Redirection and Load Balancing

Herng-Yow Chen

Outline

HTTP redirection DNS redirection Anycast routing Policy routing IP MAC forwarding IP address forwarding

Outline (cont.)

The Web Cache Coordination Protocol (WCCP) The Intercache Communication Protocol (ICP) The Hyper Text Caching Protocol (HTCP) The Network Element Control Protocol (NECP) The Cache Array Routing Protocol (CARP) The Web Proxy Autodiscovery Protocol (WPAD)

Why Redirect? Redirection is a fact of life in the modern Web

because HTTP applications always want to do three things:

Perform HTTP transactions reliably Minimize delay Conserve network bandwidth

For these reasons, web content often is distribute in multiple locations.

Redirection can be considered as a set of techniques that help to find “best” distributed content.

Redirection vs. load balancing Most redirection deployments include some form of load

balancing. Conversely, any form of load balancing involved redirection

techniques.

Where to redirect General techniques: many redirection techniques

work for web servers, proxies, caches, and gateways because of their common, server-like traits.

Specialized techniques: specially designed for a particular class of endpoint.

Web servers handle requests on a per-IP basis. Distributing requests to duplicate servers means that each request for a specific URL should be sent to an optimal web server (the one nearest to the client, or the least-loaded one, or some other optimization).

Proxies tend to handle requests on a per-protocol basis, all HTTP traffic in the neighborhood of a proxy should go through the proxy.

Overview of redirection protocols

The goal of redirection is to send HTTP messages to available web servers as quickly as possible. Several mechanisms can be provided for redirection: The browser application’s proxy configura

tion. (only for redirecting traffic to proxies) HTTP redirections. DNS resolvers. (can be used for redirecting t

raffic to any server). TCP/IP Routing (e.g. Routers and switches)

HTTP redirections

How it works (next slide)

Basis for rerouting (many options) From round-robin load balancing, to minimizing latency, To choosing the shortest path

Limitation Can be slow – every transaction involves the

extra redirect step. Also, the first server must be able to handle the request load

HTTP Redirections

Alice

Bob

Internet161.58.228.45

161.58.228.46161.58.228.47www.joes-hardware.com

(a)Alice sends HTTP request to www.joes.hardware.com

(b)Server returns 302redirect to 161.58.228.45

(c) Browser resends HTTP request,this time to

161.58.228.45

Alice

BobInternet

161.58.228.45161.58.228.46

161.58.228.47www.joes-hardware.com

(d)Bob sends HTTP request to www.joes.hardware.com

(e)Server returns 302redirect to 161.58.228.46

(f)Browser resends HTTP request,this time to 161.58.228.46

HTTP/1.0 302 RedirectionLocations: http://161....

HTTP Redirection (cont.)

Several disadvantages A significant amount of processing power

is required from the original server. (Sometime almost as much server horse-power is required to issue the redirect as would be to serve up the page itself.)

User delays are increased, because two round trips are required to access pages.

If the redirecting server is broken, the site will be broken.

DNS redirections

How it works (next slide) Try nslookup ccsun.ncnu.edu.tw, www.yaoo.com

Basis for rerouting (many options) From round-robin load balancing, to minimizing latency, To choosing the shortest path

Limitation Need to configure DNS server

DNS Redirection

router

Backbone network

Switch

Edge network

www.joes-hardware.com

www.joes-hardware.com

www.joes-hardware.comwww.joes-hardware.com

10.10.10.3

Server3

10.10.10.2

Server2

10.10.10.1

Server1

10.10.10.4

Server4

DNS server

Decides whether to resolve to

10.10.10.1,

10.10.10.2,

10.10.10.3,

10.10.10.4

DNS round robin for load balancing

Alice

Bob

Internet207.25.71.5

207.25.71.6207.25.71.7DNS Server

(a)Alice asks DNS for IP address of www.cnn.com

(b) DNS replies with 207.25.71.5

(c) Alice sends HTTP request to 207.25.71.5

Alice

BobInternet

DNS Server

(d)Bob asks DNS for IP address of www.cnn.com

(eDNS replies with 207.25.71.6

(f) Bob sends HTTP request to 207.25.71.5207.25.71.5

207.25.71.6207.25.71.7

The impact of DNS caching

DNS address rotation spreads the load around, because each DNS lookup to a server gets a different ordering of server addresses.

However, this load balancing isn’t perfect, because the results of the DNS lookup may be memorized and reused by applications, operating systems, and some primitive child DNS servers.

Many web browsers perform a DNS lookup for a host but then use the same address over and over again, to eliminate the cost of DNS lookups and because some servers prefer to keep talking to the same client.

The impact of DNS caching Furthermore, many OSs perform the DNS look

up automatically, and cache the result, but don’t rotate the addresses.

Consequently, DNS round robin generally doesn’t balance the load of a single client – one client typically will be stuck to one server for a long time.

However, it can spread the aggregate load of multiple clients. As long as there is a modestly large number of clients with similar demand, the load will be relatively well distributed across servers.

Other DNS-based redirection algorithms

Load-balancing algorithms DNS servers keep track of the load on the web

servers and place the least-loaded web servers at the front of the list.

Proximity-routing algorithms DNS servers can attempt to direct users to

nearby web servers, when the farm of web servers is geographically dispersed.

Fault-masking algorithms DNS servers can monitor the health of the

network and route requests away from service interruptions or other faults.

DNS request involving authoritative server

DNS-based redirection (cont.)

Typically, the DNS server that runs sophisticated server-tracking algorithm is an authoritative server that is under the control of the content provider.

Several distributed hosting services use this DNS redirection model.

One drawback The authoritative DNS serer uses to make its

decision is the IP address of the local DNS server, Not the IP address of the client.

Anycast Addressing

router

Backbone network

Switch

Edge network

www.joes-hardware.com

www.joes-hardware.com

www.joes-hardware.com

10.10.10.3

Server3

10.10.10.2

Server2

10.10.10.1

Server1

router

router

IP MAC Forwarding

Client MAC 1

Client MAC 2

Hub MAC3 Switch MAC4

Gateway MAC5

To Internet

IP MAC Forwarding

Client MAC 1

Client MAC 2

Hub MAC3 Switch MAC4

Caching proxy MAC6

To InternetGateway

MAC5Port 80 traffic

IP Address Forwarding

router

Backbone network

Switch

Edge network

Joe’s edge network

Joe’s server

Destination proxy

IP Address Forwarding

HTTP data ■─────────────

─────────────

─────────────

HTTP data ■─────────────

─────────────

─────────────

Form:Client:

1.1.1.1

80 To:Joe’s Server

2.2.2.2

80

Form:NAT device:

3.3.3.3.

80To:Proxy

4.4.4.4

80

Passes through network address translation (NAT) device

Proxy Auto-configuration

PAC file autodiscovery

GRE packet encapsulation

HTTP data ■─────────────

─────────────

─────────────

Form:Client:

1.1.1.1

80 To:Joe’s Server

2.2.2.2

80

Passes through WCCP router

HTTP data ■─────────────

─────────────

─────────────

Form:Client:

1.1.1.1

80 To:Joe’s Server

2.2.2.2

80

To Proxy:

3.3.3.3

8080

Proto:GRE

Cache Array Routing Protocol

Cache Array Routing Protocol

Hyper Text Caching protocol