Rapid Recovery and Backup for Virtual Environments
May 21, 2008
The benefits of virtual environments are beginning to be realized. Companies intent on cashing in on the advantages of virtualization technologies are eager to reduce the financial and physical footprint associated with racks of computers.
While many elements of the IT environment are relatively unchanged by virtualization, others are impacted more dramatically. For example, organizations with expanding virtual infrastructures now find backup and recovery a proximate and pressing concern. Challenged with meeting recovery time objectives (RTO) and recovery point objectives (RPO), even as backup windows and storage space shrink, organizations must be able to quickly back up their virtual environments and recover not just an entire virtual machine but individual files on that machine.
To that end, a growing number of companies are opting for technologies that offer dual restore capabilities from a single backup. With these tools, organizations can restore a single file or an entire image from just one backup pass. This more efficient and flexible approach to protecting virtual environments dramatically reduces server, network, and storage requirements for data protection while providing significant improvement in recovery time and reliability.
Better Backups, Better Recovery
With off-host backups of virtual machines now possible, the impact of backup processing on the server and hosted virtual machines is significantly reduced. This allows for more frequent backups.
Yet, traditional backup solutions either back up only at the vmdk level—that is, the entire virtual machine—or require two backup passes to be able to restore single operating system file restores as well as vmdk restores. Recognizing the benefits of being able to perform either type of restore, organizations prefer to have ultimate restore options when things go wrong.
For example, if a virtual machine is infected with a virus or inadvertently damaged due to user error, a single file restore is of little use; the entire virtual machine needs to be restored. However, if the user deletes and needs to recover a single file—the most common type of restore operation—restoring the entire virtual machine is not only excessive but also requires downtime.
At the same time, aggressive RPOs and RTOs remain a requirement for keeping mission-critical applications constantly available. Meeting those objectives requires tight integration of the backup and recovery process with the applications and databases they are protecting, whether in physical or virtual environments. It also requires granular recovery to improve recovery time, instant recovery options from online images, and complete system recoveries of operating environment, application, and data in minutes.
Consequently, new backup and recovery tools are making possible either type of restore while retaining the performance advantages of an off-host backup and a single backup pass. At the foundation of these capabilities is technology that backs up at the vmdk level and then maps, catalogs, and backs up individual files as well.
Particularly useful for larger virtual installations based on shared storage, the integration of this backup technology with the virtual server technology enables organizations to manipulate and control virtual snapshots just as they would array or software-based snapshots. Snapshots of virtual machines are created and then mounted to the backup proxy for backup. This approach almost completely removes the backup processing overhead from the primary virtual server and allows for rapid backup of virtual machines.
What’s more, when such backup capabilities are combined with deduplication, additional benefits emerge. By deduplicating backup data prior to transmission, the processor, network, and storage resources required for the backup process are reduced by one or more orders of magnitude, according to some studies. This comprehensive approach enables fast, low-impact virtual infrastructure backups with dramatically reduced backup windows and recovery times. It also makes virtual client backup feasible for lower-scale virtual deployments that do not employ SAN technologies
Integration and Automation
It is no surprise that backup technologies with tighter integration with virtual server technologies offer additional benefits to organizations. For example, the integration of a snapshot wizard with the virtual server can ease backup policy configuration. Also, the direct integration of a configuration wizard with the virtual infrastructure can help ensure that IT administrators have a straightforward and easy-to-use graphical interface from which to configure and manage their virtual machines. With such a GUI, administrators can quickly provide login credentials, define other types of virtual servers, and more.
A number of tools also provide for automatic discovery of virtual servers and machines. This capability is often offered as part of the backup policy to make it easier for administrators to select specific or all virtual machines associated with an enterprise-level virtual server.
Virtual Evolution
Virtualization not only provides redundancy for mission-critical applications and data, but it is also effective as a tool that enables IT to extend limited resources within overcrowded computing environments. Its use will likely continue, even as a growing number of organizations deploy virtualization not simply in test or development environments but in production environments as well. Indeed, many enterprises have virtual servers running both business and production applications today. Furthermore, enterprises that deploy virtualization are recognizing that it is not a one-time ROI-based project but an ongoing strategy for operational efficiency.
As the adoption of virtual technologies increases, businesses must take a critical look at the tools and technologies for backing up and restoring these virtual systems and their data. While traditional approaches to backup and recovery in the physical world do not translate well in virtual infrastructures, many of the requirements remain the same. Organizations must be able to continue to increase the efficiency and cost-effectiveness of their IT operations through the use of virtual technologies while also delivering on strict RTOs and RPOs.
Consequently, a growing number of enterprises are leveraging innovative backup and recovery technologies that deliver granular file-level and image-level recovery from a single backup operation. When used together with data deduplication and tight integration of backup technologies with virtual technologies, these tools enable fast, low-impact virtual backups that dramatically reduce the challenges of data protection while offering measurable improvements in reliability as well as recovery time.
The brilliant mathematician Alfred North Whitehead once said: “The art of progress is to preserve order amid change and to preserve change amid order”.
The nature of business today is that change is the only constant. Organizations, be they public or private entities, are faced with change as a result of reorganization, business expansion, competition, the impact of new technology, mergers and acquisitions, industry or government regulatory controls and a myriad of other factors.
The reality is that any change that affects an organization will have a flow-on effect to the IT organization. One can say that an organization’s ability to adapt to change is directly related to its IT system’s ability to adapt to those changes. There are many examples of organizations that have suffered considerable harm to their reputations and market values through IT disasters that resulted from poorly implemented systems, and upgrades that went wrong.
From the release of the first commercially available relational database system in 1979, to support for Very Large Database (VLDB) requirements in the late 1990s, to databases for grid computing environments in recent years, the last 30 years have seen many important innovations with new server architectures emerging to support mission critical systems.
In the past, customers had fewer choices in server architectures as symmetric multiprocessing (SMP) servers were almost the de-facto standard for UNIX-based applications. Today however, we witness the emergence of architectures such as blade servers, clustered servers and new operating systems such as Linux.
Back then, moving from one vendor’s SMP server to another was relatively simple as benchmarks could be conducted to ensure that the new server would deliver the required performance. Today, customers looking to migrate from a UNIX SMP architecture to a Linux architecture based on blade servers are faced with a significantly more complex task. The potential for errors is higher and this can lead to decisions that bring on disastrous results.
Change Assurance
Data centers have changed fundamentally in the way they look and operate with the introduction of grid computing. From silos of disparate resources to shared pools of servers and storage, organizations cluster low-cost commodity servers and modular storage arrays in a grid.
Databases built for grid environments have enabled organizations to improve user service levels, reduce downtime, and make more efficient use of their IT resources while still increasing the performance, scalability and security of their business applications.
Nevertheless, managing service level objectives continues to be an ongoing challenge. Users expect fast and secure access to business applications 24/7, and IT managers must deliver without increasing costs and resources.
Databases play a key role in ensuring high availability. In the next generation database, the ability to run real-time queries on a physical standby system for reporting, or the ability to perform online, rolling database upgrades by temporarily converting a physical standby system to a logical standby, or a snapshot standby to support test environments, can all help ensure rapid data recovery in the event of an IT disaster.
Application Performance Testing– Necessity, Not a Luxury
To understand the impact of Application Performance Testing to businesses, let us take a closer look at a key IT issue for organizations in relation to managing change. During the lifespan of any application system, changes are a fact of life but the complete impact of these changes has to be known before the application goes into production. Common system changes are:
• Updates to an application requiring it to be moved from a testing to a production environment
• Upgrading or patching the database or operating system
• Changes to the database schema
• Changes to storage or network
• Testing a potential new hardware platform (e.g. comparing UNIX platforms)
• Testing a potential new operating system (e.g. migrating from Windows to Linux)
In order to provide some structure to this process, a range of tools has been released to help customers better manage this process and provide some capability for customers to test application performance. Despite helping to make the testing process easier, it requires significant investments of time and effort to gain a functional understanding of the underlying application of many of such tools before the testing workloads can be generated. In a vast majority of cases, the bigger issue is that the resulting workloads are to a large degree, artificial.
Despite extensive testing and validation – both time consuming and expensive – the success rate has traditionally been low as many issues still go undetected and application performance can be affected, leading to potentially disastrous outcomes.
In order to help customers deal with application performance testing, the latest release of the industry’s leading database incorporates new features that allow customers to capture a production workload which can then be “replayed” on a test system to show how the application functions in a new environment.
The key difference in this approach is that all external client requests directed to the database can be captured — so the real workload is captured and can then be replayed on a test system. This will throw up any errors or unexpected results (i.e. a different number of rows returned by a query) using the comprehensive reporting system provided.
With this innovative feature, organizations will be better prepared to cope with change — without fear.
