Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
There has been a lot of debate over the past year about how storage virtualization should be done and who has the best technology. Some say it should be done in the switch, some prefer the array, others insist that it must be contained within an appliance, and a growing number believe it belongs in all three.
Vendors have taken their own positions. IBM came out the gates ahead of the rest with its SAN Volume Controller (SVC) appliance about two years ago. Hitachi Data Systems (HDS) followed up last year with its TagmaStore Universal Storage Platform (USP), an array-based option. And in recent months, EMC Invista has gained plenty of press for switch-based solutions. So which technology and which vendor are best? And which approach will ultimately win the storage virtualization race?
“It’s very hard to compare storage virtualization technologies, as they are mostly theories at this point,” says Rick Villars, a storage analyst for International Data Corp (IDC) of Framingham, Mass. “We will need time to see how well they deploy in the real world.”
Slowly Virtual
Despite the hype, the corporate world has been slow to adopt storage virtualization technology. According to an IDC study of 269 IT managers in companies of all sizes, only 8 percent are doing any virtualization at all. An average of 23 percent plan to implement some in the next 12 months.
If you focus on companies with 10,000 or more employees, usage rises to 19 percent, with 31 percent planning to add a virtual component within a year. In the mid-sized segment (1,000 or more employees), very few are using it, although 33 percent state a desire to harness the technology before the end of 2006.
Within these various camps, then, there are varying pressures and needs at play.
“Mid-size companies mainly want to manage data migration and reduce their administrative burdens,” says Villars. “Larger shops want virtualization for functions like data replication and volume management for provisioning.”
Differing Approaches
Traditionally, there have been three distinct camps in this field. On the appliance side there is SVC, StorAge, Network Appliance and DataCore. Within the array/fabric there are varying architectures offered by HDS, Sun, HP and Acopia. And in the switch, there are Invista, McData, Brocade, QLogic and Cisco.
McData, Brocade, Cisco and others, however, have made acquisitions or partnerships aimed at fabric-based virtualization, so it appears that the lines among the categories are already beginning to blur. And some of the other vendors mentioned above are now straddling at least two camps, if not extending beyond these rigid bounds.
Switch and array advocates, however, are on the attack, targeting the performance and flexibility of appliances and early virtualization engines.
“Initial implementations of storage virtualization relied on discrete solutions based on off-the-shelf components or port-based processing engines that provided the functionality required,” says Amar Kapadia, director of product management at Aarohi Communications. “The appliance approach is considered easy to deploy but tends to be application-specific.”
Aarohi believes the next generation of storage virtualization is embodied in intelligent SAN components such as its AV150 Intelligent Storage Processor. The company has formed a partnership with switch vendor McData to create fabric-based virtualization services.
HDS makes a similar attack on appliance and switch solutions.
“The Universal Storage Platform places virtualization in the storage controller at the edge of the storage network instead of at the host or in a switch or appliance at the core,” says James Bahn, director of software at HDS. “This is the best place for performance and security reasons.”
Meanwhile, Network Appliance is of the opinion that storage virtualization is best done on the network via an appliance.
“This provides customers with the most flexibility in array choice, doesn’t lock them in as an array-based solution like TagmaStore, and does not require all the complexity and costs of host-based virtualization solutions with client code,” says Jeff Hornung, vice president of storage networking at NetApp. “The appliance can be in-band or out of band within in the network.”
This article was first published on enterprisestorageforum.com. To read the full article, click here.
