be, Oracle Enterprise Manager Ops Center allows the enterprise to




                                                                                                                                         18
                           Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




determine whether its entire IT infrastructure is fully up to date and compliant with corporate
standards and the vendors’ baselines.
Oracle Enterprise Manager Ops Center can proactively recommend any patches or files that are needed
for a datacenter’s servers automatically through connections to the OS vendors, MITRE’s repository of
common vulnerability and exposures, and a local repository of approved configurations.
Conversely, IT can choose to use Oracle Enterprise Manager Ops Center to deploy software—
patches, packages, or configuration files—on demand. The difference is subtle, but it’s important that
the selected software lifecycle product allow the enterprise to take either the reactive or proactive
approach, or both.

Auditability

Due to increased regulatory consignments and obligations across certain industry sectors, organizations
are required to document compliance with regulations in an auditable manner. Specifically, datacenter
managers must be able to prove that they have a system in place that meets the following requirements:
•   Software is not contaminated.
•   Any software or hardware installation is fully documented.
•   Standard configurations of operating environments are maintained.
•   Vendor-provided fixes that affect system reliability are installed within a specific time frame.
Not adhering to these guidelines can, in some cases, affect the stability of the IT infrastructure and
result in penalties in the form of regulatory fines. Oracle Enterprise Manager Ops Center can help
avoid these regulatory risks by helping datacenter staff achieve compliance.

Coverage

Although Oracle Enterprise Manager Ops Center is a Sun product, it covers RHEL, SLES, OEL,
Oracle Solaris, and Microsoft Windows equally well. Windows coverage is achieved through a seamless
integration with Microsoft System Center Configuration Manager. From an operational perspective,
cost savings can be realized by reducing diversity and deploying as few SOEs as possible in the
datacenter. Some successful datacenters deploy only a single SOE, though this is increasingly rare in
today’s commodity-driven heterogeneous market. In addition, it’s often beneficial to stage IT solutions
against one another and select the most technically sound solution for a given business application.
Oracle Enterprise Manager Ops Center allows an organization to maintain a multiplatform strategy on
SOEs and avoid vendor lock-in.
Oracle Enterprise Manager Ops Center’s profile creation wizard reduces the overhead of moving
between vendors and helps create custom configurations. See Figure 13. Software profiles can be
implemented that mix different products from different vendors. In addition, Oracle Enterprise
Manager Ops Center can determine which payloads should be deployed to which target host, reducing
the risk of deployment errors and automating the process of separately testing compliance against
operational standards for each vendor.




                                                                                                                          19
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 13. Oracle Enterprise Manager Ops Center’s profile creation wizard uses a menu-driven interface to make it easy to move between vendor

distributions and create custom configurations.



Virtualization Management
As early as the mid-1990s, the concepts that evolved into current virtualization technologies began
taking shape. Primary among them was business service abstraction, which decouples the functionality
or service an IT solution delivers from its underlying hardware infrastructure. This meant that
consumers of IT services were required to define their needs in terms of the desired workload, and
then IT operations would deliver the resources. Technologies that enabled horizontally scaled
systems—at both the application and database levels—allowed IT operations to consolidate more of
the common workloads on the same physical server and increase average use, resulting in significant
cost savings.
Horizontally scaled application layers were containerized in resource use boxes that allowed them to be
placed on fewer physical servers. Eventually, these encapsulated application layers became VMs on
hosted hypervisors, which enabled a higher degree of separation between workloads. Hypervisors
allowed IT to stop worrying about OS or application resource conflicts. Developers of application
stacks began to investigate what OS features were needed, at a minimum, to run the applications. The
objective: to enable the applications to survive in a minimized SOE within a VM dedicated to that
application. Uses once limited to hardware appliances found their way back into a pure software world
where hardware no longer mattered.
With the development of low-cost, shared storage and continued improvements in virtualization
technologies, complex environments of the past have been transformed into mobile business services,
able to move from one server to another with ease. Subsequent to this came an age in which problems
such as form factor, datacenter real estate, power consumption, carbon emissions, and workload




                                                                                                                                       20
                                  Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




isolation were solved. Mobility increased the ability to rapidly move the business service to different
machines. And hypervisors allowed these solutions to become available on low-end architectures.
Today, virtualized solutions are becoming increasingly common, and the technology is sufficiently
mature that any datacenter that doesn’t use them for everyday workloads is at a competitive
disadvantage. Two of the most widely used virtualization technologies today are Oracle Solaris
Containers and Oracle VM Server for SPARC.
Oracle Solaris 10 is one of the most technologically advanced OSs in the datacenter today, while Oracle
Solaris Containers (referred to as “zones” in Figure 14) represents one of the best-performing
virtualization technologies available, with support for multiple instances of the OS running on a server
with minimal overhead.
In addition to Oracle Solaris Containers and hypervisors, chip multithreading (CMT) processors are a
leading driver of datacenter consolidation. The UltraSPARC T-series processor was one of the first to
bring CMT to the market, with high performance per thread and large thread counts. In small form
factors, servers are available that provide up to 256 virtual CPUs on a single small server. Today, every
server that ships with an UltraSPARC T-series processor includes a free hypervisor in the firmware on
the ILOM. This hypervisor supports multiple OS kernels with direct access to threads, memory, and
other hardware resources.




Figure 14. Oracle Solaris Containers support multiple instances of the OS running on a server with minimal overhead.


Virtualization should not increase operational costs; instead, it should lower them, because it helps
consolidate servers and drives up use. But just like physical servers, a large number of VMs create




                                                                                                                                 21
                                 Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




significant operational overhead, unless the correct processes and procedures are used to manage them.
In Centrify’s 2009 survey on the virtualization market, respondents cited organizational and
administrative issues as the main barriers to adoption of virtualization. 6
Once the operational issues are resolved, it is possible to benefit from the cost savings of virtualization
without negating these savings with increased operational costs. The main aspects of virtualization
management that generate the greatest savings in terms of datacenter costs are detailed in the
subsections that follow.

Lifecycle Control

To effectively control the VM lifecycle, a management tool should be able to
•   Discover a VM and identify its relationship to a physical machine
•   Create, destroy, stop, start, clone, copy, and change the configuration during runtime of a VM
•   Perform all these actions on logical groupings of systems
The complexity inherent in mapping VMs to physical machines is not always apparent when only a few
VMs are involved. However, once the VM count is in the hundreds, operational complexity and cost
grow rapidly. Figure 15 illustrates how Oracle Enterprise Manager Ops Center automatically discovers
and maps VMs.




Figure 15. Oracle Enterprise Manager Ops Center automatically discovers the relationship between VMs and physical hosts.




6   Centrify, Virtual Data Centers⎯Market Dynamics and Security Issues, September 2009.




                                                                                                                                22
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Resource Awareness

To effectively manage resources, a management tool must
•   Take full advantage of and easily allocate the physical resources on a server
•   Observe the usage per VM from the controlling layer
•   Discover spare capacity to run more VMs
•   Automatically migrate VMs according to a predefined policy to help maximize resource use
Figure 16 illustrates the ability to group any number of servers that offer a specific business service.
These servers share common storage and a common network identity. Once the connection to each
server is established, the servers are allocated to a pool. When a VM is created, Oracle Enterprise
Manager Ops Center automatically allocates it to a server based on the usage policy selected for the
pool. During the lifecycle of the business service, Oracle Enterprise Manager Ops Center migrates the
VMs between servers in the pool, until an optimal level of use is reached on all the servers in the pool.




Figure 16. Oracle VM Server for SPARC’s guest virtual pool helps datacenter staff use their compute resources in full.


Movements of VMs are based on the usage policy selected. See Figure 17. This helps ensure that
servers are optimally loaded and aids in abstracting business service from the underlying hardware.
Features such as this help improve TCO.




                                                                                                                                  23
                                 Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Figure 17. Oracle Enterprise Manager Ops Center allows administrators to choose between two use policies, scheduling options, and

notification rules.




Building a Business Case for Oracle Enterprise Manager Ops
Center
There are a number of ways to estimate the operational efficiencies and savings enabled by Oracle
Enterprise Manager Ops Center and discussed in this white paper. To establish a specific business case,
an organization should try to determine the potential savings. At the same time, it’s important to bear
in mind the following key benefits afforded by Oracle Enterprise Manager Ops Center:
•   It allows the systems administration staff to do more with less in a smaller amount of time, so that
    they can dedicate their time to tasks that cannot be automated.
•   It improves the operational results of IT in terms of service levels in support of the core business
    activities of the enterprise.
Oracle Enterprise Manager Ops Center is designed to improve IT operational efficiency across a range
of functions—each of which is affected to different degrees by the range of Oracle Enterprise Manager
Ops Center features. These functions are detailed in the table, along with a suggested working
assumption as to the expected savings.




                                                                                                                                    24
                            Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




 BUSINIESS CASE FACTORS FOR ORACLE ENTERPRISE MANAGER OPS CENTER


 BENEFITS                    DETAILS                                                            SUGGESTED SAVINGS

                                                                                                ASSUMPTIONS


 Reduce unplanned            Achieved by maintaining consistency across OS deployments          • 20% less outages
 downtime                    and firmware currency; implementing vendor and CERT                • 50% less downtime
                             recommendations; comparing to known good configurations;
                             proactively monitoring for power, hardware faults, and
                             performance indicators; and leveraging virtualization mobility


 Reduce maintenance          Reduce the incidents in which a staff member is dispatched to      50% savings in staff time
                             solve an operational issue of a server in production, QA, or
                             development


 Improve server              Reduce the time it takes to make a new server or virtual           60% reduction in elapsed
 time to service             machine operational                                                time


 Reduce system               Reduce the time administrators spend discovering datacenter        80% reduction in elapsed
 discovery effort            assets, updating administration, and tracking systems              time


 Reduce system               Reduce the time administrators spend rolling back systems to       80% reduction in
 rollback effort             previous state known to be good                                    administrator time and
                                                                                                system downtime


 Reduce patch analysis       Reduce the time administrators spend searching for patches,        75% reduction in
 and installation effort     assessing applicability to the organization’s systems, and         administrator time and
                             installing on all applicable systems                               system downtime


 Reduce Oracle Solaris       Reduce the time administrators spend on administering              75% reduction in
 Containers and Oracle VM    instances of Oracle Solaris Containers and Oracle VM Server        administrator time
 Server for SPARC            for SPARC
 administration effort




When establishing a business case for a datacenter automation framework in general and Oracle
Enterprise Manager Ops Center in particular, organizations need to consider their specific
circumstances and cost structure. As a result, any statements made here regarding concrete savings that
can be achieved by deploying Oracle Enterprise Manager Ops Center are rough generalizations and
could differ significantly from the actual savings realized by any specific company. The information is
simply a recommended approach to building a business case, and should be treated as such.




                                                                                                                            25
                                   Oracle White Paper—Oracle Enterprise Manager Ops Center: Changing the Economics of Datacenter Operations




Conclusion
The operations management capabilities of Oracle Enterprise Manager Ops Center, coupled with
virtualization management, place Oracle Enterprise Manager Ops Center directly within the optimal
cost savings part of the diagram in Figure 18—where the four primary elements of datacenter
operations intersect. Independently, each of the four feature sets positively impacts datacenter
operations, thus lowering TCO. When these feature sets exist in the same framework and allow the
sharing of data and common workflow, datacenter operations can achieve optimal TCO.




Figure 18. Oracle Enterprise Manager Ops Center’s four core benefits help operational standards drive efficiency.


As the cost of delivering IT grows, it becomes increasingly important to drive down operational costs
in the datacenter through automation focused on reducing TCO. Organizations can make a range of
assumptions as to the cost savings possible in their datacenter operations per the parameters
mentioned in this white paper. Whatever these assumptions are, once they’re multiplied by the actual
current operational costs, the business case for each specific organization using a solution like Oracle
Enterprise Manager Ops Center will emerge.




                                                                                                                                  26
Oracle Enterprise Manager Ops Center:   Copyright © 2009, 2010, Oracle and/or its affiliates. All rights reserved.
Changing the Economics of DataCenter    This document is provided for information purposes only and the contents hereof are subject to change without notice. This
Operations                              document is not warranted to be error-free, nor subject to any other warranties or conditions, whether expressed orally or implied in
February 2010                           law, including implied warranties and conditions of merchantability or fitness for a particular purpose. We specifically disclaim any
                                        liability with respect to this document and no contractual obligations are formed either directly or indirectly by this document. This
Oracle Corporation                      document may not be reproduced or transmitted in any form or by any means, electronic or mechanical, for any purpose, without our
World Headquarters                      prior written permission.
500 Oracle Parkway
Redwood Shores, CA 94065                Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of their respective
U.S.A.                                  owners.


Worldwide Inquiries:                    AMD, Opteron, the AMD logo, and the AMD Opteron logo are trademarks or registered trademarks of Advanced Micro Devices. Intel
Phone: +1.650.506.7000                  and Intel Xeon are trademarks or registered trademarks of Intel Corporation. All SPARC trademarks are used under license and are
Fax: +1.650.506.7200                    trademarks or registered trademarks of SPARC International, Inc. UNIX is a registered trademark licensed through X/Open
oracle.com                              Company, Ltd. 0110

								
To top