Compute resources tend to be significantly more expensive than storage. Our approach is two-fold: - We finance compute resources required for you to ingest dat

Compute resources tend to be significantly more expensive than storage. Our approach is two-fold: - We finance compute resources required for you to ingest data into BigQuery. With Redshift, you pay for ingest directly via compute cluster consumption (again, more expensive than storage). This also increases your complexity due to on-cluster contention of resources between ingest and query. - Separation of storage and compute gives you lots of options. With BigQuery ,you don't need to attach relatively expensive compute just for the luxury of getting more storage. Spectrum helps somewhat, but ultimately with Redshift you don't even get to pay for storage - you pay for compute/storage combos. - BigQuery's Long-Term Storage is not an archival storage tier - it's a discount on storage, with identical performance and durability characteristics. At only $0.01 per GB per month. This is likely a result of origins of the two technologies. BigQuery is Dremel, written and operated by Google since 2006. Redshift purchased source code to an on-premise fork of Postgres.