Virtualization The Future Is Now
Shared by: LesleyVainikolo
Virtualization: The Future Is Now. Why Now Is The Time To Move From Physical Machines To Virtual Machines Introducing Virtualization Virtualization By The Numbers • 540,000 virtual machines existed V irtualization offers the ability for multiple operating systems to worldwide in 2006. reside on the same physical machine regardless of whether the OS is an older legacy system or a newer system. Linux and • According to research firm Windows can even co-exist on the same machine. Software known Gartner, there could be as many as a Hypervisor then takes a layered approach to management, as 4,000,000 virtual machines by acting as a mediator between these Operating Systems and the 2009. hardware that they reside on. The hardware is partitioned in such a manner that the virtualized solutions can run side-by-side, sharing • A 20 watt virtual machine requires resources as needed and without interfering with each other. The only 10 watts of cooling from Hypervisor manages the drivers for the operating systems, and HVAC. keeps a strict list of hardware requirements minimizing the risk of • A 200 watt physical machine not having the drivers that it needs. requires 100 watts of power to run proper HVAC equipment. Many advocates of virtualization use a Storage Area Network (SAN) to link virtualized servers together. References are made to the popular “black box” example, which states that software doesn’t Benefits Of Having Virtual Machines Can Include: necessarily need to know what’s going on within the resource it is • Reduced need for upgrades relying on, as long as the resources are available to the software as • Simpler system upgrades when necessary it expects them to be. This approach allows applications to operate • Improved security and fault tolerance within their own “containers”, or virtual machines. • Improved redundancy • More efficient disaster recovery Virtualization has enabled Hosting.com to • Consolidated workloads separate essential applications from legacy • Allowance of legacy software to run on new hardware (under hardware and redeploy on newer, more warranty) reducing risk and costs of maintenance reliable and efficient infrastructure all while • Isolation from other virtual machines. (When one virtual reducing our data center footprint, power machine fails, other machines continue running consumption, and cooling requirements. • Independence from changes in physical hardware • Virtual machines are easy to image Wayne Shaw • A virtual machine can be cloned many times Senior Engineer • Corporate green initiatives Hosting.com Virtualization: The Future Is Now Consolidation Disaster Recovery There are many reasons for an enterprise to consider virtualization; If business is impacted with even minor downtime, then prolonged primary among them being better resource utilization. Most servers downtime due to disasters can destroy a companies reputation are under utilized, with average CPU usage at 10% at any given altogether. Most companies understand this, and have a business time. If an enterprise runs multiple servers, each using only 10% of continuity plan in place. These disaster recovery plans detail how its available resources, then 90% of the resources is squandered. to preserve business assets and functionality in the event of a Using this example, the enterprise could benefit from virtualization disaster. Disaster recovery can include preventive measures that by reducing the number of servers it relies on, and instead fully can either reduce the potential loss in a disaster or detail disaster utilize the resources from the servers it has. responses that secure the assets and functionality as soon as primary operations cease. Before virtualization, data centers would build out their infrastructure well above normal requirements in preparedness for business Due the immense potential that virtualization has on business opportunities or sudden emergencies. This largely unused surplus continuity, disaster recovery is the second most important reason takes up space and still required power, as well as cooling to to consider virtualization. counteract the heat that each machine produced. Underused investments realistically don’t provide any return for a data center. By encapsulating a server’s applications, OS, and data into a small set of files, virtual machines are positioned as hardware agnostic Consider that as present equipment ages, or warranties run out, and extremely portable. This enables an accelerated recovery new equipment must be introduced to take its place. New business period. Comparatively, data centers that have not consolidated demands often require more memory, or more space. A transition through virtualization have more data spread over more hardware, to newer equipment must happen without causing interruption to and do not have the flexibility of shared resources to expedite a ongoing business needs and with the least amount of downtime full backup, meaning a much more cumbersome, time consuming, possible. This can be a big job, considering it’s likely that the and detail intensive recovery period. company’s day-to-day operations are hosted on these systems. A virtualized solution makes it possible to migrate legacy hardware out of the datacenter, and new equipment in without suffering significant downtime. Since all datacenters have to contend with managing aging servers, upgrading to newer, more efficient solutions is a relevant concern for anyone with a web presence. By leveraging virtualization a company can: • Reduce the amount of space being used • Reduce power consumption Our Hewlett-Packard servers come with warranties and • Reduce the financial investment on hardware support that often feature three-year coverage on parts, • Reduce the load on cooling mechanisms by reducing overall labor, even onsite service. They offer next business heat production day limited global warranty, pre-failure notifications on processors, memory and SAS hard drives. IBM only offers The greenest server a fraction of these services. in the world is the one that doesn’t exist. Hewlett-Packard has the fastest growth on blade servers Darren King in the market over the past 5 years, shipping a million President & CEO servers more than IBM, and 6 times as many as Sun in 2006Hosting.com alone. Virtualization: The Future Is Now Case Study As a case study, Hosting.com consolidated 60 legacy servers Once the best virtual solution has been engineered, the conversion utilizing VMware’s ESX Server virtualization software. We can begin. A conversion from a physical machine to a virtual focused on consolidating such assets as web servers, servers machine can take anywhere from 30 minutes to 12 hours. with miscellaneous applications, and statistics servers. There are two types of conversions, a “live” conversion, or an “off • We consolidated 60 servers down to 6 physical servers --a 10 line” conversion. A “live” conversion is done while the server is to 1 ratio-- and reduced the physical server footprint by 75%. up and running. Depending on the scale of the conversion, this can take can demand an estimated downtime anywhere from 30 • We reduced our Recovery Time Objective (RTO) for a seconds to 20 minutes. An “offline” conversion is done by bring the physical machine from 24 hours to 30 minutes. server down entirely. This type of conversion is more appropriate for busy servers, and therefore, downtime varies. • We reduce the data center footprint for the consolidated equipment by 70%. Why Use Virtualization? • We realized annualized costs savings of over $100,000 due The reasons to consider virtualization continue to grow at an to the smaller data center footprint, lower power costs, and exponential rate. With the Hypervisor in place, virtual machines reduced personnel expense, which provided a tremendous become “hardware agnostic”; allowing them to run on almost all return on investment (ROI). traditional x86 servers. We did not apply virtualization to any “high transactional” servers, Using virtualization, benefits include the flexibility that one machine such as database servers (SQL), email servers (Exchange), or can run many virtual machines, resources are easier to consolidate, any I/O intensive servers. These servers are not necessarily good patches are easier to apply, changes and upgrades are easier to candidates for consolidation since up to 15% of server resources implement. Furthermore, virtual machines are easier to restore, go toward the actual process of virtualization. If a server is already streamlining disaster recovery. Simply put, virtualized machines using a higher percentage of resources, then performance could allow provisioning at a much faster rate. Consolidation through actually be sacrificed by virtualization. Servers with high resource virtualization offers effective solutions to disaster recovery, while utilization rates must be analyzed on a case by case basis, actually improving the overall available footprint in a datacenter, depending on the expectations of that server. and lowering costs. This approach enables a datacenter to position itself for growth. Ask yourself: Which would be more manageable: The key step in creating the best virtualized solution is a thorough 60 physical machines, or 60 virtual machines? analysis of the current consumption of resources. Months before we consolidated our servers using virtualization, we took detailed Simply put, there is a better way to do things. Hosting.com is performance metrics on the servers we considered candidates to already enjoying tremendous success with virtualization in our own become “virtual machines”. This is even more important in the infrastructure, and we are eager to help you take full advantage of case of resource intensive servers. Is this server a good candidate all of the great benefits that enterprise virtualization can provide for virtualization? What amount of virtualized resources would be you. appropriate for this solution? Hosting.com’s Professional Services team can be engaged to provide reporting of performance metrics, graphs, reports as well as metrics that detail historical data (before the consolidation), and current data (after the consolidation). Author Scott Brining is the Technical Writer for Hosting.com, providing over 7 years of Technical Writing experience to the company. In 2006 and 2007, Scott oversaw the SAS 70 Type II certification for Hosting.com.