# Apache Spark (including Amazon EMR, Synapse Spark, and Google Cloud Dataproc) vendor-specific interface · access · 2025

Revision 1 (11 October 2026) · Modeled snapshot · through 8 October 2026.

All figures are research-informed estimates of analytical activity. Significant errors may remain; decimal places do not imply measurement precision.

2.475% of the full access layer.

[Apache Spark (including Amazon EMR, Synapse Spark, and Google Cloud Dataproc) vendor-specific interface](https://history.columnar.tech/technologies/access/vendor-spark.md): historical names and annual shares.

Includes Spark workloads on Amazon EMR, Synapse Spark, and Google Cloud Dataproc, alongside other hosted and self-managed Spark deployments. EMR workloads using Hadoop/Hive, Presto, and Trino are counted with those engines. Databricks and Fabric LH (Spark) have separate boxes. Spark Connect, introduced in Spark 3.4 in 2023, counts as direct Arrow access because it streams results to clients as Arrow batches. Internal Arrow use within other Spark workflows does not by itself put them in that category.

Connections below describe the whole Apache Spark (including Amazon EMR, Synapse Spark, and Google Cloud Dataproc) engine. Access and storage mixes are estimated separately; they are not a joint interface-to-storage distribution.

## Access use within this engine

| Option | Share of engine activity | Share of full access layer |
| --- | --- | --- |
| [Apache Spark (including Amazon EMR, Synapse Spark, and Google Cloud Dataproc) vendor-specific interface](https://history.columnar.tech/technologies/access/vendor-spark.md) | 37.5% | 2.5% |
| [Other language/portable APIs](https://history.columnar.tech/technologies/access/other-api.md) | 24.2% | 1.6% |
| [JDBC](https://history.columnar.tech/technologies/access/jdbc.md) | 19.9% | 1.3% |
| [ODBC](https://history.columnar.tech/technologies/access/odbc.md) | 7.3% | 0.5% |
| [Python DB-API (other drivers)](https://history.columnar.tech/technologies/access/python.md) | 6.3% | 0.4% |
| [Arrow without ADBC](https://history.columnar.tech/technologies/access/arrow.md) | 4.7% | 0.3% |
| [ADBC](https://history.columnar.tech/technologies/access/adbc.md) | 0.2% | <0.1% |

## Storage use within this engine

| Option | Share of engine activity | Share of full storage layer |
| --- | --- | --- |
| [Iceberg-authoritative tables](https://history.columnar.tech/technologies/storage/iceberg.md) | 26% | 1.7% |
| [Delta-authoritative tables](https://history.columnar.tech/technologies/storage/delta.md) | 23% | 1.5% |
| [Standalone Parquet on object storage](https://history.columnar.tech/technologies/storage/parquet-object.md) | 14% | 0.9% |
| [Hudi-authoritative tables](https://history.columnar.tech/technologies/storage/hudi.md) | 8% | 0.5% |
| [Standalone Parquet on HDFS](https://history.columnar.tech/technologies/storage/parquet-hdfs.md) | 7% | 0.5% |
| [Federated relational native stores (vendor unallocated)](https://history.columnar.tech/technologies/storage/remote-relational.md) | 6% | 0.4% |
| [Standalone ORC on HDFS](https://history.columnar.tech/technologies/storage/orc-hdfs.md) | 4% | 0.3% |
| [HDFS text/SequenceFile/RCFile/Avro](https://history.columnar.tech/technologies/storage/hdfs-row.md) | 3% | 0.2% |
| [Object-store CSV/text/Avro](https://history.columnar.tech/technologies/storage/object-row.md) | 3% | 0.2% |
| [Local/shared row files](https://history.columnar.tech/technologies/storage/legacy-local.md) | 2% | 0.1% |
| [Standalone ORC on object storage](https://history.columnar.tech/technologies/storage/orc-object.md) | 2% | 0.1% |
| [Standalone Parquet on local/shared filesystems](https://history.columnar.tech/technologies/storage/parquet-local.md) | 2% | 0.1% |

[All percentages for 2025](https://history.columnar.tech/years/2025.md)

---

[Markdown overview](https://history.columnar.tech/index.md) · [Interactive visualization](https://history.columnar.tech/?2025) · [Methodology](https://history.columnar.tech/methodology.md) · [Terms of use](https://history.columnar.tech/terms.md)

© 2026 Columnar Technologies Inc.