By the end of this year, you may be using "Bluefin" -- whether you know it or not.
Bluefin -- officially known as the Storage Management Initiative Specification (SMI-S) -- is a set of interfaces designed to make it easier for storage hardware and management software from different vendors to work together.
For storage managers, this means the ability to map logical unit numbers (LUNs) of storage. It also means the ability to reassign storage volumes to different applications and monitor the performance of storage and tape arrays from different vendors using both common management tools and a common set of storage management skills.
So, where is the real benefit of using a suite of SMI-S compliant software or hardware? According to Bill North, a storage analyst at market researcher IDC in Framingham, Mass., the users who will benefit most from SMI-S are those who have heterogeneous environments and use storage products from a variety of vendors.
"You won't need to learn a new set of management tools to perform basic storage discovery, topology mapping and (potentially) provisioning just because you have different manufacturers' storage or storage network components in your mix," North said. "You'll be able to go to a single console, using the same methods, and one learning curve, rather than three or four, to be able to manage different vendors' products."
Of course, this ideal world won't dawn until the specification is finalized and every vendor offers to support it.
Where does the standard stand?
The Storage Networking Industry Association (SNIA) who is developing the standard expects to certify the first products that comply with the SMI-S standard in October.
According to Roger Reich, chairman of the SNIA committee in charge of developing SMI-S, customers can expect to see the first compliant products coming to market by the end of 2003.
SNIA has already frozen Version 1.0 of the SMI-S specification, which covers basic storage area network (SAN) functions, such as identifying and monitoring the performance of array controllers, switches and host bus adapters, and the ability to add and remove zones and volumes.
The initial versions of SMI-S are focused on storage area networks (SANs) attached to Windows and Unix servers, where the need for interoperability is perceived to be the greatest. This is also where customers are more likely to buy new tools.
In Version 1.1, SNIA hopes to extend SMI-S support to include network-attached storage (NAS) products, storage over IP networks (as opposed to the Fibre Channel focus of Version 1.0), more detailed performance statistics, added support for policy-based administration and other functions.
No formal timetable has been announced for rollout of the Version 1.1 specification. Likewise, SNIA has no current timetable for SMI-S support of core mainframe storage technologies such as the ESCON (Enterprise Systems Connection) or FICON (Fiber Connectivity) connectivity interfaces.
As management tools in the open systems world become more mature and they can be linked more easily using SMI-S, there will be an obvious need to unify the resulting open systems storage management environment with existing mainframe management tools, says Clod Barrera, IBM's storage strategy director. He doesn't see that process being completed, though, for another five years.
Open systems vendors, developers eager to join the SMI-S push
IT managers who have seen other standards efforts fail can take heart from the fact that in today's slow economy, it's in the vendors' best interest to adopt common standards. For one thing, says North, software vendors who write to the SMI-S interface can spend more time building new features into their tools rather than figuring out multiple application programming interfaces (APIs) and Command Line Interfaces (CLI) from hardware vendors.
SNIA's Reich hopes that 60 percent of new products will comply with the specification in 2004, with that number rising to 100 percent in 2005.
Virtually every major storage vendor has pledged support for SMI-S, and some (such as EMC Corp. and Hewlett-Packard Corp.) have pledged to support SMI-S in all their products. SNIA has already launched an Interoperability Conformance Testing Program that has created test specifications and a testing regime that will verify products do indeed comply with the SMS-S 1.0 specification. HP has also announced a developer's program to make it easier for vendors to develop products that support SMI-S.
In June, SNIA completed "CIM-SAN-2" which demonstrated more than 150 points of interoperability between 35 vendor products using SMI-S Version 1.0. The industry group also says it has seen "tremendous" response to its SMI-S Developers Education Course, with 51 developers trained in sold-out classes in June and July and another 25 scheduled for training in August.
IBM, which also plays a crucial role as the dominant mainframe hardware and software vendor, is moving cautiously to support SMI-S. The company has announced a SMI-S-compliant API for its "Shark" Enterprise Storage Server (ESS), but IBM requires the storage management software accessing the ESS to run on AIX, Linux or Windows 2000.
Barrera says IBM has no plans to develop a native z/OS interface for the SMI-S standard because "there's plenty of work to do to get the SMI-S standard fully functional in the open systems environment" before doing the same for the mainframe.
Is it too soon to care about SMI-S for the mainframe? That depends on how urgently you need to integrate your open systems servers (and storage) with your mainframe environments. At the very least, says North, z/Series and System/360 customers should look for SMI-S compliance whenever they upgrade or replace their storage management tools or storage hardware.
With the right SMI-S-compliant hardware in place, he says, "You will be amazed, because someday you'll install that next generation of (storage management) software and suddenly you'll be able to manage all these new devices you've acquired."
That "someday" may not come for years, but "all enterprise-class customers have this desire for integrated (storage) management with a high degree of automation," says Barrera. "The first step is to herd all the open systems cats and dogs into this SMI-S management structure," he says. "The next step would then be to reconcile these two management worlds -- the mainframe world and the SMI-S world."
The sooner SMI-S takes hold -- and the sooner customers demand SMI-S compliance from mainframe storage and software vendors -- the sooner that unified view will become real.
About the author: Robert L. Scheier writes about storage frequently from Boylston, Mass. He can be reached at firstname.lastname@example.org.