Mega-, giga-, tera-, peta-bytes of data. Where will we put it? How will we access and manage it? How much will it cost? John Dodge, editor in chief of ZCast.tv and eWeek editor at large, recently sat down with EMC Chief Technology Officer Jim Rothnie to discuss these issues and more. For the complete interview, catch The Dodge Report at www.zcast.tv.
Dodge: How has storage emerged as a key primary component of an information infrastructure?
Rothnie: The total amount of data being stored each year is actually doubling—100 percent annual growth. It creates so much emphasis on this piece of the overall infrastructure that things have to be done very well there.
[Also,] there is a fundamental change taking place in the way that storage is deployed in the typical IT operation. Storage used to be a peripheral device behind individual computers. Today, storage increasingly is organized as a separate network- connected resource to the collection of servers, and that really gives it a separate life as an element of infrastructure that customers manage and support and protect in its own way. So its really a separate pillar of IT today.
Dodge: Lets talk about just cost for a minute. Ive been covering technology for 20 years, and I remember it was a big day when the cost for a quarter megabyte of storage came down to $250,000. What is it today? I mean, its pennies per megabyte.
Rothnie: Yes, it is. For the full systems level for enterprise storage in 2001, its around 25 cents per megabyte.
Dodge: How do customers deal with these gigabytes upon terabytes upon petabytes of data?
Rothnie: The first thing is to treat it as a consolidated unit.
Dodge: Does that mean one physical location?
Rothnie: It actually doesnt need to be one physical location; its more of a logical notion of one place. By one place, what I really mean is on a single network ... that lets any of the servers reach any of the storage resources that are on the network. That gives you the capability of using an overall scheme for managing it, for protecting it, [and] if you want to replicate it at a remote location, for sharing it among all these different applications.
Dodge: So when large companies—a Procter & Gamble, a General Motors, a Citicorp—consider how much data to keep out in front and how much to archive and push in the background, is there a rule of thumb that they operate by?
Rothnie: Its actually a situation thats changing pretty radically because the value of keeping information online is increasing all the time. And the decline in the price of storage—and its something thats continuing today at 30 percent or so annually—makes it more and more feasible to keep lots and lots of stuff online.
If you think about it, how long would it take you to clear 1,000 messages out of your mailbox? Hours. And if each of those messages was 1,000 bytes, which is the typical size for a message, thats 1MB. Thats 25 cents worth of storage that youve been cleaning out. So, more and more, people are saying its not expensive to keep it online. So thats what [theyre] going to do.
Dodge: Can you quantify what the typical large enterprise needs in terms of storage—what they need now and what theyll need in a couple of years from now?
Rothnie: I can give some feeling for it. Obviously, it varies all over the place. We have customers that for the first time in 2001 have crossed the petabyte boundary, and that, as you know, is 1,000 terabytes, and a terabyte is 1,000 gigabytes. Those are huge numbers.
A more typical Global 2000 company today has around 80 to 100 terabytes in its online operations. And the typical company, the average company in the world today, is growing its storage capacity 100 percent annually. So it is accumulating at an enormous rate.
Dodge: You do the math. You see 10, 20 years out, how do you manage five, 10, and whats beyond the petabyte? How are you going to manage all this?
Rothnie: Management is really the key challenge in the storage industry today. The single largest part of [EMCs] R&D investment goes into tools for managing this resource. Everything else seems to double every year except for our brains. Maximizing the ability of people to do that management task is our key challenge.
Dodge: Obviously, a lot of your business has been driven by data centers going up from application service providers, Internet service providers, whatever kind of service provider, and weve seen some stories where the utilities cant provide the power, particularly in California. Have you guys thought much about that in terms of making storage more efficient? Because thats really where a lot of the power is eaten up.
Rothnie: The watts-per-byte part of the equation is improving very rapidly. Were able to approximately double that figure, actually bytes per watt. We can double that figure approximately every nine to 10 months.
Dodge: Whats going up faster? The amount of storage being used or the watts youre saving?
Rothnie: The reduction in power is going a little faster than the total amount of storage. Thats one factor. The other thing, of course, is we need to be able to deliver the power that the modern economy needs. I dont think efficiencies are going to be the entire answer.
Dodge: Disk technology or storage technology? Were using the same technology, just compressing it and putting more on old Winchester technology. Is there anything coming down the pike thats going to replace that?
Rothnie: There may be, but I think as we can see ahead over the next five years or so, the basic magnetic methods used in disk drives today do have a lot of headroom in front of them. The Winchester form factors [are] shrinking, but the basic concept of the Winchester drives with magnetic recording methods will still predominate in the industry over the next five years. Beyond that, there may be some optical methods that may be desired in order to continue the rapid increase in storage.
Dodge: But they wont take over for magnetic?
Rothnie: They will not for the next five years. In fact, a few years ago, optical disk drives were considered much more promising. Theyve really pretty much disappeared from the computing industry. Obviously, they are used in CDs and DVDs. But for rapid rewrite access, magnetic media is still by far the predominant method.