English German

Prevent field failures and improve end-user experience

A principle of device design is that the quality of your device will not exceed that of its weakest component. Will your company name suffer if your product corrupts user files or performs sluggishly? The extent to which this will limit the success of your product requires in-depth analysis.
 

Nearly any competent engineer can plug in flash memory and make it work for the short term, but a well-designed embedded system also ensures data reliability. Flash is ideal for storing data in embedded environments, however maintaining data integrity over the expected lifetime for the system requires careful planning on the part of the designer.
 

Things you want the end-user to experience when using your embedded devices are:

  • Fast reaction times
  • Consistently fast mount and recovery times, even after power interruption
  • Certainty of secured user-data during sudden power loss
  • Extended flash media lifetime
 

The User’s Data—Not Just Data Structures—Are Preserved

In Reliance Nitro, a transaction happens on a disk-wide basis, and protects both user data (the files you’re writing) and the meta data (the data used to keep track of where information is stored on the disk). This is unique because other file systems protect only the metadata, which, while important, does not guarantee that all of the data is safe.

Reliance Nitro minimizes the data-at-risk because it is designed to never overwrite live data, preserving a “known good state”, even if it is possible that power is lost before a transaction is complete. Data in the process of being written to the disk between transaction point could be lost, but the amount of information at risk when power is suddenly removed in a system is configurable.

What if a design needs to update or overwrite one file in the file system, but only for the changes to be committed if the entire update was complete? With Tuxera’s file systems, engineers have the control to make this happen by disabling transactions before the file update, and transacting and re-enabling them afterwards. An example of this is a video file that is not good unless the complete file is saved, or potentially an in-field update of the entire OS image or configuration files for your application stored in the file system.

In Reliance Nitro, if power is lost before all of the files have been updated and committed to with a transaction point, none of the files changed since the last successful transaction point will be updated as far as the file system is concerned. When the system restarts, the file system will be in a “known good” state, but the new files that were being written when power was lost would not appear. The file system would be in the “known good state” it was before file save started.

Download the whitepaper “Data Integrity Challenges with Flash Memory” to learn more.

= seven - two