Virtualize for best results—but be ready
Here's a familiar tale to IT managers: A technology innovation promises impressive benefits. This same technology, however, also creates unwanted side effects.
Getting best business results from the technology will therefore require a new approach that can deliver the benefits while mitigating the side effects. And pursuing that new approach, in turn, may involve new solutions.
One recent example of this story: virtualization. It's no exaggeration to say virtualization has significantly transformed enterprise data centers. However, like all innovations, virtual servers imply new ramifications for IT managers to take into account.
Consider how virtualization can affect data backup. Legacy backup solutions were certainly not designed for today's world, in which many virtual servers may be running in parallel on a shared physical host. As a result, backup administrators must ask and answer a variety of new questions such as these:
All of these questions become even more significant in the case of cloud architectures. Clouds leverage virtualization in an exceptionally deep and automated way, generating new virtual servers to fulfill unpredictable demand levels for cloud services. And as a result, the number of virtual servers existing in a given cloud may be continually in flux.
Backing up all the data in all the servers in a cloud, enough to meet or exceed recovery point objectives, is thus no simple task. It may indeed require new backup solutions or strategies more sophisticated than anything the enterprise has deployed to date.
Key to a successful approach: flexibility. IT will need to understand the challenges that apply in a given case, and address them with appropriate solutions and features capable of handling them.
Fortunately, via two IBM Tivoli solutions, data backups running in even the most advanced virtualized infrastructures can be backed up in a business-savvy, resource-optimized fashion—ensuring that all data is preserved, and can be restored on demand, while simultaneously minimizing the impact on IT services. These two solutions are IBM Tivoli Storage Manager FastBack and IBM Tivoli Storage Manager for Virtual Environments and they work in fundamentally different ways to deliver different forms of value.
Virtualized infrastructures means new backup complexities to ponder
“FastBack has been optimized specifically to minimize both the resources required to execute and the impact on service levels. Because it continually executes, at any given time it is only backing up a tiny amount of data. Accomplishing this requires correspondingly low host resources.”
To understand these issues in more detail, consider how backup processes have historically been handled in traditional infrastructures based on single-application hosts. Such hosts have only one operating system; that operating system has a backup agent installed and running in it. Backup servers communicate with the agent on each host.
Such an approach is often still used following a virtualization initiative, but typically, it will not yield best results. Among other complexities, for instance, is the fact that many virtual servers may be running simultaneously on a host. That means many operating systems and many agents—not just one of each.
If they all execute backup policies at once, the effect on that host's resources—I/O, processing capacity, and memory, among others—will be significant. So, too, will be the negative effect on all the services that host supports. While backups can be, and often are, scheduled specifically for times of reduced demand, such as after business hours, the underlying problem remains.
Other complexities apply as well. Basic management, for instance, is complicated by virtualization; there are now multiple agents executing on a given host and managers must keep careful track of these, even when virtual servers migrate from one host to another. Backup policies must similarly acknowledge and handle the new, virtualized nature of the infrastructure.
Not all backup solutions are designed with these various complexities in mind. Many will lack the comprehensive feature set required to yield the highest ROI and business resilience from virtualization.
IBM Tivoli Storage Manager Fastback: Continuous data protection
One solution that is unusually well suited to virtual servers is IBM Tivoli Storage Manager FastBack. While FastBack does use a traditional agent—an "in-guest" design—it functions continuously, instead of only at times when called to do so by policies running on the backup server.
Specifically, FastBack creates block-level, incremental backups. As new data is created on the virtual server, FastBack's agent detects that process occurring and backs up the data to a designated archive. FastBack doesn't require a backup window; it simply never stops running, and data never stops being backed up.
This approach delivers a number of clear benefits. For example, the fact that data is continually backed up means that it is continually protected. With a traditional backup solution that only backs up data at time intervals, it might be possible to lose all the data created since the last backup—an unacceptable outcome for many organizations, and one that becomes steadily less appealing as IT services become steadily more critical. FastBack delivers peace of mind, bolstering business resilience in a broad sense by reducing to nearly zero the chances of permanent data loss.
Furthermore, FastBack's performance also helps organizations address another major concern: the potential drag on IT service performance while backups occur. FastBack has been optimized specifically to minimize both the resources required to execute and the impact on service levels. Because it continually executes, at any given time it is only backing up a tiny amount of data. Accomplishing this requires correspondingly low host resources.
FastBack also boasts impressive "smart" features designed to minimize the work it must carry out. One such is deduplication. If FastBack determines that a given block of data to be backed up is actually an exact copy of data it already has backed up, it skips the backup process. Instead, it creates a pointer in the archive. When FastBack is called to restore this data from the archive, it will encounter the pointer and simply restore the original version of the data twice (or however many times it needs to do so, to mirror the original data pool).
FastBack also includes powerful compression capabilities; these recognize and leverage patterns in data. Instead of backing up all data per se, FastBack can simply back up the pattern. As a rough example, instead of backing up a thousand copies of the letter A, FastBack might only back up one letter A, and the number of times it occurred consecutively (a thousand). This is obviously a far faster process and one that requires far less storage.
FastBack's capabilities thus live up to its name—they do indeed back up data, and fast. For servers running Windows or Linux, it's a superior archiving solution that can help you get higher business value from almost any virtualization environment you may have chosen.
IBM Tivoli Storage Manager for Virtual Environments: Highest backup performance
For organizations that are using VMware (link resides outside of ibm.com) environments in particular, however, IBM has (as of early 2011) created an even more advanced option: IBM Tivoli Storage Manager for Virtual Environments (TSM for VE).
TSM for VE leverages the vStorage application programming interface (API) for Data Protection. These APIs are specifically designed to enable best-in-class backups of data in VMware environments with the least possible impact on server performance.
How do the vStorage APIs work? Essentially, they isolate the task of backing up data away from the host and virtual machines that actually have the data. Instead, a separate vStorage server is used. Given a data backup solution that uses the vStorage APIs, this separate server becomes a sort of liaison, executing backup policies largely via its own resources, instead of stealing resources from the host machine.
This means, in turn, that the host can continue to deliver best performance of all the services executing in all the virtual servers it supports—a huge win from a business standpoint. It also means that backup processes execute much more quickly. So do restoration processes, if they're required.
TSM for VE also significantly simplifies management. Because there is only one agent installed, there is only one to manage, track, and keep up to date. This solution can also be managed using the Tivoli Storage Manager console—basically, a one-stop-shop for everyday functions. Also helpful from a management standpoint: the backup target volume need not be a physical host, but can be another virtual machine. This spares IT the need to allocate and maintain dedicated hardware purely for that purpose.
Discovery of new virtual servers is another compelling feature of the solution. Because TSM for VE can detect and support new virtual servers in real time, it's a perfect match for even the most advanced virtualized infrastructures, such as clouds, in which virtual servers are constantly being generated.
Finally, TSM for VE also delivers a broad range of data recovery options; data can be recovered at the file level, the volume level, or even the level of the entire virtual machine. This extra granularity gives IT extra power to match any given backup process to any given specific business context or demand level—increasing flexibility in much the same way as virtualization itself.