Deni Connor met with Mark Bregman, executive vice president of product operations at Veritas Software, to discuss how the company views storage and how it competes with other vendors in the market.
Q: What is Veritas' vision for storage for the next year, three years - and how does it fit with the storage industry vision?The next year is clear - it gets foggier the further out you look. A trend has been taking place for some time that is coming to center stage for most of our customers - that is network storage.
There has been a lot of promise over the last several years driven by the desire to get increased data and application availability. Now, with the economic situation and focus on the bottom line, storage networking is being looked to as a way to get better utilization without buying more gear. It is also being driven by cost pressures - the hope that by having all storage on a network, you can manage it more efficiently and have fewer people dealing with it.
Those people who tried to implement advanced storage networks in the last 12 to 18 months saw that the reality didn't live up to the promise, so they could get better utilization out of the hardware and reduce their hardware costs, but instead traded it for higher management and people costs.
The realization has come about that storage networks are pretty compelling on paper but are pretty complicated, and interoperability issues continue to appear. Finally, the tools to manage, report and support the distribution of storage over the network and do charge-back are not in place yet.
There's a huge opportunity for us to help customers cut through the complexity of storage and solve interoperability issues, because as a software vendor, we're platform- and device-neutral.
Q: You say network storage is coming to fruition even though there are issues of interoperability and management. Is that what you are saying Veritas is going to be doing in the next year - solving those problems?We are already. We just announced the release of our storage resource management product, SANpoint Control, which brings all these network storage elements together - it supports NAS as well as direct-attached storage, and provides what we call active SRM. SRM has been focused on reporting, but we go beyond that in allowing users to take control and manage their storage resources, not just inspect them. So it's not just a microscope, but a surgical instrument. As things become more complex, just having inspection tools and not the tools to manage them isn't going to be sufficient.
On the three-year horizon, the thing we are looking at that will be hitting the market sometime the beginning of next year, is a product we call Global Operations Management, which raises the level of management from SRM to business-object management. What CIOs need are tools that help them understand their storage data from the view of application and data availability, and how much they are spending to store data in a certain location. We'll also add capabilities such as charge-back and service-level agreement monitoring and control, which are important as IT organizations are increasingly supplying their services to the enterprise as a utility.
Q: EMC says it is the only company that automates storage now. What is wrong with that in Veritas' view of things?EMC may automate storage management, but they only automate it in an EMC world. We automate storage independent of the hardware platform. From our point of view, storage is a commodity, and EMC is not willing to think of storage as a commodity for obvious reasons.
Q: You automate and virtualize storage from a file system level. What advantages does that have over array-level automation and virtualization?That's not exactly true. Veritas virtualizes at the host, not file level. EMC virtualizes at the array level. If you have a collection of storage, the servers can see a virtual pool of storage. EMC can do that as long as all the storage is EMC. We can do that whether it's EMC [arrays] or not. We can even add an EMC virtualized pool of storage into another pool of storage.
The thing that is happening in the broader marketplace is that there are aspects of virtualization that are done better in the fabric of the storage network - in the switches. We are working with a variety of partners to put the Veritas virtualization software into their intelligent fabric devices.
Q: Cisco [Systems Inc.] is one of those, right?Could be.
Q: You have a relationship with Cisco that has led to a lot of speculation lately - that Cisco will acquire Veritas, or if nothing else, use your virtualization in its storage switches. What is your relationship with Cisco?Aren't we going to buy Cisco? We are working with partners, and Cisco is one of them. They are members of the Veritas Powered program, which is a group of vendors that want to connect to our software. They could be array or application vendors. As the intelligent fabric is emerging, there are a set of new devices that aren't servers or arrays. They are switches or other gear that sit in the fabric. We are working with vendors who want to embed our software in their devices. Cisco, Rhapsody Networks, Adaptec and Maxxan are members of Veritas Powered.
Q: What effect will Cisco's entry into the Fibre Channel market have on Veritas?Independent of what Cisco does with us, it is going to grow the market.
The storage industry has said that CIM is the management of the future. What is Veritas doing with CIM?
We are very active. This week we are participating in the CIM-SAN I, the demonstration of a distributed network that is managed via CIM, at Storage Networking World. It is a real demonstration that CIM can work.
Q: Since Veritas is set up to manage storage on many platforms, hosts and operating systems, does a technology like CIM do all that much for you?It does matter for a couple of reasons. Since Veritas has been known as a vendor that works with anybody and our investments in the past have been in building special adapters for everyone's interfaces, wouldn't a standard put you out of business? Having a standard way of doing things makes our life easier. We can focus our resources on the services rather than the adapters. There are still going to be features and capabilities of devices that are not exposed through the CIM interface - for those, we will exploit by doing adapters.
Q: Because you don't have to duplicate effort then. Some observers will say that standards are built to the lowest common denominator and yet be critical of companies such as EMC that have its WideSky APIs. Don't you need custom extensions to the CIM standard to expose those features that Veritas can be the best at?I look at it a little differently. CIM provides a set of capabilities for a device in a standardized way and then adjacent there may be extensions a particular vendor provides. Of course, that vendor is doing it for a simple reason - they want customers to take advantage of those extensions and, therefore, be locked into using that vendor's products for the extra functions they provide. We will do what we've already done - build those extensions to exploit those features.
Q: That program is called what?It's the Veritas Enabled program, which is how we build interfaces with other vendors.
Q: Is it more important for Veritas to partner to gain capability or for you to develop to gain capability?If you look at the core of capability we provide, such as data protection, application availability and management - those functionalities we want to develop. We may partner if we really don't have the capability or for a time-to-market advantage, but that's our core asset. The stuff at the perimeter where build capability to connect to other devices or applications, there it makes sense to partner.
Q: EMC says it will focus more on software and increase its distribution of revenue to somewhere around 30 percent. How does a company like EMC threaten a company like Veritas?EMC, like a lot of other large players, is a very important partner for us in many areas of our business. Their arrays work well with our software. Having said that, EMC has stated that they are going to have to move from hardware more to software. That's a consequence of what's happening as hardware becomes more commoditized. As EMC does that, clearly there is going to be more friction and areas of overlap. I'm not overly concerned in the short-term because EMC has a major challenge making the transition. The other challenge is being neutral in their software business when you have a hardware platform to sell.
Q: Sun in the past year has started to introduce a lot of products that they formerly used Veritas for. How does that affect your company?Overall, Scott McNealy has said he doesn't want to have dependencies, therefore Sun wants to build everything they need. From the point of view of how quickly they can get there, there's a level of complexity in storage software that most people don't appreciate until they start digging into it. So, yes, in the latest release of Solaris 9, their file system is better than the previous release. Is it as good as what Veritas has? No. It doesn't perform as well, and it's not as robust. There are going to be customers who say Sun's software is good enough. That still doesn't make it better.
Q: That sounds like a comparison between Yugos and Porsches ...The same argument was made in the Unix business. Analysts said Windows NT was going to kill Unix. While NT grew, the Unix market also grew. What you can do with a Yugo or Porsche is different.
Q: Tell me where Veritas is going in respect to backup and recovery from disk instead of tape and whether you think it has the potential to change the dynamics of backing up a network.The short answer is yes. The longer answer is that backup is a funny technology because it's been around since the beginning of the computer age. It's thought of as mundane, but has evolved rapidly in the last few years with the advent of inexpensive disks and the need for instant backup and restore.
Some people have predicted that tape is going to become obsolete. I don't believe that for a number of reasons. One is that although the price gap between disk and tape continues to decrease, tape is still a very cost-effective storage medium. There are also regulatory and legal reasons why you will want to have a long-term backup instead of being able to recover from a failure.
Backup strategies are becoming much more sophisticated and they are involving disk and tape. It's not that disk is replacing tape so much as it is that disk is inserting itself as an intermediate part of the backup process.
Q; A staging platform...
Yes, disk allows you to do instant, snapshot backups and restores, but it also lets you do incremental backups. It's too early to tell if there is a single paradigm that will emerge. There may be a shake out in which people will say we tried all those different things, but this is the right approach.