One of the more welcome aspects of the new enterprise-class solid state disks (SSDs) hitting the channel is how much is already known about the technology. SSDs have a fairly lengthy history in cell phones, music players and laptops, so there is a lot of built-up knowledge as to their characteristics and performance.

Yet, there are a lot of unknowns as well. How will the technology measure up in enterprise settings? How trustworthy are they with valuable or even critical data? And before most enterprises even get to that point, there is a more fundamental concern on the table: What is the best way to deploy these devices in existing infrastructures? Or more to the point, how are they to be integrated into storage environments consisting primarily of magnetic, mechanical disks? After all, no one is seriously planning a complete overhaul to SSD in one shot.

To devise an effective deployment plan, it helps to know what SSDs can and cannot do as well as hard drives. Rich Vignes, senior manager of market development at Seagate, says the overriding attributes in determining how SSDs can benefit the enterprise are power consumption and performance .

“We see two main applications for SSDs: as boot devices for blade servers and as very high-performance devices that can be used for portions of databases,” he said.

For boot devices, power consumption is clearly in SSD’s favor. While most spinning disks consume between 4 and 8 watts of power, SSDs average just 1 or 2 watts. However, the overall savings may not be that dramatic considering storage typically consumes less than 10 percent of a standard blade’s total power budget.

The real action, he says, is when you start to apply SSDs to database loads. That’s where performance issues start to take center stage. Because SSDs are still far too expensive to deploy on a mass scale, Vignes recommends a more targeted approach that takes advantage of SSDs’ I/O superiority and hard disks’ more favorable cost per GB.

“It is reasonable to create some portion of a storage array out of SSDs – typically 5 percent, or in heavy data environments as much as 10 percent,” he says. “The remainder is different types of rotating media.”

With this configuration, you can devote the faster SSDs to the most frequently accessed data, using the slower media for bulk storage. This approach also plays into SSDs’ strengths in processing random data.

“When you’re talking about database performance, the SSD offers as much as a 10-fold increase in random IOPS performance over rotating disks,” Vignes says. “That’s what system designers are really interested in: a high-performance database device for environments when not all data is being accessed at the same frequency.”

Despite solid state’s faster performance, there is some concern regarding their durability under heavy load factors. The type of application using flash storage may have a major impact on how long your new devices last if they are, say, constantly dealing with maximum data loads.

The kind of SSD you choose will also have an effect on performance and durability. The more common type of flash, single-level cell (SLC) is less dense than the more advanced multi-level cell (MLC) variety, but it provides a greater MTBF. Sun and Micron, in fact, recently unveiled a new SLC design said to provide more than 1 million write cycles, nearly a 10-fold improvement over current systems.

It is also very likely that the solid state disk itself will only be the first wave of a flash-based invasion of enterprise storage systems. Most vendors are placing solid state technology onto familiar 2.5- and 3.5-inch form factors so they can be easily deployed into the data center. But there’s no reason the underlying solid state technology couldn’t find its way into a wide variety of hardware platforms.

Mark Peters, senior analyst at the Enterprise Strategy Group, says the possibility exists for solid state technology to put an end to external storage systems altogether.

“You can take solid state technology and boost the cache in the storage controller or somewhere else in the overall stack,” he says. “We don’t have hard disks and tiering and all the other things in a storage hierarchy just for fun. They are there because we haven’t got some magic uber-storage device. If you can put storage the size of a pinhead directly on the server, why wouldn’t you?”

Another pertinent question in all this activity is how it will affect the need for traditional spinning media, particularly the high-speed 3.5-inch disks that have at once been the workhorse of the data center but also a major consumer of energy.

Traditionally, whenever an enterprise requires speed over capacity, the practice is to short-stroke the drive, that is, to allocate space only on the outer rim of the platter so the read/write head can access it quickly. The obvious drawback is that there is a tremendous amount of capacity on the disk that goes unused, particularly on the larger devices.

Certainly, a 15k, 3.5-inch disk that is being short-stroked is a prime candidate for replacement by an SSD. But it would be a mistake to think that SSDs are going to result in wholesale replacement of traditional storage, even if the price comes down as expected.

“We see the world needing both types of storage,” says Seagate’s Vignes. “They are complementary to each other. It’s all about performance and making that choice for a solid state drive. If your usage patterns are more toward bulk storage or average workloads, the hard disks are still the right solution.”

Far from being a simple replacement solution, solid state needs to be viewed more as an enhancement to traditional storage. Under the right circumstances, it can dramatically increase the speed, flexibility and responsiveness of your storage architecture. But it’s not a cure-all. If you fail to understand how to leverage the technology’s performance advantages by deploying SSDs strictly to boost capacity, you’ll likely be in worse shape than you were before.