Snowflake was one of the first decoupled storage and compute architectures, making it the first to have nearly unlimited compute scale and workload isolation, and horizontal user scalability. It runs on AWS, Azure and GCP, and while by default it is multi-tenant compute and data, it can run in a Snowflake VPC. But it only provides 1, 2, 4, … 128 node clusters with no choice of node sizes. To get the biggest nodes you need to choose the biggest cluster.
Athena is built on a decoupled storage and compute architecture, though it only provides and controls the compute part and does not manage ingestion or storage. It is also only on multi-tenant shared resources. If you are a Redshift customer you can use Redshift Spectrum, which is dedicated Athena deployed on up to 10x the number of Redshift nodes in your own VPC, for the same price as Athena.
Snowflake delivers strong scalability with its decoupled storage and compute architecture. But it is inefficient at scaling for certain queries that require larger nodes, because the only way to get larger nodes is with larger clusters. It is better suited for batch writes and upserts as well because it requires entire micro-partitions to be rewritten for each write. Snowflake also transfers entire micro-partitions over the network, which creates a bottleneck at scale.
Athena is a shared multi-tenant resource, which means each account needs to be throttled to protect every other account’s performance. One customer was unable to handle any table or join above 5 billion rows. By default Athena supports a maximum of 20 concurrent users. If scalability is a top priority, Athena is probably the wrong choice.
Snowflake is a perfect example of a modern storage and compute architecture that is not optimized for performance. Its data access is not optimized. Snowflake has no indexing for fetching the exact data it needs. It only keeps track of data ranges within each micro-partition, which range in size from 50 MB to 150 MB uncompressed, and can overlap. Whenever Snowflake does not have the data cached locally in the virtual warehouse, it has to fetch all of the micro-partitions that might have the data, which can take seconds or longer. While Snowflake does some query plan optimization, it does not show up in the performance of queries, which are 4-6000x slower than Firebolt in customer benchmarks. The two biggest explanations are a lack of indexing, and limited query plan optimization. However, Snowflake does provide result set caching across virtual warehouses in addition to SSD caching within each virtual warehouse. This does deliver solid performance for repetitive query workloads after the first query.
Athena, and Presto, should be the worst at performance, by design. The reason is that it sacrifices storage-compute optimization to get support for federated queries across multiple data sources. But there is a reason Presto is so popular. Even with that handicap, Presto and Athena do very well. Presto can come close to Redshift and Snowflake in performance when both Presto and the external storage is managed by experts. But there is no support for indexing. Specifically with Athena, you cannot guarantee performance as a shared multi-tenant resource. In general, if performance is a top concern and you can bring data together via a data pipeline and optimize data with compute, then Athena or Presto are not the best choice.
Snowflake has broader support for use cases beyond traditional reporting and dashboards. Its decoupled storage and compute architecture enables you to isolate different workloads to meet SLAs, and it also supports high user concurrency. But Snowflake also does not provide interactive or ad hoc query performance because of inefficient data access along with a lack of extensive indexing and query optimization. Snowflake also cannot support streaming or low latency ingestion below one minute ingestion intervals. All of these limitations exclude Snowflake from many operational use cases and most customer-facing applications that require second-level performance.
Athena is one of the best “one-off” query engines; all you have to do is provide the data and pay $5 a TB. If you need to quickly pull together multiple data sources, it’s a great option. Redshift Spectrum is a great add-on option for Redshift for federated queries. But if you don’t need federated queries, need performance, and need anything other than one-off or occasional analytics, Athena is not a good option for any of these use cases. There is no data, network or query optimization, no indexing beyond pruning indexes like others.
Snowflake as a more modern cloud data warehouse with decoupled storage and compute is easier to manage for reporting and dashboards, and delivers strong user scalability. It also runs on more than AWS. But like the others, Snowflake does not deliver sub-second performance for ad hoc, interactive analytics at any reasonable scale, or support continuous ingestion well. It is also often very expensive to scale, especially for large data sets, complex queries and semi-structured data.
Athena is arguably the easiest, least expensive and best suited for “one-off analytics”. But it is also the most limited, and requires you to manage your own (external) storage and ingestion very well, which is especially hard for continuous ingestion. This makes the least-suited for any ongoing, frequent use case.