I’m excited to announce in the present day a brand new functionality of Amazon Managed Streaming for Apache Kafka (Amazon MSK) that lets you repeatedly load knowledge from an Apache Kafka cluster to Amazon Easy Storage Service (Amazon S3). We use Amazon Kinesis Information Firehose—an extract, rework, and cargo (ETL) service—to learn knowledge from a Kafka matter, rework the data, and write them to an Amazon S3 vacation spot. Kinesis Information Firehose is totally managed and you may configure it with only a few clicks within the console. No code or infrastructure is required.

Kafka is often used for constructing real-time knowledge pipelines that reliably transfer large quantities of knowledge between methods or functions. It gives a extremely scalable and fault-tolerant publish-subscribe messaging system. Many AWS prospects have adopted Kafka to seize streaming knowledge comparable to click-stream occasions, transactions, IoT occasions, and software and machine logs, and have functions that carry out real-time analytics, run steady transformations, and distribute this knowledge to knowledge lakes and databases in actual time.

Nonetheless, deploying Kafka clusters isn’t with out challenges.

The primary problem is to deploy, configure, and keep the Kafka cluster itself. Because of this we launched Amazon MSK in Could 2019. MSK reduces the work wanted to arrange, scale, and handle Apache Kafka in manufacturing. We care for the infrastructure, liberating you to focus in your knowledge and functions. The second problem is to put in writing, deploy, and handle software code that consumes knowledge from Kafka. It sometimes requires coding connectors utilizing the Kafka Join framework after which deploying, managing, and sustaining a scalable infrastructure to run the connectors. Along with the infrastructure, you additionally should code the information transformation and compression logic, handle the eventual errors, and code the retry logic to make sure no knowledge is misplaced in the course of the switch out of Kafka.

Right this moment, we announce the provision of a completely managed answer to ship knowledge from Amazon MSK to Amazon S3 utilizing Amazon Kinesis Information Firehose. The answer is serverless–there isn’t a server infrastructure to handle–and requires no code. The info transformation and error-handling logic may be configured with a number of clicks within the console.

The structure of the answer is illustrated by the next diagram.

Amazon MSK is the information supply, and Amazon S3 is the information vacation spot whereas Amazon Kinesis Information Firehose manages the information switch logic.

When utilizing this new functionality, you now not have to develop code to learn your knowledge from Amazon MSK, rework it, and write the ensuing data to Amazon S3. Kinesis Information Firehose manages the studying, the transformation and compression, and the write operations to Amazon S3. It additionally handles the error and retry logic in case one thing goes flawed. The system delivers the data that may not be processed to the S3 bucket of your alternative for guide inspection. The system additionally manages the infrastructure required to deal with the information stream. It can scale out and scale in robotically to regulate to the quantity of knowledge to switch. There aren’t any provisioning or upkeep operations required in your facet.

Kinesis Information Firehose supply streams assist each private and non-private Amazon MSK provisioned or serverless clusters. It additionally helps cross-account connections to learn from an MSK cluster and to put in writing to S3 buckets in numerous AWS accounts. The Information Firehose supply stream reads knowledge out of your MSK cluster, buffers the information for a configurable threshold dimension and time, after which writes the buffered knowledge to Amazon S3 as a single file. MSK and Information Firehose have to be in the identical AWS Area, however Information Firehose can ship knowledge to Amazon S3 buckets in different Areas.

Kinesis Information Firehose supply streams may convert knowledge sorts. It has built-in transformations to assist JSON to Apache Parquet and Apache ORC codecs. These are columnar knowledge codecs that save area and allow quicker queries on Amazon S3. For non-JSON knowledge, you should utilize AWS Lambda to remodel enter codecs comparable to CSV, XML, or structured textual content into JSON earlier than changing the information to Apache Parquet/ORC. Moreover, you may specify knowledge compression codecs from Information Firehose, comparable to GZIP, ZIP, and SNAPPY, earlier than delivering the information to Amazon S3, or you may ship the information to Amazon S3 in its uncooked type.

Let’s See How It Works
To get began, I take advantage of an AWS account the place there’s an Amazon MSK cluster already configured and a few functions streaming knowledge to it. To get began and to create your first Amazon MSK cluster, I encourage you to learn the tutorial.

For this demo, I take advantage of the console to create and configure the information supply stream. Alternatively, I can use the AWS Command Line Interface (AWS CLI), AWS SDKs, AWS CloudFormation, or Terraform.

I navigate to the Amazon Kinesis Information Firehose web page of the AWS Administration Console after which select Create supply stream.

I choose Amazon MSK as an information Supply and Amazon S3 as a supply Vacation spot. For this demo, I need to hook up with a personal cluster, so I choose Non-public bootstrap brokers underneath Amazon MSK cluster connectivity.

I have to enter the total ARN of my cluster. Like most individuals, I can not keep in mind the ARN, so I select Browse and choose my cluster from the listing.

Lastly, I enter the cluster Matter identify I would like this supply stream to learn from.

After the supply is configured, I scroll down the web page to configure the information transformation part.

On the Rework and convert data part, I can select whether or not I need to present my very own Lambda operate to remodel data that aren’t in JSON or to remodel my supply JSON data to one of many two obtainable pre-built vacation spot knowledge codecs: Apache Parquet or Apache ORC.

Apache Parquet and ORC codecs are extra environment friendly than JSON format to question knowledge from Amazon S3. You’ll be able to choose these vacation spot knowledge codecs when your supply data are in JSON format. You need to additionally present an information schema from a desk in AWS Glue.

These built-in transformations optimize your Amazon S3 price and scale back time-to-insights when downstream analytics queries are carried out with Amazon Athena, Amazon Redshift Spectrum, or different methods.

Lastly, I enter the identify of the vacation spot Amazon S3 bucket. Once more, once I can not keep in mind it, I take advantage of the Browse button to let the console information me by means of my listing of buckets. Optionally, I enter an S3 bucket prefix for the file names. For this demo, I enter aws-news-blog. After I don’t enter a prefix identify, Kinesis Information Firehose makes use of the date and time (in UTC) because the default worth.

Underneath the Buffer hints, compression and encryption part, I can modify the default values for buffering, allow knowledge compression, or choose the KMS key to encrypt the information at relaxation on Amazon S3.

When prepared, I select Create supply stream. After a number of moments, the stream standing adjustments to ✅  obtainable.

Assuming there’s an software streaming knowledge to the cluster I selected as a supply, I can now navigate to my S3 bucket and see knowledge showing within the chosen vacation spot format as Kinesis Information Firehose streams it.

As you see, no code is required to learn, rework, and write the data from my Kafka cluster. I additionally don’t should handle the underlying infrastructure to run the streaming and transformation logic.

Pricing and Availability.
This new functionality is on the market in the present day in all AWS Areas the place Amazon MSK and Kinesis Information Firehose can be found.

You pay for the quantity of knowledge going out of Amazon MSK, measured in GB per thirty days. The billing system takes under consideration the precise report dimension; there isn’t a rounding. As regular, the pricing web page has all the main points.

I can’t wait to listen to in regards to the quantity of infrastructure and code you’re going to retire after adopting this new functionality. Now go and configure your first knowledge stream between Amazon MSK and Amazon S3 in the present day.

— seb