Storage systems can present a pretty big roadblock to enterprises following the path to complete data center automation.
Server automation, when implemented correctly, can already provide big benefits to the companies that can afford it. That’s because the ability to automate the provisioning, configuring updating, patching, auditing and day-to-day management of hundreds or even thousands of servers can save vast amounts of time and remove the need for an army of technicians involved in tedious, repetitive and sometimes complex tasks. And this saves big bucks.
But compared to server automation, data center automation for storage is in its infancy. Although a few vendors, such as Onaro, Opsware and EMC, have developed fairly sophisticated storage automation products, adoption rates are far lower than for automation products targeted at servers.
At first glance, this seems strange, because the potential benefits of storage automation are huge.
“As infrastructure gets complicated, leveraging software to automate storage processes makes for tremendous efficiencies,” said Bob Laliberte, an analyst at Enterprise Strategy Group. “If you can reduce the time it takes to provision storage from 30 to 60 days to a much shorter period, you can reduce your time to market for a new application dramatically. There’s also a tremendous cost reduction, so storage automation is an inevitable destination.”
Given that both server and storage automation are highly desirable, why has the uptake of storage automation products fallen behind their server counterparts? The answer, put simplistically, is that it’s hard. And that’s what’s creating the roadblock.
The biggest problem has been the lack of standardization: the storage market is heterogeneous, and without standards it’s not possible to come out with a “one size fits all” solution. Systems have therefore been created by storage vendors for their own products, or by third parties for one or a limited number of other vendors’ products.
For data center managers who can afford it, both server automation and storage automation are great, but what would be, to use a Steve Jobsism, “insanely great,” is the ability to manage all their resources — compute, connectivity and storage — from a single unified console.
“You want to be able to say, ‘I need this new app running on a server with this much storage for performance, this much for security,’ and so on, and provision all of this automatically,” said Laliberte. “A unified system provides a holistic view of the infrastructure. You can’t manage what you can’t see.”
At the moment there are no such fully unified data automation systems, although they may not be too far over the horizon, Laliberte said. Such a system might come from a single vendor, or more likely from two or more vendors whose products are tuned to work together. For example, Onaro has a partnership with automation vendor BladeLogic, enabling BladeLogic’s product to activate some of Onaro’s storage management tools as part of its automation workflows.
Opsware also looks to be an exciting prospect, and there’s little doubt that the company will be releasing some interesting products (such as the recently announced System 7, which promises to go a long way toward unified automation) now that it is being acquired by HP.
Laliberte also highlights Cisco’s VFrame technology. Cisco’s Data Center 3.0 roadmap combines data center automation via the network with virtualization technology, with the goal, over the next couple of years, of a complete data center automation system using virtualization. In July, Cisco introduced its VFrame Data Center Appliance, which it described at the time as an orchestration platform that leverages network intelligence to provision resources together as virtualized services.
Virtualization Complicates Matters
There’s a further complicating factor to all this, and that’s virtualization. Storage virtualization has been around for years, while standard x86-based server virtualization is really only now coming into its own, thanks both to processor hardware support for virtualization, and the advanced virtual machine management systems that companies like virtualization leader VMware, a part of EMC, are supplying.
The problem is this: how do you automate the management of virtual machines and virtual storage? Ignoring storage for a moment, do you use server automation software from a server automation software vendor, or virtual machine management software from the hypervisor vendor? The first is likely to be more sophisticated, but how clearly will it be able to look inside the virtual machines to see what is actually going on?
On the storage side, the problem is automating virtualized storage devices from a variety of vendors. At present, products like IBM’s San Volume Controller (SVC) storage virtualization can be used with compatible devices from vendors such as IBM, EMC, HP and Hitachi, but the challenge then is linking the storage automation software with the physical (and virtual) server automation software it must work with.
For now, it’s likely that data center automation systems that unite server and storage views will come from technology (and therefore corporate) partnerships, aimed at companies which are primarily users of a single vendor’s hardware. The biggest gainers will be the largest companies with the highest number of servers, but paradoxically, they won’t be the ones who implement it first, since financial institutions and the like won’t want to be the early adopters. It’s the medium-sized companies which can afford it that will initially take the plunge, after a lengthy testing period, Laliberte predicted.
But one thing you can be quite sure about is that, given sufficient demand, suitable products will emerge and the roadblock will be removed. Laliberte concluded: “I think the solutions are bound to be vendor-based first — heterogeneous solutions will be difficult without a tremendous amount of work. Services like storage and server virtualization get rid of some problems and bring out new ones, but in the end, all these problems will be solved.”