Techworld had the opportunity recently to talk with James Whitemore, Sun's marketing VP for the Data Management Group (DMG). It was just after Jonathan Schwartz had received the CEO reins and the May quarterly product launch held. James talked about several things that just didn't chime with me until I realised that Sun's DMG aim is to build storage platforms for data management, not just storage products.
For Sun, storage and data management are like Siamese twins; inextricably linked.
James said that data growth looks unending and focused on aspects of this. For example, vaulting tapes was "a landfill approach - you bury the data and never find it again." He mentioned ostrich customers who stick their head in the sand and hope problems won't happen.
I disagreed with this. Businesses vault tapes for a very good reason, to be able to restore data if a disaster strikes a data centre. It's what keeps Iron Mountain in business.
James said: "The problems won't be solved by today's storage architectures." This didn't gel. What problems are these? Today's storage architectures and evolution look pretty good. SANs (storage area networks) of both Fibre Channel and IP varieties can hold more and more data. NAS (Network-attached storage) is looking good with great scalability and clustering. Storage virtualisation of both SANS and NAS is improving their management. Single-instance storage will help reduce capacity demands. What was James talking about?
He started talking about data and storage: "Storage is a shared resource ... Sun's new focus is on how customers get value from data. Data is not a liability; it is an asset." With shared storage: "There is a problem of rights management, identity management and access control."
It dawned on me that Sun isn't talking about storage problems per se but about storage and data problems. Other vendors will talk about storage resource management problems but Sun is looking at stored data management problems. James said: "We've known for a long time that storage is not that important. Data is what is important. Data management is the thing."
I realised that Sun is focused on solving the data management problems of its customers and not just the storage management problems.
On being asked how Sun competes in the storage resource management (SRM) field James replied: "We provide a management infrastructure through a portal. We don't tell a customer to throw away what they've already done. We've added identity and access tools."
Sun actually ships an SRM product based on the (HP's) AppIQ product What is has added is a portal through which additional management software can manage storage (and data management) products.
It seems to me that Sun sees storage through a data management lens. Asked if Sun is more about building a storage platform to facilitate data management, James said: "I think that encapsulates very well what we do."
Sun is focusing on active data, stored data that can be used, that needs using. Yes, it can store data on DAS, SAN, NAS and in tape libraries and vaults but it is focusing on data management issues of active, not dead, data. It wants business to be able to share access to its data, access to properly authorised people, and access from applications that delegate some data-handling capability to the storage devices themselves. Only this way, with new storage architectures, can the problems of shared data storage and access be solved.
Today's storage products in the existing DAS, NAS, SAN and tape spaces cannot solve the coming problems of shared data access. Sun is a systems company and it is combining Solaris middleware, Galaxy servers and developing storage platforms to provide better data management products. This way Sun is striving to develop its own and unique approach to using storage; it's a differentiator in development from EMC and NetApp, the dedicated storage vendors on the one hand, and HP and IBM, the other soup-to-nuts server, storage and system vendors on the other.
The development is collaborative in style and the key word is 'systems'.
In James' view: "Customers are looking to us to solve data management challenges." The problem 'is not the physical persistence of the data on the storage device'. We need to 'move beyond storage and the physical persistence of data.' Colleague Martin Warren, Sun's UK product manager for DMG, said: "Customers tell us they want to reduce complexity. Some of this is to do with data classification. Some customers don't know what (data) they have and where it is." There needs to be a reservoir of data in one repository.
Honeycomb is based on commodity hardware with a layered metadata repository. It is for customer applications that capture a huge amount of static file-based data at high-speed and need it integrated and analysed. Searches are going to be conceptually intensive and carried out by Honeycomb itself, directed by server-based applications linked to Honeycomb through an API. Sun partners may well, and it is hoped will, provide front-end applications to use Honeycomb in particular markets.
Beta testing organisations are writing their own applications to use Honeycomb's metadata layer.
James says content management vendors are excited about this, organisations such as AXS-One and Molina. Data could be photos from image-intensive online service providers. Another potential market is storing retail transaction data for business intelligence data-mining operations. Another could be the processing of RFID-based data. James said: "We're courting potential partners in this space."
It seems to me that is a substantial niche market opportunity but not a mass-market one. Honeycomb-based systems (Honeycomb HW plus software to drive it) will be relatively expensive and not volume market-priced. James Whitemore agrees and says that it is an 'absolutely viable product category'.
In the future Honeycomb's software layer will get abstracted from the hardware layer as part of an overall data management platform.
James talked about aspects of the StorageTek acquisition. At the time of it StorageTek's HW business was stable and secure. It had, for example, 69 percent of the mainframe virtual tape market. The business wasn't growing and wasn't declining. Its future was a problem because StorageTek didn't know where to go (grow).
(In my view EMC had stolen StorageTek's ILM (information lifecycle management) cloak and the StorageTek emperor had no clothes. The Sun acquisition has enabled StorageTek, now DMG, to adopt or jointly develop a Sun data management-based strategy. This will involve virtualising all the main types and layers of storage, adding in identity and access management controls, and developing very much more intelligent storage platforms on which data management tasks can be executed.)
Sun immediately bought NAS and high-end disk to the DMG sales force filling in a couple of gaping holes in their market coverage. The supply contract with HDS for high-end drive arrays has just been resigned for a further three years.
StorageTek also bought mainframe tapes and virtual tapes to Sun, very important for solidifying Sun's data centre presence in the largest enterprises. James said Sun was acutely aware of the value and sensitivity of StorageTek customers. He said: "Customers told us 'don't screw up StorageTek'. Maintain its commitment to mainframe and the high touch sales model," which is exactly what Sun is doing.
DMG will refresh its hardware. A next-generation 6920 storage controller will be announced around the end of June and use 4Gbit/s technology. James said: "It's different in price/performance characteristics from the HDS (TagmaStore) product and that's what differentiates it."
Virtual tape travails
The recent VSM virtual tape announcements by Sun did not mention VSM for Solaris, other Unix and Windows and this has been remarked upon. It turns out that the scale of the code transfer problem is simply huge. VSM was produced by StorageTek for the mainframe environment and StorageTek had 35 people working on it. Taking it to the 'open systems' arena of Unix and Windows inside Sun has involved transferring the mainframe code to a Java/Solaris base, something StorageTek was not doing in its efforts to bring VSM to Unix/Windows.
Sun has now poured an enormous resource increase into this; 3,000 people are now working on VSM for open systems. It has become a vast software project. Martin Warren said: "You'll be surprised at the speed of the result."
Sun and EMC
For Sun the top competing dog in the storage market is EMC. Like Sun EMC is also not content to just develop and sell storage hardware and storage-only systems. It wants to add software-based value and provide for future earnings growth away from steadily commoditising hardware.
James said: "We're always head-to-head with EMC in every (storage) engagement. It is formidable competition. EMC cannot compete on identity, but then EMC isn't a data management company."