White PaPer template-based Provisioning
®
Managing it infrastructure with template-based Provisioning
®
The Problem: Dynamic clouds on static infrastructure
Enterprises and service providers are increasingly moving towards public or private cloud services with the goal of lowering costs and improving service by tapping into resources “just-in-time.” But in the transition, they’re discovering that the underlying IT infrastructure wasn’t designed for this new dynamic reality. Instead, it consists of static silos of servers, networking, and storage that require weeks or months of manual configuration by specialized IT staff. Control is also becoming a significant issue, as operators struggle to keep track of data center environments that consist of a complex web of virtual and physical servers and network connections. This creates a challenge when IT infrastructure configuration needs to be audited, replicated, backed up, or restored. Given the long lead times and management challenges, IT departments are being forced to set up infrastructure “just-in-case” so they can respond quickly to customers. This reduces utilization and drives up costs related to hardware acquisition, software acquisition, power, and space consumption. However, the biggest impact is on administrative costs, which is the most significant and fastest- growing cost within the data center. According to analyst firm IDC, administrative costs for servers alone are expected to reach 63% of total IT spend by 2011 (IDC 2008). Virtualization only solves part of the problem, improving flexibility and mobility but only within available physical resources. The underlying infrastructure still has to be pre-configured and cabled using separate interfaces. Virtualization also increases management complexity. And more importantly, today close to 80% of applications are not designed to run in virtualized environments at all. The ideal approach is one in which all underlying physical IT data center resources and their interdependencies can be manipulated in software-defined templates. The entire application or service infrastructure can be provisioned or repurposed on-demand or automatically in response to real-time business needs. This results in less overbuilding, significant operational savings and improved serviceability. This white paper will discuss how unified computing can deliver these capabilities to address the challenges of provisioning and create a more responsive data center.
www.liquidcomputing.com
Pg1
White PaPer : template-based Provisioning
®
Virtualizing the entire data center
Template-based provisioning extends the concept of virtualization all the way down to the underlying physical infrastructure. Application or service environments are encapsulated in data files that can be reliably developed, tested and activated to configure servers and associated networking and storage components on-demand and in a single workflow. How does this work? Consider that in today’s most modern data centers, virtual machines, operating systems, and applications are stored as “bits on a disk” somewhere in the network. They are loaded on a physical machine on-demand, by an administrator or as-directed by the management layer. This concept can be extended to virtualize all data center IT infrastructure, including physical servers, networks, I/O ports, storage and other logical functions, such as VLANs and firewalls. In fact, infrastructure can be completely reconstituted in software as XML files. In doing so, physical hardware now becomes “bits on a disk” that can be configured and controlled completely in software. As shown in figure 1, physical hardware components are abstracted to create logical computing, networking, storage, and I/O resources. These resources can be used as building blocks for the creation of application or service-specific logical infrastructures, which are stored as templates. Template files can then be quickly created, copied, managed and deployed on the physical data center infrastructure.
Figure 1: How template-based provisioning works
For example, a “master” template could be set up to define an environment that includes a set of virtual servers, the associated network connections (i.e. one network for VMotion, one for management, one for booting and one for VM network access), and storage components. Using the master template as a starting point, an administrator could enter new customer-specific information, add or subtract servers, change security and network parameters, and save the result. At that point, the new infrastructure and associated services can be activated. This can be replicated for other environments or applications, such as Oracle RAC or Windows Server.
www.liquidcomputing.com
Pg2
White PaPer : template-based Provisioning
®
As shown in figure 2, by simplifying the workflow, IT organizations can achieve up to a 10:1 reduction in infrastructure provisioning time.
Figure 2: Provisioning time, before and after
Provisioning time without unified computing: The graph below shows the provisioning time for just one server, and doesn’t include the typically long wait times between steps, where coordination between teams is required.
Provisioning time with unified computing: Unified computing can provide a 10:1 reduction in provisioning times by simplifying processes and eliminating time-consuming manual steps.
www.liquidcomputing.com
Pg3
White PaPer : template-based Provisioning
®
Applying template-based provisioning – when and why?
With template-based provisioning, IT organizations can reduce the time and cost of managing applications using software to control and provision all resources in a single workflow. Infrastructure configuration is encapsulated in templates that can be developed, tested, and moved into production. These templates can also be backed up and restored, providing check-point and roll-back capabilities to revert infrastructure changes as needed. Rather than in weeks or months, provisioning is now measured in minutes. The result is a reduction in costs associated with power, cooling, floor space, licensing, and most significantly, administrative overhead. When should IT organizations consider template-based provisioning? Some of the most common applications include: • Cloud services hosting: Quickly set up or tear down infrastructure for “just-in-time” scalability based on customer demand. • Infrastructure consolidation: Provision resources as needed, allowing projects with disparate and out-of-sync computing requirements to share infrastructure, or to improve utilization of backup/spare pool resources. • Disaster recovery: Ensure lock-step configuration changes between production and backup by simply replicating a database of templates, and easily verifying recovery at the backup site. • Business continuity: Should a physical resource fail, services can be quickly restarted on other physical servers, with templates applied to automatically provision associated networking resources. • Remote management: Since infrastructure is completely managed in software, distributed data centers can be effectively managed remotely by trained staff located at other sites. • Change management: At any point in time, an active snapshot of the dynamic data center or any subset can be viewed to assist in debugging, change management, or to support audit requirements.
www.liquidcomputing.com
Pg4
White PaPer : template-based Provisioning
®
A new approach: unified computing
For template-based provisioning to provide these benefits, several key capabilities need to be in place. Specifically, the solution must provide: 1. Unified control: This enables all resources including servers, networking, and storage to be provisioned in a single workflow and continuously monitored for fault-tolerant operation. 2. Virtualization: Complete software control requires the ability to virtualize all hardware resources, including isolation of all physical attributes that bind functionality to specific hardware (e.g. MAC addresses). 3. High-performance networking: This eliminates bottlenecks, ensuring resources are available as infrastructure is repurposed to support both physical and virtual environments. 4. Diskless computing: This provides maximum flexibility and control, and eliminates the need for time-consuming installation by enabling servers to boot and run by referencing templates in storage. 5. Unified policy-driven management: With this, service activation can be driven ondemand by an administrator, or fully automated by a management system in response to events or other parameters. 6. Built-in security: Highly secure operations that enable virtualized multi-tenant operations to be safely delivered without threat, denial of service, or compromised data integrity. Unified computing can deliver these capabilities. It breaks down traditional silos of systems, people, and processes, creating a set of virtual resources that can be completely managed and controlled in software. All application infrastructure configuration information is encapsulated in templates, stored in a secure database, and c an be easily backed up and restored. Servers, networking, and storage can be set up in a single workflow, reducing provisioning time from weeks or months to minutes. By automating manual processes and shortening provisioning times, a unified computing system can reduce administrative costs by up to 80%, while providing the agility and responsiveness data centers need today to support demand for cloud services and more dynamic operations.
www.liquidcomputing.com
Pg5
White PaPer : template-based provisioning
Conclusion
By applying the template-based provisioning, IT infrastructure can be configured to deliver services as-needed, and projects with disparate and out-of-sync computing requirements can share infrastructure. As a result, operators can improve responsiveness to business demand through rapid service deployment and, at the same time, reduce overhead associated with power, cooling, floor space, licensing. More significantly, IT organizations can achieve up to a 10:1 time savings in administration, the largest and fastest growing cost in the data center today. Unified computing delivers these capabilities, and can be combined with virtualization to provide the flexibility and control required to achieve dynamic data center operations.
®
About Liquid Computing
Liquid Computing is a leader in unified computing infrastructure for the dynamic data center. The company’s core product, LiquidIQ(tm), is a complete “data center in a box” blade system that improves the serviceability and agility of managed business applications while significantly reducing operating costs by enabling complete real-time software control of compute, networking and storage resources. Founded in 2003, Liquid Computing is based in Stamford, Connecticut and Ottawa, Ontario. The company has customers throughout North America and has established partnerships with global industry leaders including Intel, Microsoft, VMware, Oracle, RedHat, NetApp, and AMD. For more information, visit www.liquidcomputing.com.
UNITED STATES 1 Dock Street Stamford, Connecticut 06902 Tel: 1-203-842-0631 Fax: 1-203-842-0632 sales@liquidcomputing.com www.liquidcomputing.com
CANADA 340 Terry Fox Drive, Suite 300 Ottawa, Ontario K2K 3A2 Tel: 1-613-592-2666 Tel: 1-877-592-2666 Fax: 1-613-592-2663
Pg6
White PaPer : template-based provisioning