THE Automation Challenge.

An Interview With ATL's Kevin Daly

When Bob Dylan wrote, "the times they are a' changin'," he had no idea how accurately that song title would describe today's storage landscape. Indeed, the era of social unrest that Dylan wrote about in the early 1960s may have had somewhat less of an impact on society than the changes taking place in storage today will have on enterprise computing.

LTO and Super DLT tape drives due out late this year will raise the bar for storage for the next few years. The drives will be faster than those currently available and capacities will be substantially increased. Improvements in 4mm and 8mm helical scan, and next generations of half-inch and other drives provide additional data storage options, making format, and even form factor, a difficult choice for many companies. But, in many ways, the 5.25-inch form factor is beginning to dominate tape technology in the late 90s the way it dominated disk technology in the 80s.

These faster, higher capacity devices will be necessary as a result of the meteoric increase in the amount of data created for transmission over the Internet, as well as the sustained growth of corporate databases. Digital imaging, digital audio, and digital data will fill hard drives more rapidly than ever before. Mission critical and, in many cases, business essential data will continue to be produced at increasing rates, and the data must be readily available or easily retrieved.

The rapid increases in hard drive capacities continue to push the cost per megabyte of hard drive storage ever downward, making data less expensive to store, while necessitating improved tools to manage, retrieve, and analyze the data. Without adequate management software or strategies, the task of managing data can become nearly impossible, and the costs of managing data can tax the budgets of even the largest organizations.

A wide range of other issues will impact the future of data storage. Network Attached Storage and Storage Area Networking are buzzwords referring to two alternatives to drives directly installed in servers and are perceived by many to be mutually exclusive.

Gigabit Ethernet, Fibre Channel, SCSI-3, and other emerging data transfer protocols are important issues relating to moving data between drives and servers. Next Generation I/O (NGIO) and Future I/O, also considered to be competing for mindshare, if not a place in the future of computing, offer other challenges to system developers who must map out plans for the future of computing.

In few areas of computing do all these issues converge as completely as they do in automated storage systems. Indeed, automation companies must carefully choose to direct development efforts to serve their customers (always a rapidly moving target) with the appropriate products at the right times, using the right components. Storage Inc. discussed these issues with Kevin Daly, CEO of ATL Products, to get his perspective on the present, and future, of computer storage.

Daly refers to a five-year cycle in the enterprise storage business. The value of data is considerably higher today than it was five years ago, Daly noted. Tape automation and data backup used to be what Daly called "next day storage," "where (tape) was used for backup and archiving and, in a sense, automation was really peripheral to the system. This was a system where people had modest expectations. They were looking for performance, but that was not their primary concern. Because backup was secondary, their requirements were modest: if it didn't work today, it would work tomorrow. And, if they lost the day, tomorrow they will bring it up to date," Daly reflected.

"Today, the confidence people need to have in their data protection has increased enormously. Issues like high availability and high performance have increased significantly over the past few years. This trend shows every sign of continuing to increase," Daly said.

"We see ATL evolving from a next day storage player to an external storage player. This may involve a range of technologies to allow for an overall system that meets the expectations of very high availability and very high performance. These are relatively new expectations in a lot of environments. We see that, where three or four years ago, automation was the focus of what we did; now we think of automation as the foundation of what we do. The focus will shift to providing the function of data protection, archiving, or whatever needs to be provided, using a number of tools, including automation. There will be a great deal more native intelligence to handle things like individual component failures, the ability to distribute data around an organization so that it can be used by a lot of people, so an external storage has to resolve how it will be used by a number of users. This is relatively demanding--it's more than a set of SCSI commands," Daly continued.

Daly sees tape moving from the server to external systems: "There's a nearly religious fervor around whether to have a SAN architecture or a NAS architecture in a system. While one or the other may be a more natural starting point for people, depending on their needs, many common elements will emerge between the two. In a few years, the distinction between SAN and NAS will not be as big an issue as it seems to be today."

Daly explains that "significant segments will start using NAS, while others will start in the SAN market. It is not unreasonable to expect that, over time, those two will evolve to where both will begin cross-fertilizing. The fears of choosing the wrong architecture and being on the wrong end of the power curve are stronger than they need to be. There is a lot more flexibility in how one continues to grow these architectures so that they can continue tuning them as the architectures change, and as their needs change, that make sense, depending on where you started."

Daly continues, "For us, it translates into storage systems that are becoming external storage systems with a different emphasis. The way a NAS works is one which is appropriate for appliance-type products-where the real emphasis is on ease of use and minimal local support. If you can provide functionality with as little local support as possible (this would be preferable). Such an approach will deserve consideration."

"Nothing comes for free. There is a limit to the scalability of NAS. (NAS) approaches are for people with a range of needs that won't grow without boundaries. The SAN world tends to be highly performance-oriented. The focus is on expandability. The price you pay is that the complexity will require local support, care, and feeding," Daly noted.

"From the user's point of view, they would like the data protection scheme to act the same no matter how the architecture is put together. The user would want the same approach, whether they were in Kuala Lumpur on a little network with a NAS on it or in San Francisco at a corporate headquarters with a very large SAN. They would like their interaction with the system to be very similar in both cases. That's where our focus will be. Over time, the way the technology is implemented will continue to change. One of the key challenges visible to our customers and users here is the fact that, whatever challenge we are addressing today, we know in our hearts that they will be bigger tomorrow, bigger next month, and bigger a year from now. Everybody wants to make an investment, particular capital investments, that will be good for a long period of time. Scalability is a very big buzzword in the industry. I'd like to offer someone a solution that you can look them in the eye and say `for sure, this solution will grow with you to cover your needs for the next five years.' In actual practice, it is difficult to be confident of what kind of obligation that implies. Think back 20 years ago. If you invested in a system (you may have backed up data using a) 3480 drive. This was stable for a long period of time. If your needs doubled, you needed twice as much of the same thing. When we chose DLT (as a technology to support in our automation products), we felt it had the technical legs to continue to increase in capacity and transfer rate for a considerable length of time. Though the customer's needs increase, the technology improves. You can keep from physically scaling things 1 for 1 as needs increase. We've been through three generations (of DLT drives). We've been very successful. It would be unfortunate if we had to bear the whole burden of scalability (by increasing) the number of drives (rather than replacing an earlier generation drive with a higher capacity drive that is backward compatible)," Daly said.

"We are just about to make a major step, going from DLT 7000 and 8000 to SuperDLT and LTO. From (the automation) point of view, they are essentially interchangeable. (Either one has a three or more generation roadmap for increased capacity and performance). (These new drives) allow us to continue to meet the growing needs of our customers with reasonable systems, without requiring the systems to become arbitrarily large or arbitrarily complex," continued Daly.

It should be noted here that ATL Products was acquired last year by DLT-maker Quantum. However, both Quantum and ATL executives at the time indicated that Quantum's management of ATL would be at "arm's length," allowing ATL to determine which technologies best suit its customers. Daly's inclusion of LTO along with SuperDLT as a target technology for future automation products seems to confirm this.

Finally, Daly notes a positive impact that Y2K may be having on the industry. "I'm not a big fan of the focus on Y2K. I think it's gotten more excitement than it deserves. But, one of the things it has done is that it has made people very aware of their responsibility and liability for protecting corporate information. As a company in the data protection business, we consider that to be socially redeeming behavior. It was very hard a few years ago to get people excited about not having data protection in place. Today, people recognize the importance of protecting their data. They have that obligation. If (Y2K) warrants enough (attention) to turn a CIO's head gray, if not totally around, the merging of front office and back office exposes a lot more of the corporate database to potential attack from things like viruses or worms, or things like that. We used to be able to operationally build a brick wall between the enterprise and the data systems, but the more complicated e-commerce becomes, the more vulnerable all the data is to becoming threatened from the outside," Daly observed.

Borrowing another line from a Bob Dylan song; "when you ain't got nothin' you've got nothin' to lose," it's clear that the corollary clearly applies--" when you have something, you've got something to lose." When that 'something' is data, the loss to a company can be devastating. Clearly, data backup is important, and becoming even more important as the value of the data increases.

According to Daly, we're at the beginning of a new five-year cycle. With the introduction of a handful of new drive technologies with solid roadmaps for capacity and performance improvements; a new focus on NAS and SAN; high speed data connections; and new data management tools, the next five years will provide significant challenges to IT managers, system integrators, automation product makers like ATE and others.