Amazon Redshift provides SQL data warehousing with provisioned and serverless options.
Data warehouse workloads
Amazon Redshift is a managed cloud data warehouseA store used to analyze data with SQL and business intelligence tools. for SQL analytics and business intelligence. A data warehouse is a store organized for analyzing data.
How it works
- Choose a provisioned cluster for manual resource management or Redshift Serverless for automatically provisioned and scaled warehouse capacity.
- Redshift uses massively parallel processing (MPP): compute nodes run query segments on parts of the data. Redshift ML lets analysts create, train and run models with SQL, using SageMaker AI to create the model.
- Redshift Spectrum queries S3 files directly. The cluster and files must be in the same Region. Spectrum cannot use enhanced VPC routingA Redshift setting that routes COPY and UNLOAD traffic through a VPC..
- Redshift takes automated snapshots. Enable cross-Region snapshot copy to copy cluster snapshots to another Region for restoration if the primary Region fails.
Exam cues
- If the question says warehouse capacity without managing a cluster, think Redshift Serverless, because it provisions resources and scales capacity automatically.
- If the question says run machine learning models with SQL, think Redshift ML, because it lets analysts create, train and run models.
- If the question says recover a cluster after a Region failure, think cross-Region snapshot copy, because the copied snapshots can restore the cluster in another Region.
Remember
- A Spectrum cluster and its S3 files must be in the same Region.
- Enhanced VPC routing supports clusters and Serverless workgroups, but not Spectrum.
- Automated snapshots are taken about every eight hours or after 5 GB per node of changed data, whichever comes first.
Trap: Enhanced VPC routing supports clusters and Serverless workgroups, but it cannot be used with Spectrum.
Enhanced VPC routing sends COPY or UNLOAD traffic through the VPC.
An S3 endpoint provides a path to a bucket in the same Region.
Da documentação da AWS (19)
- Amazon Redshift is a managed cloud data warehouse that supports SQL-based tools and business intelligence applications.
- Redshift Serverless automatically provisions resources and scales warehouse capacity.
- Provisioned Redshift clusters are an alternative for manual resource management.
- UNLOAD writes query results to files in S3.
- Redshift Spectrum queries data directly from files in S3.
- A Redshift Spectrum cluster and its S3 data files must be in the same AWS Region.
- A Redshift external schema references an external database in a data catalog; an external table contains metadata referring to data outside the cluster.
- The AWS Glue Data Catalog stores technical metadata, including table descriptions, for use by Redshift Spectrum.
- Redshift clusters and Serverless workgroups support enhanced VPC routing.
- Redshift Spectrum cannot use enhanced VPC routing.
- An S3 VPC endpoint can route traffic to an S3 bucket in the same Region as the Redshift cluster or workgroup.
- An S3 gateway endpoint allows access to S3 from a VPC without an internet gateway or NAT device.
- A VPC is a logically isolated virtual network for AWS resources.
- Redshift uses massively parallel processing: several compute nodes each run the same compiled query segments on part of the data, which makes complex queries on large datasets fast.
- Redshift ML lets analysts create, train and run models with SQL; Redshift creates the model in SageMaker AI from the data it is given.
- Amazon Redshift can automatically copy automated and manual snapshots of a cluster to another AWS Region, so the cluster can be restored there if the primary Region fails.
- A provisioned Redshift cluster has a leader node that receives queries and builds execution plans, and compute nodes that run them in parallel.
- Redshift automated snapshots are taken about every eight hours or every 5 GB per node of changed data, whichever comes first.
- Provisioned RA3 (and RG) clusters support Multi-AZ deployments that run compute in two Availability Zones to keep working through a zone failure.
Loading, routing and availability
Amazon Redshift is a managed cloud data warehouseA store used to analyze data with SQL and business intelligence tools. for SQL and business intelligence tools. Its loading, query, network and deployment choices determine how it accesses data and handles an Availability Zone failure.
Redshift uses columnar storage: it stores values by column so analytical queries that read a subset of columns need fewer disk reads.
How it works
- Use COPY to load data from sources such as Amazon S3 into Redshift tables in parallel. Redshift needs permission to access the source.
- To query S3 files directly, use Redshift Spectrum. Give the cluster an IAM role for the external data catalogA store of descriptions of data such as table definitions. and S3 files, then create an external schemaA collection of external tables that references a database in an external data catalog. and external tableA table definition that lets Redshift query data stored outside the warehouse. before querying.
- Enable enhanced VPC routingA Redshift setting that routes COPY and UNLOAD traffic through a VPC. to send COPY and UNLOAD traffic between the cluster and data repositories through its virtual private cloudA logically isolated virtual network for AWS resources. (VPC). Configure a network path to the data resource so the commands can reach it.
- For a zone failure, a Multi-AZ deployment runs compute in two Availability Zones so the warehouse can keep working.
Exam cues
-
If the question says analytics scans many records but needs only a few columns, think Redshift columnar storage, because it reduces the data read from disk.
-
If the question says load S3 data into warehouse tables in parallel, think COPY, because it reads and loads data in parallel.
-
If the question says query S3 files directly, think Spectrum, because it uses an external schema and table to query those files.
-
If the question says control COPY and UNLOAD traffic through the VPC, think enhanced VPC routing, because it sends that traffic through the VPC.
-
If the question says keep a provisioned warehouse working through a zone failure, think Multi-AZ, because it runs compute in two Availability Zones.
Remember
- Spectrum needs an IAM role with access to the external catalog and S3 files.
- Enhanced VPC routing allows VPC controls and flow logs, but COPY and UNLOAD can fail without a network path to the data resource.
- Provisioned RA3 and RG clusters support Multi-AZ deployments.
Trap: COPY loads data into Redshift tables. Spectrum queries S3 files directly.
Da documentação da AWS (9)
- Amazon Redshift is a managed cloud data warehouse that supports SQL-based tools and business intelligence applications.
- Redshift Spectrum queries data directly from files in S3.
- Spectrum setup includes an external schema and external table before querying S3 data.
- A cluster needs an IAM role authorizing access to the external data catalog and S3 files for Spectrum.
- Enhanced VPC routing forces COPY and UNLOAD traffic between a cluster and data repositories through its VPC, allowing VPC controls and flow logs.
- With enhanced VPC routing, COPY and UNLOAD can fail if the VPC lacks a network path to the data resource.
- COPY reads and loads data in parallel into Redshift tables from sources including S3, subject to resource permissions.
- Provisioned RA3 (and RG) clusters support Multi-AZ deployments that run compute in two Availability Zones to keep working through a zone failure.
- Amazon Redshift stores table values by column, reducing disk reads for analytical queries that need only some columns.
Flashcards
Nada para revisar aqui agora. Escolha outros tópicos ou outro filtro.
·
Todos os 12 flashcards em lista
A team needs a data warehouse for SQL analytics without manually provisioning capacity. Which Amazon Redshift option fits?
Amazon Redshift Serverless automatically provisions resources and scales warehouse capacity.
A team wants to query files in Amazon S3 directly from its Amazon Redshift analytics workflow. Which feature fits?
Amazon Redshift Spectrum queries data directly from files in S3.
A team wants Redshift warehouse capacity without manually managing a provisioned cluster. What can it choose?
Redshift Serverless automatically provisions resources and scales warehouse capacity.
Analysts load S3 data into a warehouse each morning, need fast MPP queries without managing clusters, and want to run ML predictions in SQL. What fits?
A daily Glue job that loads Amazon Redshift Serverless, with Redshift ML for SQL-based models. Serverless keeps Redshift MPP query engine with automatic capacity.
A Redshift data warehouse must survive a full Region outage. What should be turned on?
Cross-Region snapshot copy, so snapshots are copied automatically to another Region and the cluster can be restored there.
A team wants to keep its Amazon Redshift data warehouse operating through an Availability Zone failure. What deployment option can help?
A supported provisioned Multi-AZ cluster runs compute in two Availability Zones.
Which AWS service provides a managed SQL data warehouse for analytics?
Amazon Redshift provides a managed cloud warehouse for SQL and business intelligence tools.
How can Amazon Redshift query external files in S3 directly?
Use Redshift Spectrum with an external schema, an external table, and permission to access the catalog and S3.
A team must control Redshift COPY and UNLOAD network traffic through its VPC. What should it enable?
Enhanced VPC routing, then configure a network path to the data resource.
A Redshift cluster needs to load large S3 datasets into warehouse tables. Which command loads data in parallel?
COPY, with permission for Redshift to access the S3 data.
A Redshift data warehouse must survive an Availability Zone failure without a restore. What deployment fits?
A Multi-AZ deployment on RA3 nodes.
An Amazon Redshift report scans millions of records but reads only a few columns. Which storage design reduces disk I/O?
Columnar storage. Redshift reads the needed columns without loading every field in each row.