Understanding Service Degradation and Its Practical Strategies

What Is Service Degradation

Service degradation means disabling or simplifying non‑essential features when a system is under heavy load, similar to limiting the number of visitors in a scenic spot during holidays. In the Internet context, it ensures core services remain available by cutting off less important functions.

Service Level Definition

SLA (Service Level Agreement) is the key metric for judging whether a load test is abnormal. It represents the guaranteed uptime of a service, often expressed as "nines" (e.g., 99.9999% for six‑nines). Six‑nines corresponds to about 31 seconds of downtime per year.

Six‑Nines Meaning

Six‑nines = 99.9999% availability, which translates to roughly 31 seconds of service unavailability annually, indicating extremely high reliability.

Degradation Handling

Fallback Data

When a page fails, return fallback data such as default values (e.g., stock = 0), static content, or cached responses.

Rate‑Limiting Degradation

Set a maximum QPS threshold for each request type; requests exceeding the limit are rejected with friendly messages (e.g., "system busy, please try later"). This protects core services during traffic spikes.

Timeout Degradation

Define a timeout for remote calls; if a non‑critical request exceeds the timeout, degrade it by hiding optional data (e.g., product reviews) while keeping the main functionality intact.

Fault Degradation

If a remote service is unavailable (network, DNS, HTTP errors), return default values, fallback data, static pages, or cached results.

Retry / Automatic Handling

Client‑side high availability can be achieved by providing multiple service endpoints. In micro‑services, mechanisms like Dubbo retries or API call retries with a limit and idempotent handling are used. Web front‑ends may add retry buttons or automatic retries.

Degradation Switch

Operators can manually toggle a switch to disable problematic services. The switch can be stored locally or in external stores such as Redis, Zookeeper, or a configuration database, and is also useful for gray‑release rollbacks.

Crawler and Bot Handling

Detect rapid, repetitive actions to identify bots and serve static or cached pages instead of invoking backend services.

Read Degradation

When caches or DBs are unavailable, fall back to front‑end caches or static data. Strategies include temporarily switching reads to cache, disabling read endpoints, or serving static pages.

Write Degradation

During high write load, temporarily write to fast stores like Redis and later synchronize to the database, accepting eventual consistency to preserve availability.

Front‑End Degradation

When backend services are degraded, use local caches or dummy data on the client side, especially for low‑consistency scenarios such as flash sales.

JS Degradation

Embed degradation switches in JavaScript to stop sending requests once system thresholds are reached.

Ingress Layer Degradation

Use Nginx + Lua or HAProxy + Lua to filter invalid requests before they reach services, applying automatic or manual switches.

Application Layer Degradation

Configure feature flags within the application to enable automatic or manual degradation based on business needs.

In Spring Cloud, Hystrix provides circuit‑breaker and fallback mechanisms, allowing both manual and timeout‑based automatic degradation.

Fragment Degradation

If some page fragments fail to load (e.g., product listings), replace them with alternative data or omit them to keep the page functional.

Pre‑Embedding

Static data can be pre‑downloaded to devices before major events (e.g., Double 11) to reduce network load during peak times.

·END·