Estimate Azure Synapse costs for Dedicated SQL pools, Serverless SQL, Spark pools, and Data Integration pipelines.
Last verified: May 2026
Provisioned data warehouse with reserved compute (DWU)
Pay-per-query model at $5.00/TB processed. No charge when idle.
Big data processing, ML, and data engineering workloads
$1.00/1,000 runs + $1.00/1,000 orchestration
$0.25/DIU-hour
Azure Synapse Analytics is a unified analytics platform that brings together enterprise data warehousing (Dedicated SQL Pools), big data analytics (Spark Pools), serverless data exploration (Serverless SQL), and data integration pipelines into a single workspace.
Dedicated SQL Pools provide provisioned compute capacity (DWU) for predictable, high-performance queries on large datasets. Serverless SQL is pay-per-query with no infrastructure to manage, ideal for ad-hoc exploration and infrequent queries. Use dedicated pools for production workloads and serverless for exploration.
Spark Pools are best for big data processing, data engineering (ETL), machine learning model training, and complex transformations on large datasets. They support Python, Scala, SQL, R, and .NET. Auto-pause after idle time helps control costs.
Synapse replaces standalone Azure SQL Data Warehouse, HDInsight Spark, and Azure Data Factory with a unified platform. While individual services may be cheaper for single-purpose workloads, Synapse offers cost savings through shared metadata, unified security, and integrated pipelines.
Output will appear here...The estimator combines four cost components: Dedicated SQL pools (DWU level × hourly rate × hours, with pause time excluded), Serverless SQL (TB scanned × $5), Spark pools (node SKU × node count × hours), and Data Integration (DIU-hours for data movement + activity runs at $1 per 1,000). Each component is shown separately so you can identify the dominant cost.
The Azure Synapse Analytics Cost Estimator models monthly costs across Synapse's major components: Dedicated SQL pools, Serverless SQL queries, Apache Spark pools, and Data Integration pipelines. For Dedicated SQL, specify your DWU level and operating hours. For Serverless, enter expected data volumes scanned per query. Spark pool costs are based on node size, count, and execution hours. Data Integration pricing covers pipeline activity runs and data movement. This holistic estimator helps data engineering teams understand the total cost of a Synapse workspace before committing to a particular architecture.
Your data team's Synapse bill jumped from $3K to $18K last month. The estimator helps identify the breakdown: Dedicated SQL at $2,800 (expected), Spark pools at $14,000 (unexpected), Serverless SQL at $1,200, Pipelines at $200. Investigation reveals a runaway Spark notebook running 24/7 because someone forgot to set auto-pause on a new pool. You enable auto-pause at 30 minutes, kill the orphan session, and next month's bill returns to baseline.
Always pause Dedicated SQL pools when not in use. A DW100c idle pool costs ~$900/month — pausing it during nights and weekends can cut that to ~$300/month with zero data loss (storage is billed separately). The 1-2 minute resume time is acceptable for most analytics workloads.
Synapse Serverless SQL ($5/TB scanned) is great for ad-hoc exploration, but a single bad query (SELECT *) on a 10 TB Parquet dataset costs $50. Always use predicate pushdown filters and column projection in queries against external tables. Implement query cost limits on the workspace to prevent rogue queries.
Spark pool auto-pause is the difference between $50/month and $5,000/month. Always set auto-pause to 15-30 minutes for dev/test pools. Production pools running scheduled jobs should explicitly start and stop the pool in the pipeline rather than relying on auto-pause, which can leak hours of compute.
A Data Warehouse Unit (DWU) is a bundled measure of compute performance for Synapse Dedicated SQL pools. Higher DWU levels increase query parallelism and performance. You are billed per DWU-hour, and you can pause the pool when not in use to stop compute charges while retaining your data.
Synapse Serverless SQL charges per TB of data processed by your queries. There are no provisioned resources to manage. This pay-per-query model is excellent for ad-hoc exploration of data lake files but can become expensive for large recurring queries, which is when a Dedicated SQL pool may be more economical.
Was this tool helpful?
Disclaimer: This tool runs entirely in your browser. No data is sent to our servers. Always verify outputs before using them in production. AWS, Azure, and GCP are trademarks of their respective owners.