Compression sizes

By default, Sonda displays the uncompressed sizes of assets and the individual source files that make up each asset. However, you can enable the gzip and brotli options to view the compressed sizes as well.

Accuracy

Asset sizes

Modal showing asset data also contains the estimated download times using slow 3G

The GZIP and Brotli sizes shown in the assets modal are close to the actual sizes you would see when serving files from a server with GZIP or Brotli compression enabled. While the exact sizes may vary slightly depending on your specific compression settings, these differences are typically minimal.

Source sizes

The GZIP and Brotli sizes shown for individual source files are estimates, not exact measurements. This is because compression algorithms such as GZIP and Brotli are more efficient when applied to larger data sets. Larger assets have more opportunities to identify and compress repeated patterns, while smaller, isolated files tend to compress less effectively.

To estimate the compressed size of each source file:

1. Each source file is individually compressed using GZIP and Brotli to calculate its 'worst case' compressed size.
2. The sizes of all individually compressed files are summed.
3. The sum is scaled down to match the actual compressed size of the whole asset, and the difference is redistributed proportionally across all the files.

Example

Consider an asset that consists of two source files. When compressed individually, each file weighs 500 bytes, for a total of 1000 bytes. However, when the entire asset is compressed, the result is 800 bytes due to shared patterns. The 200 bytes of compression benefit are then shared proportionally so that the displayed compressed size of each file is 400 bytes. This ensures that the sum of the individual sizes equals the total compressed asset size.

While this method provides reasonable estimates, the sizes displayed for individual files may still differ slightly from their actual contribution to the compressed asset. The actual impact of each file depends on the patterns it shares with other files and how they collectively influence the compression process.