There are various ways to transfer logs from S3 to Opensearch:
Amazon Glue
Kinesis Data Firehose
Lambda as an Event Handler
What should be used in what situation? What is the cheapest? I would imagine that Kinesis and/or Event Handler method would be the quickest, but that might also put a big load on your cluster given that many calls would be made very often and there is not as much opportunity for bulk uploads. But with Glue you could for example do this operation say every 10 minutes, and then have a lot of bulk uploads, or schedule this operation in low usage periods for logs you do not need to be inserted into opensearch so quickly. I'd be interested to hear under what situation what strategy is used. I want to minimize the load on my cluster as I feel at the end of the day, putting a higher load on opensearch will cost the most.
Related
In AWS Glue job, we can write some script and execute the script via job.
In AWS Lambda too, we can write the same script and execute the same logic provided in above job.
So, my query is not whats the difference between AWS Glue Job vs AWS Lambda, BUT iam trying to undestand when AWS Glue job should be preferred over AWS Lambda, especially while when both does the same job? If both does the same job, then ideally I would blindly prefer using AWS Lambda itself, right?
Please try to understand my query..
Additional points:
Per this source and Lambda FAQ and Glue FAQ
Lambda can use a number of different languages (Node.js, Python, Go, Java, etc.) vs. Glue can only execute jobs using Scala or Python code.
Lambda can execute code from triggers by other services (SQS, Kafka, DynamoDB, Kinesis, CloudWatch, etc.) vs. Glue which can be triggered by lambda events, another Glue jobs, manually or from a schedule.
Lambda runs much faster for smaller tasks vs. Glue jobs which take longer to initialize due to the fact that it's using distributed processing. That being said, Glue leverages its parallel processing to run large workloads faster than Lambda.
Lambda looks to require more complexity/code to integrate into data sources (Redshift, RDS, S3, DBs running on ECS instances, DynamoDB, etc.) while Glue can easily integrate with these. However, with the addition of Step Functions, multiple lambda functions can be written and ordered sequentially due reduce complexity and improve modularity where each function could integrate into a aws service (Redshift, RDS, S3, DBs running on ECS instances, DynamoDB, etc.)
Glue looks to have a number of additional components, such as Data Catalog which is a central metadata repository to view your data, a flexible scheduler that handles dependency resolution/job monitoring/retries, AWS Glue DataBrew for cleaning and normalizing data with a visual interface, AWS Glue Elastic Views for combining and replicating data across multiple data stores, AWS Glue Schema Registry to validate streaming data schema.
There are other examples I am missing, so feel free to comment and I can update.
Lambda has a lifetime of fifteen minutes. It can be used to trigger a glue job as an event based activity. That is, when a file lands in S3 for example, we can have an event trigger which can run a glue job. Glue is a managed services for all data processing.
If the data is very low maybe you can do it in lambda, but for some reason the process goes beyond fifteen minutes, then data processing would fail.
The answer to this can involve some foundational design decisions. What is this job doing? What kind of data are you dealing with? Is there a decision to be made whether the task should be executed in a batch or event oriented paradigm?
Batch
This may be necessary or desirable because the task:
Is being done over large monolithic data (e.g., binary).
Relies on context of multiple records in a dataset such that they must be loaded into a single job.
Order matters.
I feel like just as often I see batch handling chosen by default because "this is the way we've always done it" but breaking from this approach could be worth consideration.
Glue is built for batch operations. With a current maximum execution time of 15 minutes and maximum memory of 10gb, Lambda has become capable of processing fairly large datasets in a single execution, as well. It can be difficult to pin down a direct cost comparison without specifics of the workload. When it comes to development, I feel that Lambda has the edge as far as tooling to build, test, deploy.
Event
In the case where your data consists of a set of records, it might behoove you to parse and "stream" them into Lambda. Consider a flow like:
CSV lands in S3.
S3 event triggers Lambda.
Lambda reads and parses CSV into discrete events, submits to another Lambda or publishes to SNS for downstream processing. Concurrent instances of this Lambda can be employed to speed up ingest, where each instance is responsible for certain lines of the S3 object.
This pushes all logic and error handling, as well as resources required, to the level of individual event/record level. Often mechanisms such as dead-letter queues are employed for remediation. While context of a given container persists across invocations - assuming the container has not been idle and torn down - Lambda should generally be considered stateless such that the processing of an event/record is thought of as occurring within its own scope, outside that of others in the dataset.
I'm storing some events into DynamoDB. I have to sync (i.e. copy incrementally) the data with Redshift. Ultimately, I want to be able to analyze the data through AWS Quicksight.
I've come across multiple solutions but those are either one-time (using the one-time COPY command) or real-time (a streaming data pipeline using Kinesis Firehose).
The real-time solution seems superior to hourly sync, but I'm worried about performance and complexity. I was wondering if there's an easier way to batch the updates on an hourly basis.
What you are looking for are DynamoDB Streams (official docs). This can seamlessly flow into the Kinesis firehose as you have correctly pointed out.
This is the most optimal way and provides the best balance between the cost, operational overhead and the functionality itself. Allow me explain how:
DynamoDB streams: Streams are triggered when any activity happens on the database. This means that unlike a process that will scan the data on a periodic basis and consume the read capacity even if there is no update, you will be notified of the new data.
Kinesis Firehose: You can configure the Firehose to batch data either by the size of the data or time. This means that if you have a good inflow, you can set the stream to batch the records received in each 2 minutes interval and then issue just one COPY command to redshift. Same goes for the size of the data in the stream buffer. Read more about it here.
The ideal way to load data into Redshift is via COPY command and Kinesis Firehose does just that. You can also configure it to automatically create backup of the data into S3.
Remember that a reactive or push based system is almost always more performant and less costly than a reactive or push based system. You save on the compute capacity needed to run a cron process and also continuously scan for updates.
I have a process that publishes data into a IoT-Core and that triggers a Lambda function that inserts the payload into an Amazon S3 bucket.
I have a process that send around 1.2 million records in some seconds, and when I check in the bucket I see I have lost around 10% of the data. If I set a sleep in the Lambda function it goes beyond 15 minutes.
What is the solution for this scenario?
It appears that your requirement is to capture the events coming into IoT-Core and save them to Amazon S3.
It also sounds like your Lambda functions are being throttled due to hitting concurrency limits and data is being lost. By default, there is a limit of 10,000 concurrent AWS Lambda functions. This could potentially be fixed by requesting an increase in the maximum number of concurrent functions.
Here is a diagram from How AWS IoT works:
As shown in the digram, the Rules engine can actually be used to send data to Amazon S3 without requiring Lambda. However, this creates a separate object in Amazon S3 for every message.
If you wish to combine messages together, you can Write to Kinesis Data Firehose Using AWS IoT. Firehose will buffer the data by time or size, and then output multiple messages to an Amazon S3 object. This could be a good way to handle large volumes of data, and it also makes it easier to work with the resulting objects in S3 because there are less objects created. This makes them faster to query and process later (eg with Amazon Athena).
Going from IoT-Core rule direct to a Lambda can be fragile.
You can use Kinesis to buffer the data or Firehose to stream it directly to S3. These are standard patterns that AWS recommend for IoT in the AWS Well-Architected framework (https://d1.awsstatic.com/whitepapers/architecture/AWS-IoT-Lens.pdf).
I have an S3 bucket with different files. I need to read those files and publish SQS msg for each row in the file.
I cannot use S3 events as the files need to be processed with a delay - put to SQS after a month.
I can write a scheduler to do this task, read and publish. But can I was AWS for this purpose?
AWS Batch or AWS data pipeline or Lambda.?
I need to pass the date(filename) of the data to be read and published.
Edit : The data volume to be dealt is huge
I can think of two ways to do this entirely using AWS serverless offerings without even having to write a scheduler.
You could use S3 events to start a Step Function that waits for a month before reading the S3 file and sending messages through SQS.
With a little more work, you could use S3 events to trigger a Lambda function which writes the messages to DynamoDB with a TTL of one month in the future. When the TTL expires, you can have another Lambda that listens to the DynamoDB streams, and when there’s a delete event, it publishes the message to SQS. (A good introduction to this general strategy can be found here.)
While the second strategy might require more effort, you might find it less expensive than using Step Functions depending on the overall message throughput and whether or not the S3 uploads occur in bursts or in a smooth distribution.
At the core, you need to do two things:
Enumerate all of the object in a bucket in S3, and perform some action on any object uploaded more than a month ago.
Can you use Lambda or Batch to do this? Sure. A Lambda could be set to trigger once a day, enumerate the files, and post the results to SQS.
Should you? No clue. A lot depends on your scale, and what you plan to do if it takes a long time to perform this work. If your S3 bucket has hundreds of objects, it won't be a problem. If it has billions, your Lambda will need to be able to handle being interrupted, and continuing paging through files from a previous run.
Alternatively, you could use S3 events to trigger a simple Lambda that adds a row to a database. Then, again, some Lambda could run on a cron job that asks the database for old rows, and publishes that set to SQS for others to consume. That's slightly cleaner, maybe, and can handle scaling up to pretty big bucket sizes.
Or, you could do the paging through files, deciding what to do, and processing old files all on a t2.micro if you just need to do some simple work to a few dozen files every day.
It all depends on your workload and needs.
We wish to build data pipeline system which tracks website interactions/events.
The goal is to track user behavior in a website so we would like to choose the right architecture to implement it having the following two constraints :
1) the system is Amazon
2) this is budgetary project so we cannot use redshift for this purpose
Based on the above two constraints my plan is to implement the following architecture:
website-javascript --> AWS-S3 -->(AWS-Lambda)--> AWS-RDS
website javascript client -
aws-firehose data delivery system to S3 - tracking user interaction and load them to aws-firehose which eventually write them in aws-S3.
AWS Lambda (Python) - Periodically task which pulls daily events from AWS-S3 and load them to AWS-RDS.
The reason I have chosen AWS-RDS is due to its cost-effectiveness for this objective
Appreciate any comment to the above mentioned implementation or any other architecture proposal that you may recommend to use instead of the above
If I understand your question correctly, you are proposing below solution to perform web analytics for your application:
WebServer --> Firehose --> AWS-S3 --> AWS-Lambda --> AWS-RDS
I see below pros and cons with above design
Pros:
low cost
easy to implement
Cons:
RDS may not be salable enough to handle analytics on massive amounts of web-streaming data, which tend to grow rapidly
Need to handle load balancing, failure scenarios and other complexities for lambda
You need to handle data transformation for RDS as it expects structured data to be ingested into relational tables
Proposal to store the data in S3 through Firehose sounds a good solution. But please keep in mind that minimum interval for Firehose is one minute, so your application needs to tolerate this minor latency. You may use Kinesis Streams to have millisecond latency, but then you need to manage your own application code and instances to handle Streams.
After ingesting data in Kinesis Firehose or Streams, you may also explore below alternatives:
Use Kinesis Analytics to track web users activity in real-time if it's available in your AWS region. It's only available in selected AWS regions currently
Within Firehose, transform your data using lambda and store it in S3 in optimized format for further analysis with AWS Athena
Use Elastic Search as a destination and perform web analytics with ELK stack instead of RDS
Though you mentioned that you can not use RedShift, it still may be the best solution for time series analysis. Exploring RedShift, RedShift Spectrum and formatted data stored in S3 may still be a cost effective solution with better cababilities
Adding few references from AWS, which you may go through before deciding on the solution:
Real-Time Web Analytics with Kinesis Data Analytics Solution
Near Real-time Analytics on Streaming Data with Amazon Kinesis and Amazon Elasticsearch
Schema-On-Read Analytics Pipeline Using Amazon Athena
Amazon Redshift Spectrum Extends Data Warehousing Out to Exabytes—No Loading Required
Hey folky this is a getting more and more common.
Generally the pattern is click events to Kinesis streams then you can monitor user interaction with the website in real time using Kinesis analytics. You can connect the stream to firehose to offload data in to an S3 bucket as well as incorporate Lambdas to transform the data.
There is some major complexity around handling Lambdas and Kinesis streams in parallel so this solution might not be as scalable as using AWS Kafka. Or perhaps run a job to move your s3 data into rds for whatever reporting you might need that is adhoc.
Here is a pattern AWS already has real-time-web-analytics-with-kinesis