I currently have a SQS queue that triggers a lambda function, but SQS have a 120k limit on in-flight messages, meaning that only 120k messages can be processed by the lambda consumer. That works fine for the most case. But imagine I have clients A and B, if A sends 500k to the queue and after that B sends only 1 message, B needs to wait all messages in the queue to be processed. It does not make sense to create one queue for each client (at least not manually). How can I, for example, process messages from clients in a round-robin manner, give everyone the same time to process their tasks?
I have been looking into Kinesis data stream, but not so sure if this will solve my problem. Would I need to create an SNS that redirects to a lambda responsible to round-robin the message to queues that do the same thing but this way I would avoid sending client B to same queue as client A?
You can check out SQS Fifo Queues. Fifo Queues have a concept of grouping using the group Id attribute. all messages in one group are processed in order. The limitations are you can only process one message per group at any given moment.
The above concept is the same as having one queue per tenant, just using one dedicated Queue.
Refer to the AWS Docs
https://aws.amazon.com/blogs/compute/solving-complex-ordering-challenges-with-amazon-sqs-fifo-queues/
Article on ordering inside SQS
https://aws.amazon.com/blogs/compute/solving-complex-ordering-challenges-with-amazon-sqs-fifo-queues/
Note:
FIFO queues are more expensive than standard queues, please refer to the pricing guide before making any decisions.
I am looking at using SNS-SQS services to deliver updates to machines running the same service. Since the plan is not for machines to communicate with each other, I was planning on creating a SQS for each machine (SQS would be created at startup).
I am however, not sure how to use a Dead-Letter Queue (DLQ) in such case. Should each SQS have its own DLQ or can I have common one which is shared across my SQS in the region? The concern I have with former approach is too many queues would be created (2x machines) and the concern with later is potential multiple copies of same message in the queue.
What is the best practice and recommended approach when using multiple SQS queues?
I wouldn't be concerned with the number of queues - they don't cost anything - so it really depends on how you plan on using the items in the dead-letter queue. I'll make the assumption that you will have some sort of process to review items in the DLQ to figure out why they were not processed before expiring.
Without knowing the details of what you plan to do, I would think a single DLQ would be better, and if you need to periodically process DLQ records, the processing app/system only needs to monitor that single queue.
Can't see the advantage of multiple DLQs in this case, at least based on your question.
As you are planning on doing a fanout process, having multiple queues is not a harm as long as they are used for asynchronous processing. Else a single queue is preferred. A fanout process in generally used when you want to process a few tasks concurrently by dividing it among several queues and working on them separately. (to read more about fanout)
The purpose of a Dead Letter Queue (DLQ) is to store messages that cannot be completed successfully by a certain process. Unless your process has a major fault, the number of elements that will be stored in a DLQ should be very less. Therefore it is okay to go ahead and use one DLQ for all other SQSs.
Having multiple DLQs will bring an overhead, where several processes will have to poll the DLQs for failed messages. Having just one DLQ can reduce this overhead.
It is recommended to use multiple DLQs if you want to store different categories of the failed messages.
We plan to use AWS SQS service to queue events created from web service and then use several workers to process those events. One event can only be processed one time. According to AWS SQS document, AWS SQS standard queue can "occasionally" produce duplicated message but with unlimited throughput. AWS SQS FIFO queue will not produce duplicated message but with throughput limitation of 300 API calls per second (with batchSize=10, equivalent of 3000 messages per second). Our current peak hour traffic is only 80 messages per second. So, both are fine in terms of throughput requirement. But, when I started to use AWS SQS FIFO queue, I found that I need to do extra work like providing extra parameters
"MessageGroupId" and "MessageDeduplicationId" or need to enable "ContentBasedDeduplication" setting. So, I am not sure which one is a better solution. We just need the message not duplicated. We don't need the message to be FIFO.
Solution #1:
Use AWS SQS FIFO queue. For each message, need to generate a UUID for "MessageGroupId" and "MessageDeduplicationId" parameters.
Solution #2:
Use AWS SQS FIFO queue with "ContentBasedDeduplcation" enabled. For each message, need to generate a UUID for "MessageGroupId".
Solution #3:
Use AWS SQS standard queue with AWS ElasticCache (either Redis or Memcached). For each message, the "MessageId" field will be saved in the cache server and checked for duplication later on. Existence means this message has been processed. (By the way, how long should the "MessageId" exists in the cache server. AWS SQS document does not mention how far back a message could be duplicated.)
You are making your systems complicated with SQS.
We have moved to Kinesis Streams, It works flawlessly. Here are the benefits we have seen,
Order of Events
Trigger an Event when data appears in stream
Deliver in Batches
Leave the responsibility to handle errors to the receiver
Go Back with time in case of issues
Buggier Implementation of the process
Higher performance than SQS
Hope it helps.
My first question would be that why is it even so important that you don't get duplicate messages? An ideal solution would be to use a standard queue and design your workers to be idempotent. For e.g., if the messages contain something like a task-ID and store the completed task's result in a database, ignore those whose task-ID already exists in DB.
Don't use receipt-handles for handling application-side deduplication, because those change every time a message is received. In other words, SQS doesn't guarantee same receipt-handle for duplicate messages.
If you insist on de-duplication, then you have to use FIFO queue.
Is it possible to dump a SQS queue to open space for "urgent" messages and then restore the dump to keep SQS queue on track?
I am not talking about aws cli commands but any possibility of doing it.
Of course I could open a new SQS and change the application to look after that new queue, but it would have some implications.
No it's not possible. The design pattern I've seen AWS recommend when you want to have "high priority" messages is this:
Create 2 queues, one for high-priority messages and one for regular-priority messages.
Have your application always scan the high-priority queue first to check for new messages.
If you don't receive any messages from the high-priority queue, scan the regular-priority queue for messages.
AWS SQS does not provide a priority based queue at the moment. But you can do certain implementations and build a priority queue for your application (consumer). Following are some implementations you can use.
1) As #Markb mentioned, you can create two SQSs, where one is for high priority messages and other is for regular messages. Make sure application polls the high-priority SQS first and then move on to the regular SQS.
2) If using a single SQS, have a few worker threads on the application side, that will collect all the messages from SQS, and process them to see which ones have a higher priority. Take them and process first.
3) Use a combination of SQS and SNS. Send all the regular messages into the SQS. If there are high priority messages, send them to SNS to direct them to a specific endpoint in your application. From you application side (consumer side), have and endpoint that listens to high priority messages coming from SNS, and then have aprocess that polls the SQS to retrieve the regular messages.
When would I use SNS versus SQS, and why are they always coupled together?
SNS is a distributed publish-subscribe system. Messages are pushed to subscribers as and when they are sent by publishers to SNS.
SQS is distributed queuing system. Messages are not pushed to receivers. Receivers have to poll or pull messages from SQS. Messages can't be received by multiple receivers at the same time. Any one receiver can receive a message, process and delete it. Other receivers do not receive the same message later. Polling inherently introduces some latency in message delivery in SQS unlike SNS where messages are immediately pushed to subscribers. SNS supports several end points such as email, SMS, HTTP end point and SQS. If you want unknown number and type of subscribers to receive messages, you need SNS.
You don't have to couple SNS and SQS always. You can have SNS send messages to email, SMS or HTTP end point apart from SQS. There are advantages to coupling SNS with SQS. You may not want an external service to make connections to your hosts (a firewall may block all incoming connections to your host from outside).
Your end point may just die because of heavy volume of messages. Email and SMS maybe not your choice of processing messages quickly. By coupling SNS with SQS, you can receive messages at your pace. It allows clients to be offline, tolerant to network and host failures. You also achieve guaranteed delivery. If you configure SNS to send messages to an HTTP end point or email or SMS, several failures to send message may result in messages being dropped.
SQS is mainly used to decouple applications or integrate applications. Messages can be stored in SQS for a short duration of time (maximum 14 days). SNS distributes several copies of messages to several subscribers. For example, let’s say you want to replicate data generated by an application to several storage systems. You could use SNS and send this data to multiple subscribers, each replicating the messages it receives to different storage systems (S3, hard disk on your host, database, etc.).
Here's a comparison of the two:
Entity Type
SQS: Queue (Similar to JMS)
SNS: Topic (Pub/Sub system)
Message consumption
SQS: Pull Mechanism - Consumers poll and pull messages from SQS
SNS: Push Mechanism - SNS Pushes messages to consumers
Use Case
SQS: Decoupling two applications and allowing parallel asynchronous processing
SNS: Fanout - Processing the same message in multiple ways
Persistence
SQS: Messages are persisted for some (configurable) duration if no consumer is available (maximum two weeks), so the consumer does not have to be up when messages are added to queue.
SNS: No persistence. Whichever consumer is present at the time of message arrival gets the message and the message is deleted. If no consumers are available then the message is lost after a few retries.
Consumer Type
SQS: All the consumers are typically identical and hence process the messages in the exact same way (each message is processed once by one consumer, though in rare cases messages may be resent)
SNS: The consumers might process the messages in different ways
Sample applications
SQS: Jobs framework: The Jobs are submitted to SQS and the consumers at the other end can process the jobs asynchronously. If the job frequency increases, the number of consumers can simply be increased to achieve better throughput.
SNS: Image processing. If someone uploads an image to S3 then watermark that image, create a thumbnail and also send a Thank You email. In that case S3 can publish notifications to an SNS topic with three consumers listening to it. The first one watermarks the image, the second one creates a thumbnail and the third one sends a Thank You email. All of them receive the same message (image URL) and do their processing in parallel.
You can see SNS as a traditional topic which you can have multiple Subscribers. You can have heterogeneous subscribers for one given SNS topic, including Lambda and SQS, for example. You can also send SMS messages or even e-mails out of the box using SNS. One thing to consider in SNS is only one message (notification) is received at once, so you cannot take advantage from batching.
SQS, on the other hand, is nothing but a queue, where you store messages and subscribe one consumer (yes, you can have N consumers to one SQS queue, but it would get messy very quickly and way harder to manage considering all consumers would need to read the message at least once, so one is better off with SNS combined with SQS for this use case, where SNS would push notifications to N SQS queues and every queue would have one subscriber, only) to process these messages. As of Jun 28, 2018, AWS Supports Lambda Triggers for SQS, meaning you don't have to poll for messages any more.
Furthermore, you can configure a DLQ on your source SQS queue to send messages to in case of failure. In case of success, messages are automatically deleted (this is another great improvement), so you don't have to worry about the already processed messages being read again in case you forgot to delete them manually. I suggest taking a look at Lambda Retry Behaviour to better understand how it works.
One great benefit of using SQS is that it enables batch processing. Each batch can contain up to 10 messages, so if 100 messages arrive at once in your SQS queue, then 10 Lambda functions will spin up (considering the default auto-scaling behaviour for Lambda) and they'll process these 100 messages (keep in mind this is the happy path as in practice, a few more Lambda functions could spin up reading less than the 10 messages in the batch, but you get the idea). If you posted these same 100 messages to SNS, however, 100 Lambda functions would spin up, unnecessarily increasing costs and using up your Lambda concurrency.
However, if you are still running traditional servers (like EC2 instances), you will still need to poll for messages and manage them manually.
You also have FIFO SQS queues, which guarantee the delivery order of the messages. SQS FIFO is also supported as an event source for Lambda as of November 2019
Even though there's some overlap in their use cases, both SQS and SNS have their own spotlight.
Use SNS if:
multiple subscribers is a requirement
sending SMS/E-mail out of the box is handy
Use SQS if:
only one subscriber is needed
batching is important
AWS SNS is a publisher subscriber network, where subscribers can subscribe to topics and will receive messages whenever a publisher publishes to that topic.
AWS SQS is a queue service, which stores messages in a queue. SQS cannot deliver any messages, where an external service (lambda, EC2, etc.) is needed to poll SQS and grab messages from SQS.
SNS and SQS can be used together for multiple reasons.
There may be different kinds of subscribers where some need the
immediate delivery of messages, where some would require the message
to persist, for later usage via polling. See this link.
The "Fanout Pattern." This is for the asynchronous processing of
messages. When a message is published to SNS, it can distribute it
to multiple SQS queues in parallel. This can be great when loading
thumbnails in an application in parallel, when images are being
published. See this link.
Persistent storage. When a service that is going to process a message is not reliable. In a case like this, if SNS pushes a
notification to a Service, and that service is unavailable, then the
notification will be lost. Therefore we can use SQS as a persistent
storage and then process it afterwards.
From the AWS documentation:
Amazon SNS allows applications to send time-critical messages to
multiple subscribers through a “push” mechanism, eliminating the need
to periodically check or “poll” for updates.
Amazon SQS is a message queue service used by distributed applications
to exchange messages through a polling model, and can be used to
decouple sending and receiving components—without requiring each
component to be concurrently available.
Fanout to Amazon SQS queues
Following are the major differences between the main messaging technologies on AWS (SQS, SNS, +EventBridge). In order to choose a particular AWS service, we should know the functionalities a service provides as well as its comparison with other services.
The below diagram summarizes the main similarities as well as differences between this service.
In simple terms,
SNS - sends messages to the subscriber using push mechanism and no need of pull.
SQS - it is a message queue service used by distributed applications to exchange messages through a polling model, and can be used to decouple sending and receiving components.
A common pattern is to use SNS to publish messages to Amazon SQS queues to reliably send messages to one or many system components asynchronously.
Reference from Amazon SNS FAQs.
One reason for coupling SQS and SNS would be for data processing pipelines.
Let's say you are generating three kinds of product, and that products B & C are both derived from the same intermediate product A. For each kind of product (i.e., for each segment of the pipeline) you set up:
a compute resource (maybe a lambda function, or a cluster of virtual machines, or an autoscaling kubernetes job) to generate the product.
a queue (describing units of work that need to be performed) to partition the work across the compute resource (so that each unit of work is processed exactly once, but separate units of work can be processed separately in parallel and asynchronously with each other).
a news feed (announcing outputs that have been produced).
Then arrange so that the input queues for B & C are both subscribing to the output announcements of A.
This makes the pipeline modular on the level of infrastructure. Rather than having a monolithic server application that generates all three products together, different stages of the pipeline can utilise different hardware resources (for example, perhaps stage B is very memory intensive, but the two other stages can be performed with cheaper hardware/services). This also makes it easier to iterate on the development of one pipeline segment without disrupting delivery of the other products.
There are some key distinctions between SNS and SQS:
SNS supports A2A and A2P communication, while SQS supports only A2A
communication.
SNS is a pub/sub system, while SQS is a queuing system. You'd
typically use SNS to send the same message to multiple consumers via
topics. In comparison, in most scenarios, each message in an SQS
queue is processed by only one consumer. With SQS, messages are
delivered through a long polling (pull) mechanism, while SNS uses a
push mechanism to immediately deliver messages to subscribed
endpoints.
SNS is typically used for applications that need real time
notifications, while SQS is more suited for message processing use
cases.
SNS does not persist messages - it delivers them to subscribers that
are present, and then deletes them. In comparison, SQS can persist
messages (from 1 minute to 14 days).
Individually, Amazon SQS and SNS are used for different use cases. You can, however, use them together in some scenarios.