I'm creating a (modern) data warehouse in redshift. All of our infrastructure is hosted at Amazon. So far, I have setup DMS to ingest data (including changed data) from some tables of our business database (SQL Server on EC2, not RDS) and store it directly to S3.
Now I must transform and enrich this data from the S3 before I can write it to Redshift. Our DW have some tables for facts and dimensions (star schema), so, imagine a Customer dimension, it should contain not only the customer basic info, but address info, city, state, etc. This data is spread amongst a few tables in our business database.
So here's my problem, I don't have a clear idea of how to query the S3 staging area in order to join these tables and write it to my redshift DW. I want to do it using AWS services like Glue, Kinesis, etc. i.e. full serverless.
Can Kinesis accomplish this task? Would it make things easier if I moved my staging area from S3 to Redshift since all of our data is highly relational in nature anyway? If so, the question remains, how to transform/enrich data before saving it on our DW schemas? I have searched everywhere for this particular topic but information on it is scarse.
Any help is appreciated.
There are a lot of ways to do this but one idea is to query the data using Redshift Spectrum. Spectrum is a way to query S3 (called an external database) using your Redshift cluster.
Really high-level, one way to do this would be to create a Glue Crawler job to crawl your S3 bucket, which creates the External Database that Redshift Spectrum can query.
This way, you don't need to move your data into Redshift itself. Likely, you will want to keep your "staging" area in S3 and only bring into Redshift the data that is ready to be used for reporting or analytics, which would be your Customer Dim table.
Here is the documentation to do this: https://docs.aws.amazon.com/redshift/latest/dg/c-getting-started-using-spectrum.html
To schedule the ETL SQL: I don't believe there is a scheduling tool built into Redshift but you can do that in a few ways:
1) Get an ETL tool or set up CRON jobs on a server or Glue that schedules SQL scripts to be ran. I do this with a Python script that connects to the database then runs the SQL text. This would be a little bit more of a bulk operation. You can also do this in a Lambda function and have it be scheduled on a Cloudwatch trigger which can be on a cron schedule
2) Use a Lambda function that runs the SQL script that you want that triggers for S3 PUTs into that bucket. That way the script will run right when the file drops. This would be basically a realtime operation. DMS drops files very quickly so you will have files dropping multiple times per minute so that might be more difficult to handle.
One option is to load the 'raw' data into Redshift as 'staging' tables. Then, run SQL commands to manipulate the data (JOINs, etc) into the desired format.
Finally, copy the resulting data into the 'public' tables that users query.
This is a normal Extract-Load-Transform process (slightly different to ETL) that uses the capabilities of Redshift to do the transform.
Related
We are exploring usecases where we want to achieve in-place transformation and querying of S3 data lake data. We don't want to provision database and create tables (so we are not keen to consider Redshift or Athena) and we want the querying to be most cost-efficient. While we can use S3 Select to directly query S3 data, it has its own limitations such as we can query single object using S3 Select, etc. Are there any alternatives to achieve this? Please guide
I have daily MySQL DB snapshots stored on S3. This daily DB snapshot is a backup of 1000 tables in our DB, using mysqldump, size is about 300M daily (stored 1 year of snapshots, which is about 110G).
Now we want to load these snapshots daily to snowflake for reporting purpose. How do we create tables in snowflake? Shall we create 1000 tables? Will snowflake be able to handle this scenario?
All comments are welcome. Thanks!
One comment before I look at possible solutions: your statement "Our purpose is to avoid creating dimension or fact tables (typical data warehouse approach) to save cost at the beginning" is the sort of thinking that can get companies into real trouble. Once you build something and start using it, in 99% of cases you will be stuck with it - so not designing a proper, supportable, reporting solution (whether it is a Kimball model or something else) from the start is always a false economy. If you take a "quick and dirty" approach now you will regret it in a year's time.
With that out of the way, there seem to be 2 issues you need to address:
How to store your data
How to process your data (to produce you metrics and whatever else you want to do with it)
Data Storage
(Probably stating the obvious) Any tables that you create to hold metrics or which will be accessed by BI tools (including direct SQL) I would hold in Snowflake - otherwise you wont get the performance that Snowflake can deliver and there is little point using Snowflake - you might as well be using Athena directly against your S3 buckets.
For your source tables (currently in S3), in an ideal world I would also copy them into Snowflake and treat S3 as your staging area - so once the data has been copied from S3 to Snowflake you can drop the data from S3 (or archive it or do whatever you want to it).
However, if you need the S3 versions of the data for other purposes (and so can't delete it once it has been copied to Snowflake) then rather than keep duplicate copies of the data you could create External Tables in Snowflake that point to your S3 buckets and don't require you to move the data into Snowflake. Query performance against External Tables will be worse than if the tables were within Snowflake, but performance may be good enough for your purposes - especially if they are "just" being used as data sources rather than for analytical queries.
Computation
There are a number of options for the technologies you use to calculate your metrics - which one you choose is probably down to your existing skillset, cost, supportability, etc.
Snowflake functionality - Stored Procedures, External Functions (still in Preview rather than GA, I believe), etc.
External coding tools: anything that can connect to Snowflake and read/write data (e.g. Python, Spark, etc.)
ETL/ELT tool - probably overkill for your specific use case but if you are building a proper reporting platform that requires an ETL tool then obviously you could use this to create your metrics as well as move your data around
Hope this helps?
I have huge data from different DB sources ( Oracle, Mongo, Cassandra ) and also eventing data available in Kafka. Using Tableau for analytics and facing performance issue with huge data. So, planning to store data in some other way and use Tableau for visualization also. Have multiple options now and need some help to finalize the approach.
Option 1:-
Read DB data and store them in Parquet file and then expose it over Spark SQL or HiveQL or Presto SQL and let Tableau connect to this SQL.
Option 2:-
Read DB data and store them in Parquet file in S3 and then use AWS Athena for analytics and let Tableau connect to Athena.
Option 3:-
Read DB data and store them in Parquet file in S3 and then move to Redshift for analytics and let Tableau connect to Redshift.
Not sure if any of the above approach will be a good solution for streaming data( Kafka ) analytics as well.
Note:- I have multiple big tables and need joins b/w them.
I understand you have huge data from different sources, and you also have access to AWS. Then, you plan to use this data for analytics and dashboarding via Tableau.
Option 1 and 2
Your Options 1 and 2 are basically the same, as AWS Athena and Hive are based on the same principle of creating tables over flat files via a metastore which stores table definition. Both Athena's Presto engine and Spark are distributed and highly efficient on huge data (TB data). The main difference is the pricing model (Athena is based on price per data processed per request and is serverless, whereas Spark may imply infrastructure cost).
Then, both options may not perform well as they are not OLAP systems designed for self service BI (they are better use for ad hoc queries over huge data regarding).
Then, you may have trouble in managing your data model using flat files and table or views over them (data storage and compression won't be optimized for each table which may impact Tableau performance).
Option 3
Option 3 is better as it is based on Redshift which is designed to support OLAP system. You can connect Tableau directly to Redshift but you'll suffer from latency and you may have trouble managing your cluster load depending on the number of users and/or requests. But it can work the way you describe it.
Then, if you have performance issues, you'll be able to create data source extracts from Redshift to Tableau later on. You can also implement an intermediate database to store pre-aggregated queries (= datamarts) and connect Tableau directly to it which will avoid performing the same query on Redshift each time a dashboard is opened in Tableau (in that case Redshift also caches queries).
Then, as you need to perform multiple joins, you'll be able to optimize data storage for such queries using Redshift by setting the right partition and sort keys.
To conclude, you can also directly access flat files from Redshift using Redshift Spectrum (via Athena/Glue metastore).
Documentations:
https://docs.aws.amazon.com/redshift/latest/dg/best-practices.html
https://aws.amazon.com/fr/athena/pricing/
It seems dbt only works for a single database.
If my data is in a different database, will that still work? For example, if my datalake is using delta, but I want to run dbt using Redshift, would dbt still work for this case?
To use dbt, you need to already be able to select from your raw data in your warehouse.
In general, dbt is not an ETL tool:
[dbt] doesn’t extract or load data, but it’s extremely good at transforming data that’s already loaded into your warehouse. This “transform after load” architecture is becoming known as ELT (extract, load, transform). dbt is the T in ELT. [reference]
So no, you cannot use dbt with Redshift and Deltalake at the same time. Instead, use a separate service to extract and load data into your Redshift cluster — dbt is agnostic about which tool you use to do this.
There is a nuance to this answer - you could use dbt to select from external files in S3 or GCS, so long as you've set up your data warehouse to be able to read those files. For Redshift, this means setting up Redshift Spectrum. (For Snowflake, this means setting up an external table and on BigQuery, you can also query cloud storage data)
So, if the data you read in Deltalake lives in S3, if you set up your Redshift cluster to be able to read it, you can use dbt to transform the data!
You can use Trino with dbt to connect to multiple databases in the same project.
The Github example project https://github.com/victorcouste/trino-dbt-demo contains a fully working setup, that you can replicate and adapt to your needs.
I would say that DBT doesn't have an execution engine, so you can not use it to move data from one source to another as it isn't processing data itself, it only sends the SQL commands to the database.
In any case, if you want to move data from S3 to Redshift, maybe you could use Redshift Spectrum where you can query S3 as external tables. There you'll be able to use DBT on S3 and Redshift data from the same system.
#willie Chen the short answer is yes you can. The more accurate answer that is not the intent of dbt. As a tool it is intended for the transform part of ETL. It serves as a transform that is already existing in a data warehouse. I agree that you should use Redshift Spectrum for ETL.
Luther
I have huge CSV files in the zipped format in S3 storage. I need just a subset of columns from the data for Machine learning purposes. How should I extract those columns into EMR then to Redshift without transferring the whole files?
My idea is to process all files into EMR then extract subset and push the required columns into Redshift. But this taking a lot of time. Please let me know if there is an optimized way of handling this data.
Edit: I am trying to automate this pipeline using Kafka. Let say a new folder in added into S3, it should be processed in EMR using spark and stored into redshift without any manual intervention.
Edit 2: Thanks for input guys, I was able to create a pipeline From S3 to Redshift using Pyspark in EMR. Currently, I am trying to integrate Kafka into this pipeline.
I would suggest:
Create an external table in Amazon Athena (An AWS Glue crawler can do this for you) that points to where your data is stored
Use CREATE TABLE AS to select the desired columns and store them in a new table (with the data automatically stored in Amazon S3)
Amazon Athena can handle gzip format, but you'll have to check whether this includes zip format.
See:
CREATE TABLE - Amazon Athena
Examples of CTAS Queries - Amazon Athena
Compression Formats - Amazon Athena
If the goal is to materialise a subset of the file columns in a table in Redshift then one option you have is Redshift Spectrum, which will allow you to define an "external table" over the CSV files in S3.
You can then select the relevant columns from the external tables and insert them into actual Redshift tables.
You'll have an initial cost hit when Spectrum scans the CSV files to query them, which will vary depending on how big the files are, but that's likely to be significantly less than spinning up an EMR cluster to process the data.
Getting Started with Amazon Redshift Spectrum