ALKHOBAR, 25 February 2003 — If you take hardware from many vendors, install on it software from any vendor, and make it all work together in a network, what do you have?
A heterogeneous IT environment, that’s what. There are a lot of heterogeneous IT environments out there today, due to a rapid change in the patterns of corporate IT spending. Businesses want to capitalize on previous IT spending. Forget the days of writing off equipment that’s just a few years old.
Internal policies at many firms prohibit the early retirement of hardware. Companies want to maximize the return on investment from heterogeneous systems already installed and reduce the total cost of ownership for the information technology they use now and in the future.
The amount of data we store electronically, according to some claims, is doubling every year. But companies cannot afford to continue handling data in ways that were traditional just a few years ago.
Enterprises are looking for data management solutions, which help meet the needs for scalability, efficiency and manageability. Many companies have turned to Storage Area Networks (SANs) to help pool heterogeneous storage devices and manage them centrally, more effectively and more efficiently.
In a heterogeneous environment, a SAN allows different kinds of servers running Windows NT, UNIX, and Linux to share different kinds of storage. With this shared capacity, organizations can acquire, deploy, protect and use storage devices more cost-effectively.
On a SAN, any data, at any network location is accessible. SANs allow you to virtually centralize storage. With a SAN you can separate actual physical storage hardware from the virtual representation that users see. This technique, called virtualization, provides flexibility in distributing and reconfiguring storage resources. Virtualization fulfills a role for storage similar to that which an operating system does for a computer system — namely making programming and operation easier by automating resource management housekeeping. Virtualization hides the complexity of the underlying management from the server that is simply trying to connect to that network and access storage.
“Let’s talk about virtualization for a moment,” said Scott Gready, director, Virtualization Business Segment, Network Storage Solutions, Hewlett-Packard. “One area of confusion is that people want to think of virtualization as a product. When really it is not a product in and of itself. You will always see HP talk about our virtualization technology.
Unfortunately I think this can be confusing because we do not have a product that is purely virtualization. The best way that I have found to explain it is to say that virtualization is a technique we use in storage, and it is a technique that has been used for many years, but we haven’t always called it virtualization.”
“In fact, in the past it really wasn’t called anything,” Gready continued. “It was simply several advancements in technology that I can point back to that really are virtualization, but they weren’t called that. One example is that, a long time ago, you had to tell your computer how many heads and sectors and cylinders a disk drive had so that it would know the internal geometry of that disk drive. No longer do we have to do that. Now we have something called logical block addressing (LBA). With LBA, a disk drive looks like block one through one million, and the computer doesn’t have to know all the details. We hide the underlying complexity of what’s inside the disk drive from the computer. Any time you insert a layer that hides the complexity underneath, that is the essence of what virtualization is.”
Another example of how virtualization has changed storage management can be found in regards to redundant arrays of independent disks (RAID). According to Gready, servers and computers used to have many, many disk drives, and more and more disk drives were being added all the time. Connecting and configuring all these many disk drives individually became more and more complex. In the late 1980s RAID controllers appeared. These array controllers not only striped data across multiple disk drives and provided more performance and more fault tolerance across individual disk drives, but array controllers hid the underlying complexity of how disk drives were connected. All the servers had to see was the array controller. They had their storage assigned by the array controller without requiring knowledge of what was underneath. This was virtualization — a layer of abstraction between the consumer of storage and the physical devices.
HP likes to talk about the storage utility or delivering storage to the consumer of storage as a utility service. This is much like building a new house and just hooking up to the electric and water utility.
You don’t have to know what kind of power generators the electric company is using or how many of them are up and how many of them are down. Just simply plug in to the electricity network and get a service.
HP believes that storage specifically and computing in general should be delivered in this utility model, and that’s where virtualization comes in.
“The utility company has virtualized their power grid and their generating capacity for me the consumer,” said Gready. “Network based virtualization does much the same. That’s sort of an abstract academic description of what we’re doing. Why is this of benefit to the customer? Customers are benefiting from this technique because due to mergers, acquisitions and consolidation, customers wind up with a heterogeneous, varied set of storage resources. They have array controllers from multiple vendors. They have a big legacy, really a hodgepodge of equipment, and they need help managing that heterogeneous set of things especially when it comes to applying storage services on top of that infrastructure.”
As an example, Gready described a business that has two sites with different types of array controllers at both sites. The business wants to replicate data from one site to another. “This is a situation where it is very helpful to take those array controllers and treat them as a pool of storage at each site and employ a network based device to do the replication between the two sites,” he explained.
“HP sells array controllers that support replication and we will continue to do so because many of our customers very much like the model of buying the same array controller at two different sites and connecting them and replicating data.”
This is an easy way to replicate data. However data replication depending on the technology in arrays alone requires that there are identical twins on opposite sides of the wire. “That does not work when you move into very heterogeneous environments,” Gready said. “So we sell a product today called the continuous access storage appliance (CASA). For those environments where array to array replication won’t work because the sites are just too different, CASA is the answer. You put a CASA appliance on the network at each site and it performs the replication function at the network level. The technique that CASA uses to pool those heterogeneous array controllers is virtualization and we call it virtualization, but again it’s just a technique to deliver an advantage to the customer at that layer. We’ve added it up and found that we are able to support 90 percent of the installed storage in the world behind the CASA appliance.”
For years, Compaq and HP were both developing their own virtualization strategies. When HP merged with Compaq, a new business unit, Network Storage Solutions, was formed from the two companies former storage businesses to handle customers’ storage requirements. According to Gready, pre-merger HP had been focused on enterprise storage services across a heterogeneous pool of storage controllers with a limited focus on scale. On the other hand, pre-merger Compaq was concentrating less on heterogeneity and enterprise services, and more on the very high scalability of storage environments.
‘The analogy we use,” Gready said with a smile, “is that each company was painting the same wall with Compaq starting on the left and HP on the right and now since the merger we’re getting complete coverage.”
If you are wondering if utilizing virtualization technology is right for your company consider the potential return on IT investment. According to Gartner Research, without storage virtualization one person is able to manage up to 300GB of data. With virtualization, the amount of data one person can manage jumps to 2TB. Therefore, without virtualization, the long-term cost of managing and maintaining traditional data storage infrastructure can radically outweigh the cost of acquiring and installing storage. And IDC claims that organizations can achieve many of the benefits of uniform centralized storage through a virtual approach, which masks the complexity of the actual storage by making it appear to be a single accessible pool of information. This permits organizations to leverage existing storage and network infrastructure to gain maximum return on investment (ROI).
Comments to: [email protected]
