All about the info

EMC exec Mark Lewis tells why the future is about information, not storage.

Say the name EMC today and most people think of storage. Mark Lewis, EMC chief development officer, would like this to change. His marketing message of late has EMC morphing from storage vendor into an information infrastructure company. Naturally, information life-cycle management (ILM) technology plays a role, but so do two other new data center linchpins: service-oriented architecture and virtualization. Lewis describes his big-picture view in this recent interview with Signature Series Editor Beth Schultz .

Storage today is all about managing data based on business value. What's up with ILM?

The value is understood, and now we're talking about implementation in three major steps. Step 1 is tiering storage - defining applications and their uses and then attaching a storage type to them. Step 2 is looking at specific implementations of ILM, with archive and compliance strategies, for the 20% of major applications that are responsible for 80% of the data - your structured data, or database applications, your e-mail and unstructured content files. The third step is a vision EMC is developing toward. EMC's goal is to deliver ILM as a service-oriented architecture capability, where it becomes a part of the infrastructure and viewed just like the network or any other service.

How would that work?

An enterprise plugs an application in to its service-oriented architecture, which would deliver compliance, archiving, data tiering, automated data movement - even data protection and recovery - in an automated way.

With VMware for servers and now Invista for the storage network, virtualization is of obvious strategic importance to EMC. What role does it play in your ILM strategy?

Information life-cycle management has multiple characteristics in which virtualization is absolutely necessary. Imagine if an enterprise has hundreds of applications in this architecture running in a storage array. It can't ever reboot or move the storage array. It needs the virtual layer so it can deliver on the services expectation [of 100% uptime] and keep the infrastructure up without impacting those applications. That's a part of the service delivery of ILM.

You mentioned virtualization is important to ILM for multiple reasons. What's another example?

The second use for virtualization is to enable tiered storage, for improving the class of service and optimization. You can't get this with today's storage.

What if storage could self-optimize? If an enterprise creates a volume that has a silver requirement but gold storage is available, why not let the volume be on gold for delivery of a better service level? And then when gold-level storage is needed, that volume simply moves to silver.

To get there requires a "data object model" - meaning applications, and even users and administrators, would be able to describe their data requirements in a very structured way. From there, a combination of management and virtualization technology will enable the automation and the optimization.

So on the virtualization side, do VMware and Invista ever meet?

No. They're different products, built from different software that happen to share a common technology model. But they're complementary, and from a management perspective, EMC will offer tools and connectivity in creating a single architecture from a services orientation to provision storage or applications. And there's no reason an enterprise can't take advantage of storage virtualization when it also has a virtual server environment.

Take the example of VMotion, the VMware add-on that helps enterprises migrate an application from one physical machine to another while it's running. Imagine being able to move storage volumes while they're being accessed across storage arrays - that's a marrying of those two technologies from a process and production standpoint. It lets an IT manager think, "Wow, I could dynamically move my Exchange server from New York to New Jersey while I'm processing e-mail. I can move the computing resources and even the storage that's used." Think about the amazing capability of physically moving a data center someday while applications are up and running.

Today, we have customers that have hundreds of applications running in a single array. They can't take that down for even 10 minutes - imagine having to coordinate that downtime with 100 application owners. People spend years doing firmware upgrades, not because they don't want or need the hardware, they simply don't have the time to take the application offline for a couple of minutes even to do an upgrade. Change management is a major pain point.

So how do you tie all of these - ILM, virtualization, service-oriented architecture - into one grand vision?

Through the services-oriented architecture and virtualization, we want to deliver an information infrastructure. An information infrastructure is fully capable, when given service-level requirements on data, to become solely responsible for providing the protection, migration, lowest cost of ownership, recovery, archiving, compliance. It applies all of the characteristics needed in a holistic way so that the application and computing environments are relieved from ever having to worry about anything associated with data and data protection. That compares to the storage infrastructure of disk drives, tape drives, the arrays and some point software for replication.

Learn more about this topic

ILM in action


IT execs prepare for ILM, but the tools aren't all there yet


Organize to automate


Net intelligence a balancing act


The latest storage virtualization news

Network World's Storage in the Enterprise newsletter

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2005 IDG Communications, Inc.