com
Technical White Paper
DATA CENTER AUTOMATION
4 . . . . . Components
5 . . . . . Compute Servers
7 . . . . . Storage Servers
7 . . . . . Management Servers
10 . . . . . Summary
p. 1
Easing Pain in the Data Center
Data centers are being squeezed by a variety source-based platform that offers value through
of internal and external pressures such as sophisticated integration of otherwise isolated
power consumption, HVAC requirements, new components. This solution addresses the
“Data Center Managers
servers, human errors, patching, asset track- workloads shown in Figure 1. Consider the
are on the hot seat lately.
ing and more. In fact, the average data center evolution of computing from mainframe to
They not only have to
consumes enough power in a month to power mini to client/server. Now modularize, stan-
cram more servers per
1,000 homes! On top of all this, you have to dardize, commoditize and virtualize. Next,
square inch than they
keep up with dynamically changing business add integrated intelligence and you have
ever thought they’d need,
requirements. You need a solution that will a modern computer, comprising virtualized
they also have to figure
allow you to align IT to your business, control computing and storage that is controlled by
out how to do it without
costs and minimize risks. Data center man- a distributed operating system realized by
sending the electricity
agers are looking for a variety of ways to grid-inspired resource management software.
bill through the roof.”
address these dilemmas. One of the key ways This new solution enables mainframe-class
is server consolidation using virtualization. capabilities for commodity scale-out data
eWeek
“The Greening of the Data Center” center architectures. All workloads are
by Kevin Fogarty Data Center Automation supported by a common modular Linux*
August 2006 Solutions from Novell® foundation: SUSE Linux Enterprise,
®
Figure 1. SUSE Linux Enterprise is the foundation for commercial high-performance cluster computing, data center,
enterprise workgroup and desktop.
p. 2
Automation for the New Data Center www.novell.com
p. 3
“For virtualization to
truly work in real-world
applications, users must
also focus strongly on
automation, the policy-
based administrative
tools used to deploy
virtualized instances
and manage them.”
John Enck
Gartner
Figure 2. A platform for executing SOA Applications can be represented using the following formula:
Virtual Storage + Virtual Machines + Resource Management + Identity Management
Virtualized Storage. Active control of SAN devices to accomplish data protection or storage provisioning goals
on behalf of host resident services. A cluster file system stores virtual machine images and other persistent data
in a highly available way.
Virtual Machines. Execution containers for NetWare®, Linux, .Net and Java*, on Linux (for example, a physical
server that has been imaged and is now running on a Linux cluster as a managed virtual machine).
Resource Management. Management software that images blades, orchestrates remote installs and package
updates across a farm of (blade) servers; includes Grid resource scheduling algorithms, for example, for migrating
workloads and even large datasets from one blade server to another or one data center SAN to another for disaster
recovery; policy-driven data protection.
Identity Management. The owner of services and data, the entity that creates purpose and ownership for the
above activities.
p. 4
Automation for the New Data Center www.novell.com
p. 5
SUSE Linux Enterprise Server 10—Virtuously Virtual
SUSE Linux Enterprise 10 offers virtualiza- integrated high-availability clustering with
tion capabilities like no other OS. It can support for LinuxHA v2 and the Oracle*
provision, deprovision, install, monitor and Cluster File System (OCFS). Inn a clus-
manage multiple guest operating systems. tered environment, SUSE Linux Enterprise
It provides the out-of-the-box ability to cre- Server 10 (plus the Xen hypervisor, YaST2,
ate Xen* virtual machines running modified, CIM-based monitoring tools and other
highly tuned, paravirtualized guest oper- built-in, standards-based management
ating systems for optimal performance. solutions) is the foundation for allowing
What’s more, with CPU hardware assist resources to be pooled, allocated and
plus Xen functionality, SUSE Linux Enterprise utilized like never before. In effect, VM
Server can play host to several guest OSes management becomes synonymous with
operating on a single server at speeds that workload management. The data center
are generally faster than those obtained becomes an asset manager that is aware
when the OSes were operating solo in of all physical and virtual servers in the
a 1:1 configuration. environment and their characteristics.
This information is acted upon in real time
Data center managers can maintain a to allocate resources as appropriately and
centralized store of virtual machines (VMs) efficiently as possible.
and deploy them over the network by iden-
tifying a physical computer at deployment Data center managers can configure a
time, copying the VM image, and making clustered environment based on standard
it available to run on that particular physi- servers and storage and running SUSE
cal server. The VM can specify a set of Linux Enterprise 10 that is free of single
constraints such as 32-bit or 64-bit server points of failure (see Figure 4 on the follow-
or SAN connectivity. The VM might contain ing page).
an unmodified Windows* OS and declare
a deployment constraint for hardware that Shared-disk clusters enable high availability
must support virtualization technology. for VM hosting, as all VM OS image files
The data center can maintain a veritable reside in a central location and access is
catalog of available VMs in an offline possible by each server. VMs can be failed
repository and send OS images upon the over if the physical server on which they’re
request of an individual, a workgroup or— running fails. With future support for live
soon—when a business policy, a service VM state migration—a real-time transfer of
level agreement or a server failure neces- live OS state from one physical server to
sitates deploying a new image. another—there is virtually no server down-
time; applications continue to operate
In addition to virtualization capabilities, uninterrupted, and end users are unaware
SUSE Linux Enterprise Server 10 provides that a migration even took place.
p. 6
Automation for the New Data Center www.novell.com
_______________________________
† Note: According to
Wikipedia.com, SMI-S,
or the Storage Management
Initiative—Specification, is a
storage standard developed
and maintained by the Storage
Networking Industry Associ-
ation (SNIA) and is a model,
or guide to building systems
using modules that plug
together. SMI-S-compliant
storage modules interoperate
in a system and function in
consistent, predictable ways,
regardless of which vendor
built them, provided that
the modules use Common
Information Model (CIM)
language and adhere to sets
of specifications called CIM
schema. The main objective
of SMI-S is to enable broad
interoperability among
heterogeneous storage
vendor systems.
Figure 4. You can create a SUSE Linux Enterprise 10 cluster for hosting Xen VMs. Using the YaST2 tool, VMs take
on cluster-wide capabilities. VM images and creation files are placed where they are accessible to other servers in
a clustered configuration.
p. 7
This namespace, plus any hierarchical struc- Orchestrator
ture created by the data center administrator
The Orchestrator is the brain behind the
(e.g., organizing servers into groups), will be
data center automation system; it interacts
federated with the Universal Model Facility
with the configuration and storage resource
(UMF; also see Universal Model Facility on management servers to manage physical
the following page) to support CIM-based compute and storage resources and the rela-
server health monitoring. The management tionships between them. The Orchestrator
server can track the registration of virtual also manages virtual resources. It’s respon-
machines, and considers them to be man- sible for the entire lifecycle of individual virtual
aged assets just as physical servers are. machines comprising control information,
Virtual machines, once created, will also OS image and storage resources from initial
appear in the managed server namespace. creation to deployment and monitored execu-
Once the systems are in a managed state, tion, to final destruction. Physical constraints,
you need a way to orchestrate them to align dependencies, live performance trends and
to business needs. other real-time execution states monitored by
p. 8
Automation for the New Data Center www.novell.com
What is Grid?
Electricity supply utilities depend on a A good way to think about Grid is as the
high-voltage grid. Therefore, Grid software next generation of resource management
is nominally considered the foundation for software. A better way might be to con-
Utility computing. A Grid runs distributed sider Grid the next distributed operating
resource management software capable system. Grid Computing manages appli-
of allocating capacity from virtualized cations comprising a collection of Web
computers and storage devices. Instead services consuming virtualized storage
of statically installing application software and computational optimal use of physical
onto a computer, grid software dynamically resources relative to consumers. Therefore,
binds services and data to computers the classic definition of an OS hasn’t
at execution time. This makes individual changed; it too has been virtualized
computers anonymous relative to process- across multiple computers.
ing. Grid “jobs” that comprise program
logic and dependent data are scheduled
to be combined and enable processing.
the UMF are considered by the Orchestrator a summary view of vital-signs metrics to the
when scheduling virtual machines to compute Orchestrator, the UMF could be considered the
servers for execution. nervous system wired to the Orchestrator’s
brain. A monitored variable may go above
Storage Resource Manager and dip below thresholds, but isn’t consid-
ered noteworthy until it has stayed above a
The storage resource manager is responsible
threshold for a certain period of time.
for managing network storage assets. It main-
tains a configuration inventory of storage
devices. Compute servers will dynamically Image Creation
access SAN storage on behalf of virtual An image-creation server is a special kind
machines that are scheduled to run on them. of compute server dedicated to the creation
The storage resource manager also supports and installation of virtual machines. In large
provisioning of network storage when creating environments that depend on frequent virtual
a new virtual machine. machine creation, you might have multiple
image-creation servers. In other scenarios,
Universal Model Facility the Orchestration server may decide to define
The Universal Model Facility (UMF) is another and install a virtual machine “in-place,” effec-
new component responsible for aggregating tively incubating the virtual machine on the
and associating management models and compute server that will eventually also exe-
monitoring data from managed devices. cute it. The result of providing image-creation
Managed devices are either compute servers, services is the automated control and creation
virtual machines or SMI-S-enabled storage of a new virtual machine comprising control
servers. The UMF collects and records health information, OS image and optional external
information in the context of the relationships storage references. Infant virtual machines
that exist between managed devices. By con- are ready to execute. They actually run as a
suming status events, applying hysteresis result of Orchestrator-driven deployment to
thresholds to monitored devices and exporting an assigned compute server.
p. 9
www.novell.com
462-002042-001 | 11/06 | © 2006 Novell, Inc. All rights reserved. Novell, the Novell logo, the N logo, NetWare and SUSE are
registered trademarks of Novell, Inc. in the United States and other countries.
*Linux is a registered trademark of Linus Torvalds. All other third-party trademarks are the property of their respective owners.