All Other Metrics Are Useless

When it comes to queues, whether they’re implemented as JMS, database tables (i.e. what Ruby’s Delayed::Job uses for a queue), or even Amazon’s SQS, the most common metric used to evaluate the state of a queue is its length. In essence, one derives an efficiency metric based upon how many messages are residing in a queue at any given time. If there are just a few messages, the queue is operating efficiently. If there are numerous messages, things are inefficient and alarms must be sounded.time1

But what if you’re in a consistently busy environment with extreme bursts where queues have the tendency to rapidly fill up? If you have sufficient workers already running to handle that burst, do you need to fire up more?

You can fire up more workers, but doing so might cost you. That is, you might have to provision new worker instances, such as Heroku worker dynos or AWS AMIs, which will end up costing you tangible money. And sometimes those worker instances take a few moments to fire up and when they’re operational, the burst of activity is over and the queue is back to normal – the initially available workers handled the load adequately.

It turns out that the queue’s length was a lagging indicator. You spun up unneeded resources. False alarm!

If you already have sufficient capacity to handle the influx of messages on a queue, then monitoring a queue’s length isn’t too helpful. In fact, it’s a misleading metric and can cause you to take unneeded actions.

Consequently, a queue’s length is not indicative of a system’s efficiency when there’s already sufficient workers present. Rather, the metric that means something in a high capacity environment is how long a message resides in a queue. That is an actionable metric: if messages are stuck in a queue waiting to be processed then you need more processors!

Moo over queue length and let queue wait time in

By default, Amazon’s SQS doesn’t provide the ability to query how long a message has been residing in a queue. Therefore, I wrote Moo.

Moo provides an interface for clients to obtain and take action on the message time in queue metric. This is done by augmenting an SQS message with a time stamp. That time stamp is then checked when a message is popped off of an SQS queue. If a threshold difference is exceed, then a callback is invoked.

Users of Moo will find its usage similar to Ahoy!, which is an asynchronous callback oriented facade on top of AWS’s Java SDK. In fact, Moo uses Ahoy! underneath, with the added feature of attaching a “maximum time in queue” asynchronous callback.

Moo supports multiple time in queue thresholds and setting a maximum time in queue threshold is done like so:

Adding a maximum threshold for time in queue

//adds a 1 second max threshold
sqs.addQueueWaitTimeCallback(1000, new QueueWaitTimeCallback() {
  public void onThresholdExceeded(long waitTime) {
    //waitTime is the actual time in queue
    //do something... like fire off a web hook, etc
  }
});

Note the addQueueWaitTimeCallback method takes a millisecond maximum time in queue value and an accompanying QueueWaitTimeCallback callback implementation. The onThresholdExceeded method will be invoked asynchronously during a message receive if the maximum threshold value is exceeded; what’s more, the onThresholdExceeded will receive as a parameter the actual queue wait time.

Show me the Moo

To fire up an instance of Moo, you have a number of options, including configuring an instance of AWS’s AmazonSQS or just passing along a key, secret, and queue name like so:

Adding a maximum threshold for time in queue

SQS sqs = new SQS(System.getProperty("key"), System.getProperty("secret"), System.getProperty("queue"));

Next, you can attach zero to many QueueWaitTimeCallback instances like so:

Adding a maximum threshold for time in queue

sqs.addQueueWaitTimeCallback(600000, new QueueWaitTimeCallback() {
  public void onThresholdExceeded(long actualWaitTime) {
    //do something -- fire off SNS message?
  }
});

In this case, I’ve added a callback to be invoked if messages are in a queue longer than 10 minutes. Note, these QueueWaitTimeCallback callbacks are fired by the queue reader instance; accordingly, a QueueWaitTimeCallback can certainly fire up more instances of itself, for example.

Here’s a sample JSON document that you might want to throw onto an SQS queue:

Adding a maximum threshold for time in queue

{ "employees":[
      { "firstName":"John", "lastName":"Doe" },
      { "firstName":"Anna", "lastName":"Smith" },
      { "firstName":"Peter", "lastName":"Jones" }
]}

Sending and receiving this message are exactly like you’d do if you were using Ahoy!. For example, to send a message, just pass along a String to the send method:

Adding a maximum threshold for time in queue

sqs.send(json, new SendCallback() {
  public void onSend(String messageId) {
    //messageId is from SQS
  }
});

Note, the send method takes an optional SendCallback.

Receiving a message is via the receive method, which takes a mandatory ReceiveCallback – this callback will be invoked asynchronously for each message received off of a queue. Each instance will receive the message placed upon the queue and the message’s SQS id.

Adding a maximum threshold for time in queue

sqs.receive(new ReceiveCallback() {
  public void onReceive(String messageId, String message) {
    //do something w/the message -- in this case it's JSON
  }
});

Note, if upon the receive of a message, Moo notices that a message has been waiting in a queue for more than the max queue wait time threshold configured for an associated QueueWaitTimeCallback, Moo will invoke it. Note, Moo can invoke more than one instance; thus, you can set up a chain to take various actions as times increase.

Remember, a queue’s length is usually a lagging indicator. The metric that actually means something is how long a message resides in a queue. That’s an actionable metric and Moo gives you the ability to do something about it! Can you dig it?
 

Reference: All Other Metrics Are Useless from our JCG partner Andrew Glover at the The Disco Blog blog.

Do you want to know how to develop your skillset to become a Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

JPA Mini Book

Learn how to leverage the power of JPA in order to create robust and flexible Java applications. With this Mini Book, you will get introduced to JPA and smoothly transition to more advanced concepts.

JVM Troubleshooting Guide

The Java virtual machine is really the foundation of any Java EE platform. Learn how to master it with this advanced guide!

Given email address is already subscribed, thank you!
Oops. Something went wrong. Please try again later.
Please provide a valid email address.
Thank you, your sign-up request was successful! Please check your e-mail inbox.
Please complete the CAPTCHA.
Please fill in the required fields.

Leave a Reply


five − = 0



Java Code Geeks and all content copyright © 2010-2014, Exelixis Media Ltd | Terms of Use | Privacy Policy | Contact
All trademarks and registered trademarks appearing on Java Code Geeks are the property of their respective owners.
Java is a trademark or registered trademark of Oracle Corporation in the United States and other countries.
Java Code Geeks is not connected to Oracle Corporation and is not sponsored by Oracle Corporation.
Do you want to know how to develop your skillset and become a ...
Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

Get ready to Rock!
You can download the complementary eBooks using the links below:
Close