GIS Analytics Cost Comparison

BigQuery vs Snowflake vs Wherobots vs PostGIS on the same workload.

Summary

#EngineMonthly costManaged?
1BigQuery~$400Yes
2Wherobots~$410Yes
3Hetzner PostGIS~$500No (self-hosted)
4Snowflake (Small WH)~$1,000Yes
5Cloud SQL PostGIS~$3,500Yes

Assumptions

Fixed scenario for all engines:

ParameterValue
Team size10 analysts, 6-8 hours/day
Queries per analyst15/day (analytical + batch)
Total queries150/day
Avg data scanned/query12 GiB (well-pruned spatial analytics)
Working days22/month
Data scanned/month~39.6 TiB
DatasetOverture Maps, ~7 TB compressed (Parquet)

BigQuery - ~$400/month

ComponentCalculationCost
Storage7 TB x $0.02/GB/month$140
Queries (on-demand)38.6 TiB x $6.25/TiB (first 1 TiB free)$241
Total~$400

See how BigQuery spatial queries look in practice: BigQuery map examples in Dekart.

Wherobots - ~$410/month

ComponentCalculationCost
Engine165 SU-hours x $1.50/SU-hour (US)$248
Storage (S3)7,000 GB x $0.023/GB$161
Total~$410

1 SU = 32 vCPU Spark/Sedona cluster. 150 queries/day x 3 min each = 7.5 SU-hours/day x 22 days = 165 SU-hours/month. $300/month minimum on Professional plan.

Hetzner PostGIS - ~$500/month

ComponentCalculationCost
ServerAX102: 16 cores, 128 GB DDR5, 2x 1.92 TB NVMe$115 (€104)
Additional NVMe2x 7.68 TB for ~10 TB Postgres data$205 (€186)
BackupsWAL-G to S3-compatible, ~7-10 TB$180
Total~$500

You own all ops: upgrades, PostGIS extensions, monitoring, failover, no SLA.

Snowflake - ~$1,000/month

ComponentCalculationCost
Storage7 TB x $23/TB$161
Compute264 credits x $3/credit (Enterprise)$792
Total~$1,000

Small WH = 2 credits/hour. 6 hours/day x 22 days = 132 hours x 2 = 264 credits. Medium WH doubles the cost.

See Snowflake spatial queries in action: Snowflake Kepler.gl map examples.

Cloud SQL PostGIS - ~$3,500/month

ComponentCalculationCost
Storage (SSD)10 TB x $0.22/GB (incl. indexes/bloat)$2,200
Compute8 vCPU / 32 GB RAM, 24x7$518
Backups10 TB x $0.08/GB$800
Total~$3,500

SSD storage pricing makes Cloud SQL brutal for large datasets. Built for OLTP, not planetary analytics.

Key takeaways

  • BigQuery is cheapest fully-managed. Pay per bytes scanned - well-clustered tables with partition pruning keep costs low.
  • Wherobots is similar cost, serverless, geo-specialized. Better for heavy spatial joins.
  • Hetzner is cheap raw metal, but you own everything.
  • Snowflake is ~2.5x more because you pay for warehouse time, not bytes.
  • Cloud SQL PostGIS is 8x more expensive than BigQuery at planet scale.

Pragmatic architecture: BigQuery or Wherobots for planetary analytics + small PostGIS for low-latency APIs.

Caveats

  • Published list prices, no committed-use discounts applied
  • Egress costs not included
  • HA/replication doubles PostGIS costs
  • Query pruning efficiency varies - these assume well-optimized spatial queries

Try it yourself

Edit this page on GitHub