FAQ: What are some best practices for raster storage?


What are some best practices for raster storage?


There are many different approaches for storing and managing raster data, and each approach offers different advantages and disadvantages. The optimal approach to use depends on the needs and the raster data in question.

The following are a few common considerations and objectives to be considered when selecting an approach for storing raster data:

  • Accessibility to data
  • Size of data
  • Functionality of data
Accessibility to data
The number of users that must have access the raster data should determine the type of geodatabase used to store the data. A personal or file geodatabase is more suited for single users or small workgroups that only have one writer, depending on the size of the data being stored. Such data can be hosted in a shared network drive, using UNC paths to the data for a few readers. An enterprise geodatabase hosted on a server is used when there are many users with writing privileges who need to access and modify the data.

For comparisons of the different types of geodatabases to store raster data, refer to the following documentation: How raster data is stored and managed.

If data accessibility is not an important factor, use a traditional folder-based approach. For more information, refer to the following documentation: Should I load my raster data into a geodatabase?
Size of data
The size of the data may determine the method of compression and the type of geodatabase to use (if needed). Each type of geodatabase has a size limit to store data, with the smallest being the personal geodatabase and the largest being the enterprise geodatabase.

The size of the raster data can be reduced through compression, but this reduces the accuracy of the data if the pixel values are altered. When balancing the trade-off between the size and accuracy of the raster data, consider either a lossless or lossy compression. For analyzing data, a lossless compression is recommended, while lossy compression is useful for improving the time taken to generate a dataset for display that does not require high precision.

For more information and comparison between lossless and lossy compression, refer to the following documentation: Raster compression.
When storing raster data in the cloud, there is another type of compression called Limited Error Raster Compression (LERC). For more information on this type of compression, refer to the following documentation: Storing large volumes of imagery and raster in the cloud.
Besides compression, raster dataset sizes can be reduced by down-sampling the dataset. Down-sampled data is used to improve display speed and performance. Using down-sampling creates a dataset that is 2 to 10 percent the size of the original dataset.

For more information, refer to the following documentation: Compression, pyramids, and tile size.
Functionality of data
Deciding whether the raster data has better performance or portability is also another factor when determining the desired raster data storage model. If portability is a priority, it is better to use the most basic data storage model - raster datasets, stored in a disk or a geodatabase. Raster datasets are useful for quick viewing or large raster data that are not modified often.

Mosaic datasets allow for better performance, especially when the raster data is expected to be queried, storing metadata, and overlapping raster data.

For more information and comparison between using raster datasets or mosaic datasets, refer to the following documentation: Comparing raster data storage models.

Related Information