Why Observability with service mesh in Kubernetes? - Purpose & Use Cases

The Big Idea

What if you could see every conversation in your system without asking each app separately?

The Scenario

Imagine you have many small apps talking to each other inside a big system. You want to know if they are working well, but you have to check each app one by one, like calling every friend to ask how they are doing.

The Problem

Checking each app manually is slow and confusing. You might miss problems or get wrong info because apps don't share their health easily. It's like trying to find a lost item in a messy room without any light.

The Solution

A service mesh adds a smart helper between apps that watches all the talks and shares clear reports. It shows you who is talking to whom, how fast, and if there are any problems, all in one place.

Before vs After

✗ Before

kubectl logs app1
kubectl logs app2
kubectl logs app3

✓ After

istioctl dashboard kiali

What It Enables

With observability in a service mesh, you can quickly see the health and flow of your whole system, making it easy to fix issues before users notice.

Real Life Example

A company running many microservices uses a service mesh to spot slow responses and errors instantly, so their app stays fast and reliable for customers.

Key Takeaways

Manual checks are slow and miss the big picture.

Service mesh gives a clear view of all app interactions.

This helps catch and fix problems faster and smarter.

Practice

1. What is the main purpose of using a service mesh for observability in Kubernetes?

easy

A. To replace Kubernetes networking completely

B. To deploy applications faster without monitoring

C. To automatically collect metrics, logs, and traces from microservices

D. To store application data persistently

Solution

1. Step 1: Understand service mesh role in observability

   A service mesh helps by automatically collecting data like metrics, logs, and traces from microservices without manual setup.

2. Step 2: Compare options with this role

   Only To automatically collect metrics, logs, and traces from microservices describes this automatic collection for observability. Other options describe unrelated tasks.

3. Final Answer:

   To automatically collect metrics, logs, and traces from microservices -> Option C

4. Quick Check:

   Service mesh observability = automatic data collection [OK]

Hint: Service mesh = automatic monitoring data collection [OK]

Common Mistakes:

• Thinking service mesh replaces Kubernetes networking
• Confusing observability with deployment speed
• Assuming service mesh stores application data

2. Which of the following is the correct command to install Istio's observability components using istioctl?

easy

A. istioctl install --set profile=demo

B. istioctl deploy --profile=observability

C. kubectl apply -f istio-observability.yaml

D. istioctl setup observability

Solution

1. Step 1: Recall Istio installation syntax

   The correct command to install Istio with observability features is 'istioctl install' with a profile like 'demo' that includes observability tools.

2. Step 2: Check options for correct syntax

   istioctl install --set profile=demo matches the correct syntax. Options A and B use invalid commands. kubectl apply -f istio-observability.yaml is generic and not specific to istioctl.

3. Final Answer:

   istioctl install --set profile=demo -> Option A

4. Quick Check:

   Istio install command = istioctl install --set profile=demo [OK]

Hint: Use 'istioctl install --set profile=demo' for observability [OK]

Common Mistakes:

• Using 'deploy' instead of 'install' with istioctl
• Trying kubectl apply without correct manifest
• Assuming 'setup observability' is a valid command

3. Given the following Istio configuration snippet for telemetry, what will be the effect?

apiVersion: telemetry.istio.io/v1alpha1
kind: Telemetry
metadata:
  name: example-telemetry
spec:
  metrics:
  - providers:
    - name: prometheus
    overrides:
      prometheus:
        defaultHistogramBuckets: [0.1, 0.5, 1, 5]

medium

A. Prometheus will ignore histogram buckets and use defaults

B. Prometheus will collect metrics with custom histogram buckets 0.1, 0.5, 1, and 5

C. Telemetry resource will cause an error due to invalid syntax

D. Metrics will be sent to Jaeger instead of Prometheus

Solution

1. Step 1: Analyze the Telemetry resource configuration

   The snippet sets a Telemetry resource specifying Prometheus as the metrics provider and overrides histogram buckets to [0.1, 0.5, 1, 5].

2. Step 2: Understand the effect on Prometheus metrics

   This means Prometheus will collect metrics using these custom histogram buckets instead of defaults.

3. Final Answer:

   Prometheus will collect metrics with custom histogram buckets 0.1, 0.5, 1, and 5 -> Option B

4. Quick Check:

   Telemetry config with overrides = custom Prometheus buckets [OK]

Hint: Overrides in Telemetry change Prometheus buckets [OK]

Common Mistakes:

• Assuming default buckets remain unchanged
• Confusing metrics destination as Jaeger
• Thinking syntax is invalid without error

4. You deployed Istio with observability enabled but notice no traces appear in Jaeger UI. Which of the following is the most likely cause?

medium

A. The application logs are too verbose

B. Prometheus is not scraping metrics correctly

C. The Kubernetes cluster is out of storage

D. Istio sidecar proxy injection is missing on your application pods

Solution

1. Step 1: Identify cause of missing traces in Jaeger

   Jaeger receives traces from Istio sidecar proxies. If sidecars are missing, no traces are sent.

2. Step 2: Evaluate options for trace absence

   Istio sidecar proxy injection is missing on your application pods correctly identifies missing sidecar injection as the cause. Prometheus scraping affects metrics, not traces. Storage or log verbosity do not directly cause missing traces.

3. Final Answer:

   Istio sidecar proxy injection is missing on your application pods -> Option D

4. Quick Check:

   Missing sidecar = no traces in Jaeger [OK]

Hint: No Jaeger traces? Check sidecar injection on pods [OK]

Common Mistakes:

• Blaming Prometheus for trace issues
• Assuming storage issues cause missing traces
• Thinking log verbosity affects tracing

5. You want to monitor request latency across multiple microservices in your Kubernetes cluster using Istio and Prometheus. Which combination of configurations will best achieve this?

hard

A. Enable Istio sidecar injection, configure Prometheus scrape for Istio metrics, and use Grafana dashboards for latency visualization

B. Disable Istio sidecar injection and install Jaeger only

C. Use only Kubernetes native metrics without Istio or Prometheus

D. Configure Prometheus to scrape application logs directly

Solution

1. Step 1: Identify components needed for latency monitoring

   Istio sidecars collect telemetry data. Prometheus scrapes these metrics. Grafana visualizes latency metrics effectively.

2. Step 2: Evaluate options for best observability setup

   Enable Istio sidecar injection, configure Prometheus scrape for Istio metrics, and use Grafana dashboards for latency visualization combines sidecar injection, Prometheus scraping, and Grafana dashboards, which is the standard approach. Other options miss key components or use incorrect methods.

3. Final Answer:

   Enable Istio sidecar injection, configure Prometheus scrape for Istio metrics, and use Grafana dashboards for latency visualization -> Option A

4. Quick Check:

   Sidecar + Prometheus + Grafana = latency monitoring [OK]

Hint: Use sidecar, Prometheus, and Grafana for latency monitoring [OK]

Common Mistakes:

• Disabling sidecar injection breaks telemetry collection
• Relying only on Jaeger for latency metrics
• Scraping logs instead of metrics for latency