---
title: "Proxy load balancing benefits; non-redundant array"
subject: "Internet and Intranet"
chapter: 5
key: "semester_8/internet_and_intranet:proxy-lb-benefits-non-redundant"
source: https://www.bctengineer.com/q/semester_8/internet_and_intranet/proxy-lb-benefits-non-redundant
site: BCT Engineering
---

# Proxy load balancing benefits; non-redundant array

Internet and Intranet · Chapter 5: Designing Internet Systems and Servers

- Asked in: 2070 Bhadra, 5+6 marks
- Video: [What is a Load Balancer?](https://www.youtube.com/watch?v=sCR3SAVdyCc) — IBM Technology
- Answer checked by Claude Opus 5.5 (high effort) on 2026-10-14
- Page: https://www.bctengineer.com/q/semester_8/internet_and_intranet/proxy-lb-benefits-non-redundant
- Whole chapter: https://www.bctengineer.com/md/semester_8/internet_and_intranet/chapter-5

## Question

What are the benefits of proxy load balancing? Discuss non-redundant proxy array load balancing technique with its features.

## Short answer

Proxy load balancing improves performance, scalability, availability, effective cache, bandwidth saving, maintenance, and cost. Non-redundant array (CARP) hashes each URL to exactly one proxy, so no duplicate copies and total cache is the sum; it has high hit rate, no inter-proxy traffic, but no redundancy if a proxy fails.

## Answer

#### Benefits of proxy load balancing

1. **Better performance.** Requests are spread over several proxies, so no single server is overloaded and response time falls.
2. **Scalability.** Capacity grows by adding more proxy servers without changing client settings.
3. **Higher availability (fault tolerance).** If a proxy fails, requests go to the remaining ones.
4. **Larger effective cache.** With hash-based distribution the disks of all proxies add up, giving a higher hit ratio and less Internet traffic.
5. **Bandwidth saving.** More hits mean fewer fetches from the origin servers over the costly WAN link.
6. **Easy maintenance.** A proxy can be taken out for upgrade without a service stop.
7. **Cost effective.** Several inexpensive PCs can replace one costly large server.

#### Non-redundant proxy array load balancing

In a **non-redundant array**, the URL space is partitioned among the proxies, so every object is cached in **exactly one** proxy (no duplicates). The partitioning is done by a hash function, as used in CARP.

**Working**
1. The client (or front end) computes a hash of the requested URL combined with each proxy's name.
2. The proxy with the highest score receives the request.
3. If it has the object, it returns it (hit). Otherwise it fetches it from the origin server, caches it and returns it (miss).

```text
 URL --> hash --> P2 only
 Clients ----> [P1] [P2] [P3] [P4] ---> Origin servers
              (each holds a different slice of URLs)
```

**Features**
- No duplicate copies: total cache size = sum of all proxies.
- High hit rate and efficient disk use.
- No communication between proxies (no ICP queries), so low overhead.
- Deterministic: the same URL always goes to the same proxy.
- Scales easily; adding or removing a proxy re-maps only about 1/n of the URLs.
- Weak point: no redundancy. If a proxy fails, its cached objects are lost (cold cache, objects must be fetched again), though requests are re-hashed and service continues.
- Needs the hash to be known to the clients (a PAC script) or a front-end device.

## Check yourself

The answer's key points as multiple-choice questions, with the correct option and why.

1. Why does hash-based distribution give a larger effective cache?
   A. Each proxy caches everything
   B. Objects are compressed
   C. The disks of all proxies add up because each object is stored once
   D. Cache size is fixed by the origin server
   Answer: C. With no duplicates, total cache size equals the sum of all proxies, raising the hit ratio. (Key point: Larger effective cache)

2. What does non-redundant proxy array mean?
   A. Every object is cached in exactly one proxy
   B. Every object is cached in two proxies
   C. The array has only one proxy
   D. Proxies never fail
   Answer: A. The URL space is partitioned so each object lives in one proxy, with no duplicate copies. (Key point: Non-redundant meaning)

3. How is the proxy for a request chosen in the working described?
   A. The proxy with the fewest connections
   B. Randomly
   C. In rotation, one after another
   D. The proxy with the highest score from hashing the URL combined with each proxy's name
   Answer: D. The URL hash is combined with each proxy name and the highest score wins, so the same URL always goes to the same proxy. (Key point: CARP selection)

4. What is the weakness of a non-redundant array when a proxy fails?
   A. Service stops completely
   B. Its cached objects are lost and must be fetched again, though requests are re-hashed and service continues
   C. All proxies lose their caches
   D. Clients must reboot
   Answer: B. There are no copies, so the failed proxy's cache is lost (cold cache), but re-hashing keeps service running. (Key point: Weak point)

## Related in Internet and Intranet

- [Proxy array load balancing types for big ISP](https://www.bctengineer.com/question.md?key=semester_8%2Finternet_and_intranet%3Aproxy-array-lb-mechanisms-isp) — Internet and Intranet, chapter 5
- [Short note: Proxy load balancing](https://www.bctengineer.com/question.md?key=semester_8%2Finternet_and_intranet%3Anote-proxy-load-balancing) — Internet and Intranet, chapter 5, asked 2×
- [RADIUS; proxy load balancing with CARP](https://www.bctengineer.com/question.md?key=semester_8%2Finternet_and_intranet%3Aradius-proxy-lb-carp) — Internet and Intranet, chapter 5

## Related in other subjects

- [Features of end-fire array](https://www.bctengineer.com/question.md?key=semester_6%2Fpropagation_and_antenna%3Aendfire-array-features) — Propagation and Antenna, chapter 2