top | item 13072255

Amazon Athena: Query S3 Using SQL

176 points| polmolea | 9 years ago |aws.amazon.com

44 comments

order

danso|9 years ago

This looks very neat. I'm someone who deals with a lot of plaintext data from a variety of sources, and so I find using ack/grep and csvkit to be efficient enough for my purposes of exploration. I love using SQL and SQLite but rarely do it for "fun" -- that is, I'll use it when I've committed to building a project, but not for exploration. This seems like it could lighten the friction quite a bit.

If anyone from AWS is here: how is this used internally at Amazon?

ktamura|9 years ago

The real question to ask is, will Amazon contribute back to open source? Presto itself is plenty proven and scalable: after all, it was created at Facebook.

kermatt|9 years ago

"Amazon Athena uses Presto with ANSI SQL support and works with a variety of standard data formats, including CSV, JSON, ORC, and Parquet."

I wonder if this is essentially a Presto SaaS product?

jakozaur|9 years ago

Looks very similar to Google Big Query.

Even the pricing is same: $5 / TB of data scanned.

estefan|9 years ago

When I tried it it was slower than bigquery. Plus you've got to mess about creating hive schemas.

buremba|9 years ago

I hope that Amazon contributes back to the Presto community.

nimrody|9 years ago

Would be useful if AVRO files were supported. This was the data can also be imported into Redshift if needed (Redshift does support Avro).

Other formats are schema-less (JSON,CSV, etc.) or not supported by Redshift (ORC, Parquet). Perhaps less efficient for some queries (AVRO is not a columnar format) but still useful.

dhananjayc|9 years ago

Is it possible to connect Athena to existing Hive Metastore?

nodesocket|9 years ago

Anybody have an example of storing NGINX access logs and using Athena to search them?

neximo64|9 years ago

Any examples of queries and what this can do? S3 was file storage as far as i thought?

raghavsethi|9 years ago

Athena (Presto) supports standard ANSI SQL - you can query data that's stored in S3.

cdevs|9 years ago

$5 a terabyte jeebus...don't f that query up

bsg75|9 years ago

Just like with BigQuery, a carefully thought out partitioning scheme is critical, or your queries need to be carefully locked down to prevent excessive table scanning. I burned through my BigQuery trial credit fast, by not using partitions during a quick-and-dirty test.

nulagrithom|9 years ago

Wondering if I could use this like SQLite for Lambdas. I'd like to build some serverless apps, but the commitment to a monthly fee from DynamoDB puts me off. Could I use Athena to drive down my cost to zero as long as the app is unused?

asteadman|9 years ago

Note the 10MB minimum "charge" per Query. For small datasets under 10MB, you'd only get up to 200 Queries for the minimum billable $0.01. That would be a fairly small number of queries, so probably not that useful. Plus you'd have all kinds of issues regarding consistency if your data was dynamic (s3 is a blob store, not a database, normal s3 consistency guarantees still apply).

I'm confused though. The monthly fees for dynamodb only apply after you exceed the free tier, and for someone who is unable to commit to a monthly fee because they envision low usage, shouldn't the free tier be sufficient? (Honest question, I'm looking at using dynamodb, but comments like this make me think I'm missing something)

brilliantcode|9 years ago

DynamoDB is like $5 or $10 bucks a month? but I understand the need to keep it to a minimum.

Athena is really interesting and if it can be as it is advertised "Serverless SQL" then they've got a killer product in the pipes: A future where developers no longer need to spend time on scaling, configuring, maintaining, strategizing deployments but upload code and instantly begin reaping the benefits of serverless tech.

The only missing component that would be a killer feature is something that answers to Azure's Active Directory. It would be nice if we had serverless plug-and-play user authentication and access control that integrated with Lambda and Athena.

I'd imagine some sort of "RoR on Serverless" type of framework that will scaffold out CRUD, User Management & REST Api is going to be in the works as well.

The only potential downside I see at the moment for Serverless is the uncertainty surrounding cold boots, it will directly affect user experience. It's fine when you got enough traffic to keep things in the "warm" state but there needs to be no dead zone when the call to the API Gateway is taking many seconds waiting for Lambda function to fire.

balls187|9 years ago

Tried it twice, and it crashed big time.

balls187|9 years ago

Also gives me a 500 on US-WEST-2

asafm|9 years ago

I wonder why they haven't chose Apache Drill over Presto. Anyone knows?

intrasight|9 years ago

what does "point your data in S3" mean?

justinsaccount|9 years ago

Are you talking about this? you left out a word.

Simply point to your data in Amazon S3