Skip to main content

Build Ultra-Scalable Backends with Pekko and Kubernetes

Image
By
Listen:

Update May 21, 2024: 

Lightbend, the company behind Akka, changed the license from Apache 2.0 to BSL, which makes Akka not really attractive anymore. The new license is in place from Akka 2.7 onwards. 

Pekko is a 2.6x fork of Akka, with the same functionalities,  maintained and further developed by the Apache Software Foundation, and is under Apache Software Licence 2 available.

Ractor is written in Rust, has network communication, single messaging, named actors and a cluster mode, and is under MIT license available.

------------------------------

Building a backend that can handle massive traffic and keep your users happy is tough. We're not talking about a simple website with a few hundred visitors a day. We're talking about high-performance systems that can handle spikes in traffic, process huge amounts of data, and keep chugging along even when things go wrong. I have made some great experiences with Akka as micro service in kubernetes, we use k8s to scale our IoT platform across different regions and hybrid setups. 

Running Akka in k8s is a great match from my point of view, they complement each other perfectly, allowing you to build backend systems that are not only scalable and resilient but also incredibly efficient and easy to manage.

Why Akka?

If you're not familiar with Akka, it's a toolkit and runtime for building highly concurrent, distributed, and resilient message-driven applications on the JVM. In plain English, this means it's a framework that helps you build apps that can handle a ton of stuff happening at once, spread the load across multiple machines, and keep running even when things go wrong.

The secret to Akka's power is the Actor Model. Actors are like little independent workers that can communicate with each other by sending messages. This makes it super easy to build complex systems that can scale up or down as needed.

Think of it like this: instead of having one overworked employee trying to juggle a thousand tasks, you have a team of specialists, each handling their own piece of the puzzle. This not only makes things more efficient, but it also makes it easier to recover from failures. If one actor crashes, the rest of the system can keep on running.

Why Kubernetes?

Now, let's talk about Kubernetes, or short k8s. Kubernetes is an open-source system for automating the deployment, scaling, and management of containerized applications, initially developed by Google to handle the infrastructure and applications for GCP.

What are containers? In a nutshell, containers are a way to package up an application and all its dependencies so that it can run reliably across different computing environments. There are multiple container management platforms out there, like Docker Swarm, Openshift, or Kubernetes. I personally like k8s, since it's easier to scale and distribute, plus you can better manage multiple installations in deployment sets, using Namespaces to distinguish between applications, etc etc. Kubernetes handles everything from scheduling containers on different machines to load balancing traffic to ensuring that your applications are always available.

Run Akka with k8s

There are multiple pros running Akka in k8s, first Akka actors are naturally suited to a microservices architecture. Each actor can be thought of as an independent service, handling specific tasks and communicating with other actors through messages. Kubernetes provides the perfect platform for deploying and managing these microservices.

2nd, Kubernetes excels at scaling applications up or down based on demand. With Akka actors running as microservices, you can easily add or remove actor instances to meet changing workloads. This ensures your backend can handle traffic spikes without breaking (given the cluster is set to autoscale). Adding a loadbalancer and scale out to the actors, k8s takes care to start new pods at a given load (scale-out), and also kills pods when the load goes back (scale-down). 

3rd, with Akka's built-in fault-tolerance mechanisms, combined with Kubernetes' self-healing capabilities, I create a backend that can withstand failures. If an actor crashes, Kubernetes can automatically restart it on a different node, ensuring uninterrupted operation. 

Here's a simplified ASCII diagram to illustrate the concept:

Akka with k8s simple architecture diagram


A yaml file for a simple setup looks like this one:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-akka-actor
spec:
  replicas: 3 
  selector:
    matchLabels:
      app: my-akka-actor
  template:
    metadata:
      labels:
        app: my-akka-actor
    spec:
      containers:
      - name: my-akka-actor
        image: my-akka-actor-image:latest
        ports:
        - containerPort: 8080 

---
apiVersion: v1
kind: Service
metadata:
  name: my-akka-actor-service
spec:
  selector:
    app: my-akka-actor
  ports:
    - protocol: TCP
      port: 80  # External port (can be different from containerPort)
      targetPort: 8080 
  type: LoadBalancer

This yaml tells Kubernetes to create three instances of your Akka actor, each running in its own container. Kubernetes will automatically manage these instances, ensuring they are spread across different nodes for high availability. The loadbalancer distributes the load, to have persistent session add a sticky mechanism, ideally with a session token.

If you want to build a backend that can handle anything you throw at it, Akka and Kubernetes can be your new best friends. They offer a scalable, resilient, and efficient solution that can help you deliver amazing user experiences.


Comments

Post a Comment

Popular posts from this blog

Deal with corrupted messages in Apache Kafka

By
Under some strange circumstances, it can happen that a message in a Kafka topic is corrupted. This often happens when using 3rd party frameworks with Kafka. In addition, Kafka < 0.9 does not have a lock on Log.read() at the consumer read level, but does have a lock on Log.write(). This can lead to a rare race condition as described in KAKFA-2477 [1]. A likely log entry looks like this: ERROR Error processing message, stopping consumer: (kafka.tools.ConsoleConsumer$) kafka.message.InvalidMessageException: Message is corrupt (stored crc = xxxxxxxxxx, computed crc = yyyyyyyyyy Kafka-Tools Kafka stores the offset of each consumer in Zookeeper. To read the offsets, Kafka provides handy tools [2]. But you can also use zkCli.sh, at least to display the consumer and the stored offsets. First we need to find the consumer for a topic (> Kafka 0.9): bin/kafka-consumer-groups.sh --zookeeper management01:2181 --describe --group test Prior to Kafka 0.9, the only way to get this inform
Post a Comment
Read more

Hive query shows ERROR "too many counters"

By Anonymous
A hive job face the odd " Too many counters:"  like Ended Job = job_xxxxxx with exception 'org.apache.hadoop.mapreduce.counters.LimitExceededException(Too many counters: 201 max=200)' FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.MapRedTask Intercepting System.exit(1) These happens when operators are used in queries ( Hive Operators ). Hive creates 4 counters per operator, max upto 1000, plus a few additional counters like file read/write, partitions and tables. Hence the number of counter required is going to be dependent upon the query.  To avoid such exception, configure " mapreduce.job.counters.max " in mapreduce-site.xml to a value above 1000. Hive will fail when he is hitting the 1k counts, but other MR jobs not. A number around 1120 should be a good choice. Using " EXPLAIN EXTENDED " and " grep -ri operators | wc -l " print out the used numbers of operators. Use this value to tweak the MR s
2 comments
Read more

AI's False Reality: Understanding Hallucination

By
Artificial Intelligence (AI) has leapfrogged to the poster child of technological innovation, on track to transform industries in a scale similar to the Industrial Revolution of the 1800s. But in this case, as cutting-edge technology, AI presents its own unique challenge, exploiting our human behavior of "love to trust", we as humans face a challenge: AI hallucinations. This phenomenon, where AI models generate outputs that are factually incorrect, misleading, or entirely fabricated, raises complex questions about the reliability and trust of AI models and larger systems. The tendency for AI to hallucinate comes from several interrelated factors. Overfitting – a condition where models become overly specialized to their training data – can lead to confident but wildly inaccurate responses when presented with novel scenarios (Guo et al., 2017). Moreover, biases embedded within datasets shape the models' understanding of the world; if these datasets are flawed or unreprese
Post a Comment
Read more