What is storage virtualization?
Storage virtualization is not synonymous with cloud storage. It is an approach to data storage that abstracts the name of the data from the actual data itself. The naming information, or metadata, is searchable by applications and other IT resources, but the physical data behind the logical metadata can exist in any addressable resource. That means that suddenly it is not important where the physical data is, so long as the metadata points accurately to its location.
Storage virtualization benefits: An end to overprovisioning?
Abstracting data using storage virtualization enables dynamic management of data resources. It becomes possible to aggregate data, move data and even place it in remote storage facilities (cloud storage). Then the data can be utilized as though it were a local resource.
Among storage virtualization’s many virtues is that it enables IT to avoid the dreaded stair-step-of-doom demand curve, where adding additional storage capacity significantly overprovisions enterprise storage resources. This is because storage devices no longer need be mapped to a particular application where significant resources remain unused because the application may not need them immediately. Instead, storage devices can be loaded to near 100% levels, more efficiently utilizing storage resources.
Additionally, because storage virtualization allows for the transparent movement of data, the technology improves data integrity by simplifying off-site backup and archival. Since a metadata directory doesn’t care about the physical location of data, it is possible to index not only current data, but archival data, making both equally accessible.
Storage virtualization complications
Storage virtualization, though, does come with some baggage. It is a complex technology, requiring the maintenance of a metadata directory, significant tools to manage the mapping of metadata labels to physical data storage, as well as strategies for securing metadata. If the metadata directory is corrupted, the entire data storage facility can be rendered unusable.
Additionally, storage virtualization can introduce latency and security issues, especially when it is being applied to network- attached storage resources that may be accessed through the Internet cloud. If the network connection dies, so can access to mission-critical data. What's more, as data traverses the network, it can become subject to intercept or corruption. Encryption becomes an essential aspect of storage virtualization when it's used for an external cloud environment.
Steps to storage virtualization implementation
Implementation of storage virtualization requires that the enterprise first develop a plan for storage that includes acquiring servers for virtualization and metadata directories, building out a SAN and more.
Essential questions to ask about storage virtualization implementation include the following:
- What is the maximum amount of storage that is desired for online access?
- What are the backup and archival requirements?
- Which applications will be accessing the virtualized data store?
- How secure must the storage facility be?
- How will storage need to evolve over time to meet business needs?
The answers to these questions determine how complex and extensive the storage virtualization process will be. For example, if security requirements trump all others, then the enterprise is likely to deploy a storage area network (SAN) that is highly interconnected but does not include network access to the outside world.
In any case, enterprises seeking to deploy a virtualized storage environment will need to seek servers that can run virtualization applications and manage a metadata directory.
Once these servers have been configured, existing storage is then mapped to the directory. Additionally, a SAN needs to be deployed to manage the dynamic transfer of data across devices. This typically involves interfacing storage devices with high-speed network interfaces that interconnect through fiber or high speed Ethernet. Whether this requires substantial investments will largely depend on the kind of storage environment and the sorts of latency that can be accepted by the applications that access the storage facility.
Additional configuration requirements include determining how often data will require backup, whether data needs to mirrored across different storage facilities, and to what degree data must be encrypted. As part of this process, the IT shop will also need to determine whether cloud storage services have to be part of the virtualized environment and to what degree that will include working data or archival data.
Ultimately, it is likely that most IT operations will decide that storage virtualization is worth the complexity, since businesses increasingly demand that data be collected and maintained for protracted periods. A dynamic storage environment can accommodate vast amounts of data and enable the use of cloud services.
This was first published in September 2011