Our team is trying to create an ETL into Redshift to be our data warehouse for some reporting. We are using Microsoft SQL Server and have partitioned out our database into 40+ datasources. We are looking for a way to be able to pipe the data from all of these identical data sources into 1 Redshift DB.
Looking at AWS Glue it doesn't seem possible to achieve this. Since they open up the job script to be edited by developers, I was wondering if anyone else has had experience with looping through multiple databases and transfering the same table into a single data warehouse. We are trying to prevent ourselves from having to create a job for each database... Unless we can programmatically loop through and create multiple jobs for each database.
We've taken a look at DMS as well, which is helpful for getting the schema and current data over to redshift, but it doesn't seem like it would work for the multiple partitioned datasource issue as well.
This sounds like an excellent use-case for Matillion ETL for Redshift.
(Full disclosure: I am the product manager for Matillion ETL for Redshift)
Matillion is an ELT tool - it will Extract data from your (numerous) SQL server databases and Load them, via an efficient Redshift COPY, into some staging tables (which can be stored inside Redshift in the usual way, or can be held on S3 and accessed from Redshift via Spectrum). From there you can add Transformation jobs to clean/filter/join (and much more!) into nice queryable star-schemas for your reporting users.
If the table schemas on your 40+ databases are very similar (your question doesn't clarify how you are breaking your data down into those servers - horizontal or vertical) you can parameterise the connection details in your jobs and use iteration to run them over each source database, either serially or with a level of parallelism.
Pushing down transformations to Redshift works nicely because all of those transformation queries can utilize the power of a massively parallel, scalable compute architecture. Workload Management configuration can be used to ensure ETL and User queries can happen concurrently.
Also, you may have other sources of data you want to mash-up inside your Redshift cluster, and Matillion supports many more - see https://www.matillion.com/etl-for-redshift/integrations/.
You can use AWS DMS for this.
Steps:
set up and configure DMS instance
set up target endpoint for redshift
set up source endpoints for each sql server instance see
https://docs.aws.amazon.com/dms/latest/userguide/CHAP_Source.SQLServer.html
set up a task for each sql server source, you can specify the tables
to copy/synchronise and you can use a transformation to specify
which schema name(s) on redshift you want to write to.
You will then have all of the data in identical schemas on redshift.
If you want to query all those together, you can do that by wither running some transformation code inside redsshift to combine and make new tables. Or you may be able to use views.
Related
I am trying to find a tool, or methodology to store when an update is done against an specific table and column in AWS Redshift.
In PostgreSQL there is a way of doing this with triggers, but Redshift does not support these triggers.
Can we monitor updates statements and store the timestamp, the old value, the new one, and the table affected?
There is no in-built capability in Amazon Redshift to do change detection.
Amazon Redshift is intended as a Data Warehouse, which typically means that bulk information is loaded from external sources. It should be relatively rare for data to be updated within Amazon Redshift because it is not intended to be used as an OLTP database.
Thus, it would be better to put change detection in the source database or in the ETL pipeline, rather than Redshift.
I have to connect multiple tables that are part of single or multiple databases. Approximately 10-15 tables in each query have to be connected to generate data for the analysis in SQL Server 2014.
I don't have access to the database diagram or architecture and these reports are to be sent out weekly. I want to understand the approach on how to begin writing these kind of queries which are of basic and advanced level and identify the relationship between tables and what kind of advanced level queries I can learn or utilize like CTE, Rank Partition, Subqueries etc.
Anybody who can provide a rough flow diagram or structure about the approach will be really helpful.
It's very unlikely that owners of those source systems want to be directly queried every time someone runs a report. Since you already have access to SQL Server, I would suggest building a data warehouse with that.
You haven't provided a whole lot of information to go on, but SSIS packages could be created to connect to the source systems and load into your data warehouse. And furthermore, those packages can be scheduled through Agent.
As for modeling... Again it is difficult with the lack of information, but generally the star model works great for reporting, which is a fact table surrounded by dimension (or attribute) tables.
As for figuring out relationships without a diagram, this will have to be done via experimentation and tieing to existing reports to make sure your joins aren't dropping records or cascading.
Good luck.
I'm a SSIS Developer. I do lots of SQL stored procedure lookup concepts in SSIS. But when coming to Azure Data Factory I haven't any idea how to perform a lookup using a SQL stored procedure.
Could anyone please guide me on this?
Thanks in advance !
Jay
Azure Data Factory (ADF) is more of an ELT tool rather than ETL, therefore direct lookups are not supported. Instead, this type of operation, along with other transforms is pushed down into the compute you are actually using. For example, if you are moving data to SQL Server, Azure SQL Database or Azure SQL Data Warehouse, you would ensure all data is on the same server and use a Stored Procedure task to execute the lookups using T-SQL and joins. If you are using Azure Data Lake Analytics (ADLA) you would use the U-SQL Activity to run U-SQL or execute ADLA stored procedures, again doing lookups via joins or custom U-SQL code such as Combiner, Applier, Reducer. In fact you can use any of the ADF compute options like SQL, HDInsight (including Hive, Pig, Map Reduce, Streaming and Spark script), Machiine Learning or custom .net activities.
So you need to think about things differently with ADF. Have a look through this article to gain greater understanding of transforming data in ADF:
Transform data in Azure Data Factory
https://learn.microsoft.com/en-us/azure/data-factory/data-factory-data-transformation-activities
As an aside, I would rarely use Lookups in SSIS as performance in early versions used to be poor. Although this has been improved in later versions, generally if you can do it in SQL you probably should. This pattern harnesses the power of SQL Server, rather than dragging data up into the SSIS pipeline, eg for the purposes of lookups (which are essentially joins) and pushing the data back out again. I reserve Data Flow transformations mainly when non-relational data is involved, eg xml or joining your email server with relational data. This is my personal view anyway : )
I'm currently trying to build a data flow in SSIS to select all records from a mapping table where an ID column exists in the related Item table. There are two complications:
The two tables are currently in different databases on different servers.
The databases are in Azure, for which I've read Linked Servers are not supported.
To be more clear, the job to migrate data from Staging environment to Production. I only want to push lookup records into prod if the associated Item IDs are in there. Here's some psudo-TSQL to give a clear goal of what I'm trying to achieve:
SELECT *
FROM [Staging_Server].[SourceDB].[dbo].[Lookup] L
WHERE L.[ID] IN (
SELECT P.[Item]
FROM [Production_Server].[TargetDB].[dbo].[Item] P
)
I haven't found a good way to create this in SSIS. I think I've created a work-around that involves sorting both tables and performing a merge join, but sorting both sides is an unnecessary hit on performance. I'm looking for a more direct and intuitive design for this seemingly simple data flow.
Doing this in a data flow, you'd have your Source query, sans filter, fed into a Lookup Component which is the subquery.
The challenge with this is SSIS is likely on-premises so that means you are going to pull all of your data out of Stage Azure to the server running SSIS and push it back to the Prod Azure instance.
That's a lot of network activity and as I'm reading the Azure pricing guide, I guess as long as you have the appropriate DTUs, you'd be fine. Back in the day, you were charged for Reads and not Writes so the idiom was to just push all your data to target server and then do the comparison there, much as ElendaDBA mentions. Only suggestion I'd make on the implementation is to avoid temporary tables or ad-hoc creation/destruction of them. Just implement as a physical table and truncate and reload prior to transmission to production.
You could create a temp table on staging server to copy production data into. Then you could create a query joining those two tables. After SSIS package runs, you could delete the temp table on staging server
I'm faced with needing access for reporting to some data that lives in Oracle and other data that lives in a SQL Server 2000 database. For various reasons these live on different sides of a firewall. Now we're looking at doing an export/import from sql server to oracle and I'd like some advice on the best way to go about it... The procedure will need to be fully automated and run nightly, so that excludes using the SQL developer tools. I also can't make a live link between databases from our (oracle) side as the firewall is in the way. The data needs to be transformed in the process from a star schema to a de-normalised table ready for reporting.
What I'm thinking about is writing a monster query for SQL Server (which I mostly have already) that will denormalise and read out the data from SQL Server into a flat file using the sql server equivalent of sqlplus as a scheduled task, dump into a Well Known Location, then on the oracle side have a cron job that copies down the file and loads it with sql loader and rebuilds indexes etc.
This is all doable, but very manual. Is there one or a combination of FOSS or standard oracle/SQL Server tools that could automate this for me? the Irreducible complexity is the query on one side and building indexes on the other, but I would love to not have to write the CSV dumping detail or the SQL loader script, just say dump this view out to CSV on one side, and on the other truncate and insert into this table from CSV and not worry about mapping column names and all other arcane sqlldr voodoo...
best practices? thoughts? comments?
edit: I have about 50+ columns all of varying types and lengths in my dataset, which is why I'd prefer to not have to write out how to generate and map each single column...
"The data needs to be transformed in the process from a star schema to a de-normalised table ready for reporting."
You are really looking for an ETL tool. If you have no money in the till, I suggest you check out the Open Source Talend and Pentaho offerings.