This dataset provides global fixed broadband and mobile (cellular) network performance metrics in zoom level 16 web mercator tiles (approximately 610.8 meters by 610.8 meters at the equator). Data is provided in both Shapefile format as well as Apache Parquet with geometries represented in Well Known Text (WKT) projected in EPSG:4326. Download speed, upload speed, and latency are collected via the Speedtest by Ookla applications for Android and iOS and averaged for each tile. Measurements are filtered to results containing GPS-quality location accuracy.

Speedtest data is used today by commercial fixed and mobile network operators around the world to inform network buildout, improve global Internet quality, and increase Internet accessibility. Government regulators such as the United States Federal Communications Commission and the Malaysian Communications and Multimedia Commission use Speedtest data to hold telecommunications entities accountable and direct funds for rural and urban connectivity development. Ookla licenses data to NGOs and educational institutions to fulfill its mission: to measure, understand, and help improve connected experiences for all. Ookla hopes to further this mission by distributing the data to make it easier for individuals and organizations to use it for the purposes of bridging the social and economic gaps between those with and without modern Internet access.

Hundreds of millions of measurements are taken on Ookla's Speedtest platform each month. In order to create a manageable dataset, we aggregate raw data into tiles. The size of a data tile is defined as a function of "zoom level" (or "z"). At z=0, the size of a tile is the size of the whole world. At z=1, the tile is split in half vertically and horizontally, creating 4 tiles that cover the globe. This tile-splitting continues as zoom level increases, causing tiles to become exponentially smaller as we zoom into a given region. By this definition, tile sizes are actually some fraction of the width/height of Earth according to Web Mercator projection (EPSG:3857). As such, tile size varies slightly depending on latitude, but tile sizes can be estimated in meters.

For the purposes of these layers, a zoom level of 16 (z=16) is used for the tiling. This equates to a tile that is approximately 610.8 meters by 610.8 meters at the equator (18 arcsecond blocks). The geometry of each tile is represented in WGS 84 (EPSG:4326) in the tile field.

The tile aggregates start in Q1 2019 and go through the most recently completed quarter (Q2 2023). They will be updated shortly after the conclusion of the quarter.

Each tile contains the following adjoining attributes:

Quadkeys can act as a unique identifier for the tile. This can be useful for joining data spatially from multiple periods (quarters), creating coarser spatial aggregations without using geospatial functions, spatial indexing, partitioning, and an alternative for storing and deriving the tile geometry.

Two layers are distributed as separate sets of files:

• performance_mobile_tiles - Tiles containing tests taken from mobile devices with GPS-quality location and a cellular connection type (e.g. 4G LTE, 5G NR).
• performance_fixed_tiles - Tiles containing tests taken from mobile devices with GPS-quality location and a non-cellular connection type (e.g. WiFi, ethernet).

Layers are generated based on a quarter year of data (three months) and files will be updated and added on a quarterly basis. A /year=2020/quarter=1/ period, the first quarter of the year 2020, would include all data generated on or after 2020-01-01 and before 2020-04-01.

Data is subject to be reaggregated regularly in order to honor Data Subject Access Requests (DSAR) as is applicable in certain jurisdictions under laws including but not limited to General Data Protection Regulation (GDPR), California Consumer Privacy Act (CCPA), and Lei Geral de Proteção de Dados (LGPD). Therefore, data accessed at different times may result in variation in the total number of tests, tiles, and resulting performance metrics.

The Shapefile (SHP) is a widely-adopted format for sharing geospatial data, and is supported by nearly every GIS-capable software and library. There is a .zip file for each layer:

• 2020-04-01_performance_mobile_tiles.zip
• 2020-04-01_performance_fixed_tiles.zip

Each layer .zip unarchives to contain a directory of the same name and the necessary files. E.g. gps_mobile_tiles.zip unarchives with the following structure:

2020-04-01_performance_mobile_tiles.zip
+-- 2020-04-01_performance_mobile_tiles
|   +-- 2020-04-01_performance_mobile_tiles.dbf
|   +-- 2020-04-01_performance_mobile_tiles.prj
|   +-- 2020-04-01_performance_mobile_tiles.shp
|   +-- 2020-04-01_performance_mobile_tiles.shx

These files serve the following purpose:

• .dbf dBase file containing attributes
• .prj defines the spatial projection
• .shp contains the geometries
• .shx spatial index

The data is also available in Apache Parquet format. The available fields are described in the Attributes section. An additional field, tiles contains a Well Known Text (WKT) representation of the tile geometry. WKT is readable by many cloud tools with spatial extensions, such as Athena and Redshift.

The performance datasets are available via AWS S3 at the s3:https://ookla-open-data bucket, where individual Parquet time series and compressed Shapefiles are organized by

1. file format (shapefiles or parquet)
2. service type (fixed or mobile)
3. year (2020)
4. quarter (for example, 1 corresponds to the Q1 period starting 2020-01-01)

Individual downloads for fixed or mobile network performance aggregates (map tiles) for a given quarter can be located based on the following object key naming pattern:

s3:https://ookla-open-data/FORMAT/performance/type=TYPE/year=YYYY/quarter=Q/FILENAME

For example, to access all of the files for fixed and mobile service types for the second quarter of the year 2023, one would use the following S3 URIs:

Shapefiles:

• s3:https://ookla-open-data/shapefiles/performance/type=mobile/year=2023/quarter=2/2023-04-01_performance_mobile_tiles.zip
• s3:https://ookla-open-data/shapefiles/performance/type=fixed/year=2023/quarter=2/2023-04-01_performance_fixed_tiles.zip

parquet files:

• s3:https://ookla-open-data/parquet/performance/type=mobile/year=2023/quarter=2/2023-04-01_performance_mobile_tiles.parquet
• s3:https://ookla-open-data/parquet/performance/type=fixed/year=2023/quarter=2/2023-04-01_performance_fixed_tiles.parquet

Files can also be downloaded directly by clicking on the following URLs:

Esri Shapefiles:

• https://ookla-open-data.s3.amazonaws.com/shapefiles/performance/type=mobile/year=2023/quarter=2/2023-04-01_performance_mobile_tiles.zip
• https://ookla-open-data.s3.amazonaws.com/shapefiles/performance/type=fixed/year=2023/quarter=2/2023-04-01_performance_fixed_tiles.zip

Apache Parquet:

• https://ookla-open-data.s3.amazonaws.com/parquet/performance/type=mobile/year=2023/quarter=2/2023-04-01_performance_mobile_tiles.parquet
• https://ookla-open-data.s3.amazonaws.com/parquet/performance/type=fixed/year=2023/quarter=2/2023-04-01_performance_fixed_tiles.parquet

S3 objects can also be downloaded via the AWS CLI. See these instructions for installing AWS CLI v2.

Using the object keys described above, the following bash script downloads a shapefile (2023-04-01_performance_fixed_tiles.zip) for fixed performance tiles aggregated over Q2 2023 using aws s3 cp.

#!/usr/bin/env bash
export FORMAT='shapefiles' # (shapefiles|parquet)
export TYPE='fixed'        # (fixed|mobile)
export YYYY='2020'         # 2019,2020,2021,2022,2023
export Q='1'               # 1,2,3,4 (to date)
aws s3 cp s3:https://ookla-open-data/${FORMAT}/performance/type=${TYPE}/year=${YYYY}/quarter=${Q}/ . \
--recursive \
--no-sign-request

To download the 2023 Q2 mobile and fixed time series datasets, we can also specify the full S3 URI to download the objects:

#!/usr/bin/env bash
# Mobile 2023 Q2
aws s3 cp s3:https://ookla-open-data/parquet/performance/type=mobile/year=2023/quarter=2/2023-04-01_performance_mobile_tiles.parquet --no-sign-request
# Fixed 2023 Q2
aws s3 cp s3:https://ookla-open-data/parquet/performance/type=fixed/year=2023/quarter=2/2023-04-01_performance_fixed_tiles.parquet --no-sign-request

If using R, the ooklaOpenDataR package provides functions for accessing and working with the datasets.

Using R to analyze download speeds in Kentucky counties

Using Python to analyze download speeds in Kentucky counties

Using R to filter tiles by location with parquet

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

• 2023-02-23 Loaded latency values added as avg_lat_down_ms and avg_lat_up_ms. Loaded latency measures how network load impacts network responsiveness, or latency. These values are only available in Parquet format.

• 2023-07-01 Centroid coordinate values for tiles added as tile_x and tile_y.