The following is excerpted from the executive roundtable discussion at this summer's Network Storage 2001 conference. The session was moderated by Farid Neema, president of Peripheral Concepts (www.periconcepts.com) and the conference organizer, and Don Fulgham, chief information officer at Equinix. The panel included
Mark Cree-general manager, Storage Router Business Unit, Cisco Systems; Nora Denzel-VP and general manager, Network Storage Solutions Organization, Hewlett-Packard; Mark Lewis-VP and general manager, Enterprise Storage Group, Compaq; Stephen Luczo-CEO, Seagate Technology; John McDonnell-chairman, president, and CEO, McData; and Linda Sanford-senior VP and group executive, Storage Systems Group, IBM.
What are the most significant factors or movements that affected the storage industry in the past year?
McDonnell: We saw the roll-out of enterprise-size Fibre Channel [storage area network] configurations. I can think of five examples, four of which grew from 50 to 1,000 ports in 18 months. The fifth one has installed an initial SAN configuration of 1,500 ports with 100TB of storage as a starting point. This is a different mind-set than we saw in previous years.
Luczo: This was the year of availability of the underlying technologies. [However,] the downside of storage being hot is that everyone wants to label everything SAN or NAS [network-attached storage]. We have to differentiate what's real technology from what's just integration. The risk is that the hype may cause a bubble that we won't be able to sustain.
Lewis: The emphasis this year is on the value of information, as opposed to speed, performance, or technology. Customers are driving vendors to solutions that solve their problems. Last year, we did drive some ideas in networking that did not exist before and demonstrated our ability to interoperate.
Sanford: The most important factor I see is the e-business revolution. This is leading to more data that needs to be stored, protected, and managed, and above all, is driving the need for universal access.
Cree: Last year, everybody expected IP storage would take 36 months to mature. The iSCSI first draft was created 18 months ago, and both Cisco and IBM [and others] are delivering products today. We now have more than 250 companies committed to delivering IP-storage products. A larger percentage of storage continues to be sold outside the server. Storage networking is in its infancy. But the tools to manage data over distance are not really here, and there are no agreed-upon standards for storage virtualization or storage consolidation.
What are the top priorities of IT administrators? Have they changed from last year?
Denzel: The top priority is to manage data. The one thing that resonates across the world is staff shortage. How do you allow customers to scale 10 times with the same people print? What person-per-gigabyte ratio do you use? From one of our customers, the answer was "3.5: We have 3.5 administrators to do the job, regardless of the number of terabytes."
McDonnell: Return on investment [ROI] is the issue that we are seeing most. Last year, users said, "I see that technology; I want to try it." Today, they say, "Give me a compelling ROI," and "How do I reduce my costs?" The only way to reduce cost is to cut down management costs.
Lewis: Early SAN adopters tell us that just by consolidating, they manage three to four times the amount of storage per person compared to a couple of years ago. ROI will continue to improve. Managing data and storage is where the problem lies. You see an emphasis on the total cost of ownership now.
Last year, the unanimous answer was "availability." Are you saying it has shifted to "cost?"
Sanford: Availability is still important, but cost is more on the mind of administrators. Today, half of the hardware IT spending is on storage, whereas it was only 25% two years ago. Enterprises look at the biggest chunk of their spending and zero in on it. We are only seeing the tip of the iceberg, and total cost-meaning manageability-is the important parameter.
What can users look forward to in the next two years?
Cree: As users deploy storage, their concern is how to manage it. It is more expensive to manage it than to own it. Over the next two years, IP will be one way to make storage networking less painful to users. Speaking of cost of ownership and shortage of staff, finding IP expertise is not that difficult. You can look forward to that in the next two years.
Sanford: For the next two years we have identified five key technologies that will directly or indirectly impact users: virtualization, IP storage, storage management, policy-based automated storage management tools, and SAN-wide file systems.
Is Fibre Channel threatened by IP storage?
Cree: Fibre Channel and IP solve different problems. Fibre Channel is the premier data-center storage technology. It will continue to be predominant in data centers, but we see IP for the rest of the problems. Users that don't want Fibre Channel because of the perceived complexity will buy IP storage, and this is a much larger market. In the immediate future, they are very complementary. iSCSI opens a whole new set of customers to block-level access to storage. It does not really threaten Fibre Channel-at least not until 10 Gigabit Ethernet shows up.
McDonnell: I think that Fibre Channel is the best storage technology in a data center. Does that mean that iSCSI is not going to happen? No, I agree that iSCSI will be here. The issue is: How are we going to solve problems specifically related to storage? We have not solved storage-related problems in IP. It's not trivial, and it will take time. Storage over IP and iSCSI will fundamentally change the equation, connecting geographically dispersed SANs. This will drive demand on all storage. Rather than speaking about speeds and feeds, we have to look at it from a solution perspective.
Sanford: We will have to deal with a multi-protocol world. No one technology will win over the other.
Is the profile of distribution channels changing significantly with the complexity of storage? How is your company dealing with the change?
Sanford: We're using a hybrid channel. We're counting on the channels very much. They are getting us in places that we can't reach otherwise. This year, we will do more business through our channel than selling direct. We're also finding that there are competencies that our channel partners bring. They have opened total solutions centers where customers can come in and test in a typical environment and the entire behavior of their application.
Lewis: It's much more about providing services and total solutions. The market belongs to those who provide value-added. Channels are solution partners.
McDonnell: There are many Fortune 100 companies that don't go through channels, but channels can add great value. We have to support both models. Today, the solutions are complex enough that the user says, "I need you to put the entire thing together."
Cree: Storage by itself is a horizontal proposition. Users need a vertical solution, and that's what VARs and integrators bring to the party.
What are the most significant changes that will affect the storage industry in the next 12 months? Or, what will 2001 be remembered for?
Denzel: Technologically, we'll find out what iSCSI is made of this year. We'll start to see products. It will not replace Fibre Channel. It will not be a tornado market.
Cree: I agree, except that I think iSCSI will be more of a tornado. Standards will not be approved before the end of this year, but next year it will be the real revenue producer.
Lewis: I'm more on the conservative side. I don't believe this is going to be the year of iSCSI. We have barely scratched the surface of SANs. My guess is that the technology will start taking hold this year.
Sanford: Interoperability will take more of a center stage. Customers are putting real pressure on us to make things flexible and interoperable.