How is it that we can do this now, where we couldn't in the past?
Well, technology wasn't there to do this in the past. I mean, you didn't have this class of ATA storage. It really got introduced into the market in 2003.
Is that a serial ATA you are talking about?
Right. So now there's a place at about a penny a megabyte that gives you an alternative to tape. But again the speed between the two, the reliability between the two is night and day.
So is the ATA much more reliable besides being much faster?
Yeah. When you put a tape in a vault, there is a chance -- and not a remote chance -- that you're not going to be able to read it. With this technology, that virtually never happens.
What about the idea of how to judge that value of data. Isn't that something that has been kind of a holy grail also? How are you going to keep track of which is the data that is important and which isn't?
Our customers have to know their own business processes right? But we can tell them how often it is accessed, what kind of utilisation they are getting from it. And then they should know from their business process how does that relate back to their business process. They can set policies. And then we came out with software that helps them automatically move the information.
For example, if you're a bank and your business process is funds transfer, in a fraction of a second you could move a billion dollars. That information is kind of critical and the bank would know that, right? We can give them the data about the data. From those two points, they can develop policies and then move the data using our software.
To what extent can ILM be done right now, or is it still sort of a dream?
The tiered storage is there today. Everything I talked about in storage. This is not in the future. These pieces are there today.
Give me an example of what is not there now that could be helpful down the line. Is there a particular kind of metadata that is missing right now? (Metadata is descriptive information about data.)
I mean there are a lot of kinds. It's not that it is missing; it is how easy it is to get. You know there have been a lot of advances made there, but we are certainly not where we want to be. I mean it is a journey in that area.
I'll give you a statistic. For every copy of information that exists on disks, there's eight and a half copies on tape. This has to do with incremental backup. I mean there is no real good way to basically have this metadata engine so powerful that I could say that I will only need one copy on tape. And I will let the disk handle the incremental adds and when I get enough of them, I'll update it again, and get rid of the old copy, you follow?
