SSDs Are in the Cards
Flash cards are going to be very cost effective, and that makes it possible for an SSD purchase to fly under the radar
March 11, 2009
![Network Computing logo Network Computing logo](https://eu-images.contentstack.com/v3/assets/bltde8121fc52c5c8f3/blt3f3d0318f746b1c2/65a530e4187606040a1d8b8c/placeholder.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Now it's a category. PCI-Express-based solid-state disk technology first pioneered by FusionIO has now been joined by Texas Memory Systems with its announcement today of the RamSan-20, a 450-GB Flash-based PCIe card. I'll let you read today's press release for all the speed and feeds. To save you time, these solutions are fast and cheap, but where do they fit in? That's what Im going to examine.
These cards fly in the face of a "typical storage guy's" view of Utopia, the one where everything is shared on a Fibre Channel SAN. But in today's market, they may be one of the most logical inroads for SSD technology. In most cases, these cards are going to be the SSD equivalent of a direct-attached hard disk. So why do they make sense?
First and foremost, it is because they are going to be very cost effective. The RamSan-20 is going to retail for about $18,000 and the FusionIO products are in that same price category. This price makes it possible for an SSD purchase to fly under the radar.
For example, an IT manager in charge of the Oracle environment has been complaining about performance to the storage team for the last year or so and may very well decide that the best way to solve the problem is to take care of it himself. DBAs -- and especially Oracle guys -- really know their databases; they understand what hot files are, and they know how to relocate them. They also in many cases own the server hardware that the database happens to be running on. Them accessing that server, installing a sub-$20k SSD could go totally unnoticed especially when you compare the cost of various Oracle modules.
Even if you are not sure what the problem I/O files are or if they can't be eliminated, for many environments the mission critical data is less than 450 GBs per server and the entire data set could be placed in the card.Doing this also makes SSDs simple. Just install a card. There are no concerns if the network is fast enough -- you are now direct attached right onto the PCI-E bus of the server accessing the SSD. If you really did have a storage I/O problem, it is now likely eliminated. Kind of like cracking an egg with a sledgehammer.
Another obvious use case is for these cards to be used by storage OEMs, especially those that have distinct lines already drawn between the storage hardware and the storage software. Companies like DataCore, FalconStor, NetApp, and Compellent could with relative ease put these cards right into their storage controller head units. This would eliminate all of the I/O bandwidth-to-the-shelf concerns that I talk about in our "State of SSD Storage" presentation.
This certainly is not the end of shared SAN-based SSDs. For customers that have a SAN, the ability to share the SSD memory across multiple applications will remain appealing. But for one-off, quick solutions to a problem PCIe-based SSD solutions are hard to argue with.
InformationWeek Analytics has published an independent analysis of data center strategies. Download the report here (registration required).
— George Crump is founder of Storage Switzerland , which provides strategic consulting and analysis to storage users, suppliers, and integrators. Prior to Storage Switzerland, he was CTO at one of the nation's largest integrators.6668
You May Also Like
Maximizing cloud potential: Building and operating an effective Cloud Center of Excellence (CCoE)
September 10, 2024Radical Automation of ITSM
September 19, 2024Unleash the power of the browser to secure any device in minutes
September 24, 2024