January 18, 2013
Sometimes vendors come out with products I just don't see the need for. The latest example came in the form of an announcement from DISUK that its Paranoia 3 encryption appliance now supports the LTFS tape format.
While I strongly recommend encrypting data stored on tapes, especially if those tapes ever leave your data center, I don't see why storage administrators would invest in an external encryption appliance to do so. After all, LTFS is only supported on LTO 5 and later tape drives, which themselves include encryption capabilities. Organizations would be better off using the native encryption chip in the tape drive.
- Forrester Study: The Total Economic Impact of VMware View
- Industrialization of Business Operations in the Insurance Industry
- State of Cloud 2011: Time for Process Maturation
- Research: Federal Government Cloud Computing Survey
There are a few reasons for this. For one, an external appliance will cost several thousand dollars more than using the encryption chip that came with the tape. Second, adding an external appliance complicates the infrastructure. It means the tape is encrypted in a form that requires the external appliance to read, which not only means administrators need to buy an additional encryption appliance for the disaster recovery site.
It also puts the data at the mercy of the appliance vendor. Ask anyone who bought encryption appliances from Kasten-Chase, Neoscale or Decru how much fun it is dealing with a warehouse full of tapes encrypted with an appliance whose vendor has discontinued support or gone out of business. I'd much rather deal with the key management issues that arise from using the encryption capabilities built into tape drives.
Patents on Multipath?
Safe Storage, an entity with which I'm not familiar, filed a patent infringement suit in Delaware late last year against several leading storage vendors, including Dell, HP and HDS, claiming that they infringed on a patent entitled "Apparatus for redundant interconnection between multiple hosts and RAID" originally issued to Electronics and Telecommunications Research Institute in Korea in December 2000.
My quick reading of the patent abstract leads me to believe that this patent would apply to any form of what we now call multipathing to a storage system with multiple controllers. The Korean institute may really have invented multipathing, and the industry may have adopted it as a standard in violation of this patent.
Whether that's true or not, I'm disturbed by the fact that an unknown entity, which apparently exists only for the purpose of bringing patent infringement suits, has decided to act many years after multipath became a standard operating procedure in the industry. Perhaps patents should be like trademarks, where the failure to defend a trademark for a period of years invalidates the protection, which would prevent suits like this.
Holographic Storage Is the Future--And Always Will Be
Solid 3-D storage media have been a mainstay of science fiction for decades. However, despite the millions of dollars researchers have spent over the years on holographic storage, no one has delivered a practical product. The last serious attempt was InPhase Technologies, which ran out of cash before delivering its promised 1.6-Tbyte Tapestry product in 2011. VC firm Acadia Woods Partners was an investor in InPhase and was left with some of InPhase's technology. The investors have put another $10.7 million into Akonia Holographics, which is now promising a 6-Tbyte holographic disk.
Even if Akonia gets its product out the door in 2014 or 2015, it will deliver capacities and performance comparable to the LTO-7 format tape systems expected to ship at about the same time. To gain market traction, a new technology like holographic storage from an unknown startup would have to deliver significantly better capacity, performance and/or longevity than an established format. Hitachi's quartz glass disks, promised for 2015, could solve the longevity problem by etching data into the incredibly stable fused quartz, but only at CD-like data densities.
I'm convinced all these technologies will turn out to be too little too late as the constantly growing amount of data we need to store will exceed the ability to optically encode it.