EMC CTO speaks out about AutoIS, openness, and more

By Heidi Biggar

In the midst of one of EMC's longest and broadest product announcement cycles on record, CTO Mark Lewis sat down with InfoStor to address questions about the company's future course. An excerpt of that conversation follows.

At EMC Analyst Day this past June, CEO Joe Tucci said that the industry could expect to see "the richest set of product introductions in the company's history" over the next three quarters. That puts you at about the halfway mark. How would you describe your progress, and what type of product announcement can we expect to see over the new two quarters?

We're still very much on track. In the coming weeks and months, you can expect to see announcements across the board.

Of your recent announcements, which do you consider to be most important, and what type of message do you hope to convey to the user community?

To date, the most significant is the Auto IS Chapter 2 launch, mostly because it gives users a proof point for managing heterogeneous storage networks.

The three fundamental messages we want to convey to users are 1) that even in the economic downturn, we continue to be a technology-oriented company, one that invests in R&D and that drives a complete portfolio of products and solutions, 2) that we [continue to] drive the open message, and 3) that we are spending the effort to build an independent software stream [independent of our hardware] and an independent software strategy that are a credible value proposition to our customers.

It has been reported that EMC has invested as much as $1 billion in its software business over the last year. Is this accurate and, if so, do you expect to continue this pace going forward and where will the bulk of dollars go?

We've [been spending about] $800 million per year in total R&D, of which roughly 75% [has been in] software. We intend to remain a technology company, [realizing] that our core value proposition to companies remains our R&D and the technology that we possess across that R&D. We don't intend to economize for the sake of getting out a technology. Whether our R&D spending goes up or down a little, I won't comment.

EMC announced its acquisition of Prisa and its intent to acquire the assets of SANrise. How do the two acquisitions fit strategically into EMC's overall strategy?

Our acquisition strategy is to focus on software in key growth areas. I can't comment on SANrise because the acquisition is still pending. However, Prisa has excellent software (e.g., visualization, topology, performance management, and configuration management) for the entry-to-mid-tier SAN management space. As we move more into this market, we [have realized that we need] a complete, but very easy to use, tool suite. Prisa's technology will supplement our overall open software strategy in these markets and will become a fundamental part of our Auto IS portfolio.

Of today's hot trends (e.g., unified management, automation, SAN/NAS convergence, etc.), which is EMC investing greatest in?

Fundamentally, we've invested 100% in automated network storage because we believe the overarching macro growth is in network, not direct-connect, storage. We believe that, in terms of the value equation, software is becoming more and more important. That's why we are biasing our R&D toward software.

In terms of our business model, while we continue to focus the bulk of our R&D effort in software, we'll still do hardware where it is needed, but we'll leverage off-the-shelf components wherever we can.

Our key "outside-the-box" market is fixed content. The cost of storage has moved to where it's actually less expensive to store most analog content [on digital media]. Exabytes of data simply aren't being stored efficiently.

How would you rate EMC on the following: scope of software portfolio and ability to execute Auto IS strategy?

Let me tell you where we think we lag, as well as where we have executed. Coming in from HP, where I saw EMC lag [its competitors] was in its ability to articulate its vision and strategy to customers. And, in software, that is an important part of the overall sale. One of my goals is to better articulate our vision.

[On the execution front], though, I think EMC ranks a solid "A," when you compare it feature by feature to [competitive products] in the areas we cover (e.g., provisioning, SRM technology, storage infrastructure technology, and replication management technology). In these areas, ...we have out-executed our competitors in terms of functionality [and support].

Historically, EMC has been criticized for being proprietary. Is EMC any less proprietary today than it was two years ago and, if so, what steps were taken and will continue to be taken to make strides in this area?

The answer is, yes. EMC is a lot less proprietary in the sense that it is de-coupling more and more hardware and software functionality. We are still focused on adding value with software. Being open doesn't mean you don't want to add value. It simply means that you want to provide APIs and interface layers so customers can have choices at more levels.

[We are building the Auto IS strategy with the same goal]. For example, our new provisioning, visualization, and topology management tools support IBM, StorageWorks, HDS, and Network Appliance today. Veritas would argue that they have more coverage, but [in many instances] their tools are volume managers or host-based SRM that works with any storage. They're heterogeneous by matter of course. That's like saying Microsoft is heterogeneous because it supports all storage. The fact is it doesn't touch storage directly. The operations that we perform require that we directly touch the array. This is a much more difficult task, but we are still working to make these operations heterogeneous.

When do you think users can expect to see CIM-compliant software that will allow them to manage all the devices in their network? How do you expect the CIM/API scenario to unfold?

I believe that realistic products will begin to roll out by the middle of next year. Products will begin to appear before then, but you'll begin to see good compliance in the middle of next year.

The concern for customers is how [to take advantage of] these standards [without having] to replace their entire set of data-center software and hardware to do it. That's why we've adopted a two-prong strategy: Support all the standards because that makes the engineering easier, but still write to the APIs or do the direct coding that we need to do into the unique and existing APIs so that we can offer the best of both worlds.

This article was originally published on November 01, 2002