White Paper
®
Designing Switches & Routers with APEX CAM
Introduction
Content-addressable memory (CAM) allows a design to quickly search a table for a particular item. The data is
supplied by the user or the system, and CAM uses the information provided to determine the location of the requested
data.
CAM is especially useful in speeding up search operations in switches and routers. This white paper describes how
APEX™ CAM can be used in specific applications such as:
■
■
■
■
■
Data compression
Network switching
Internet protocol filtering
Internet protocol address resolution
Multi-protocol label switching.
Data Compression
Data compression increases system bandwidth by removing redundant information and allowing a faster flow of
traffic. Most compression algorithm time is spent searching and maintaining data structures to manage the
compression; however, replacing the data structure with a hardware search engine can increase the throughput of the
algorithm.
In data compression, a shorter bit of data can replace a longer bit of data. For example, an eight-bit number can be
respresented with a two-bit number. These conversions are translated in CAM. The eight-bit number is compared
against the data in CAM, and the addresses in the CAM block are two-bit condensed numbers. In Figure 1, CAM
receives the number 42 as data. This number is compared against all the data in the CAM block, then CAM finds the
compressed result.
Figure 1. Using CAM for Data Compression
CAM Contents
Uncompressed
Data
Compressed
Data
42
0
59
1
83
2
42
0
83
2
27
3
Data
Address
42
0
59
1
83
2
27
3
59
1
Sequence
Token
M-WP-APEXCAM-02
October 2000
1
Designing Switches & Routers with APEX CAM White Paper
Altera Corporation
In one compression algorithm, commonly recurring sequences are replaced by tokens. Tokens are shorter than
sequences, compressing the data stream. When a word is driven into the CAM block as data, the CAM block returns
the address where the word was found. This address is now the token for the sequence. If CAM finds the code
corresponding to the word-bit pattern in the input register, then it clears the appropriate symbol or token output and
input register. If the code is not found in CAM, then another word is shifted in. The word that was not found can be
stored with an escape code to indicate that it was not compressed. CAM generates a result in a single transaction,
regardless of table size or length of a search list. This makes CAM ideal for data compression schemes that use
sparsely populated tables as part of their algorithm.
RAM in another device can be used to uncompress the compressed data or token from the first device’s CAM block.
In RAM, the compressed data or token is saved as the uncompressed data’s address. Figure 2 illustrates the
compression and decompression of data in two devices through CAM and RAM.
Figure 2. Compression & Decompression in Two Devices through CAM & RAM
Device 1
Device 2
CAM
Address
Data
Uncompressed
Data
RAM
Token
Address
Data
0
142
142
0
259
1
1
259
833
2
2
833
247
3
3
247
Uncompressed
Data
Network Switch
CAM can be used in switch applications to extract and process address information from incoming data packets.
CAM compares the stored destination address in order to switch the packet to the correct outgoing port. In Figure 3, a
network switch routes incoming packets to one of four destinations based on ethernet addresses and switch port
numbers stored in CAM. CAM then compares the address of each received packet against the stored addresses and
port numbers. If the comparison yields a match, routing control forwards the packet to the proper destination port.
Figure 3. Using CAM as a Network Switch
Ethernet Address
19:07:4A:3B:CA:E1
Ethernet Address
07:20:1A:3F:08:28
1
Ethernet Address
03:65:A1:4B:09:7F
0
2
Switch
from Internet
3
Ethernet Address
74:61:12:E4:C5:42
CAM Contents
Port
2
Address
Data
0
03:65:A1:4B:09:7F
1
07:20:1A:3F:08:28
2
19:07:4A:3B:CA:E1
3
74:61:12:E4:C5:42
Ethernet Address
Altera Corporation
Altera Corporation
Designing Switches & Routers with APEX CAM White Paper
IP Filter
An IP filter restricts unauthorized local area network (LAN) access or traffic on a wide area network (WAN).
Restricting Internet traffic across a LAN allows the administrator to give the LAN workstations permission to specific
Internet-based applications, such as e-mail. Filters can be implemented in both software and hardware to improve
performance and control security.
CAM can improve the performance of the IP filter by storing the source or destination addresses of packets that are
allowed to pass along the network. When CAM receives the address information, it compares the address to the list of
stored addresses. If a match is found, the packet is sent through to its destination. If no match is found, the packet is
not transmitted. Figure 4 displays an example of this process.
Figure 4. CAM as an IP Filter
Addresses Stored in CAM
CAM
Status
Packet Address
Address
192.2.41.53
Pass
192.2.41.53
0
192.21.42.3
Pass
192.63.12.3
1
192.57.11.101
Denied
192.21.42.3
2
192.57.12.144
Denied
Data
If more than one application needs to access the LAN, a combination of CAM and RAM can perform a multiple
application filter. CAM stores the source or destination address of a packet, and RAM stores the permissions. Each bit
of RAM stores one specific permission.
Figure 5 shows an example with requested packets having explicit permission to access the Internet and e-mail.
Figure 5. Multiple-Permission IP Filter
192.63.12.3
Requested
Data
Address
Data
Address
Web
192.2.41.53
0
192.63.12.3
1
192.21.42.3
2
192.57.11.101
3
...
Altera Corporation
Web &
E-mail
Permission
RAM
CAM
...
E-mail
FTP
Telnet
1
0
1
1
1
1
1
1
0
0
2
0
0
0
1
3
1
0
1
0
...
...
...
...
...
3
Designing Switches & Routers with APEX CAM White Paper
Altera Corporation
Internet Protocol Address Resolution
Internet protocol address resolution is used by layer 3 switches to convert Internet protocol addresses to Ethernet or
media access control (MAC) addresses and vice versa. An Internet protocol address is a 32-bit value that identifies
each individual device on a network that is able to send or receive data. When a source attempts to send data over the
Internet, the Internet protocol portion of the transmission control protocol/Internet protocol (TCP/IP) includes the
source Internet protocol address and the destination Internet protocol address in the packet. Once all the required
information for delivery is included in the packet, the TCP/IP stack sends the entire packet across the Internet to the
destination. The destination receives the data, and a response can be sent to the source by using the existing Internet
protocol address.
A MAC address, or Ethernet address, is the unique physical address designated to a hardware device (such as a
personal computer) when it is manufactured. When a computer is connected to a LAN, a correspondence table relates
the Internet protocol address of the device to its MAC address. A combination of CAM and RAM can implement a
table to maintain a correlation between each MAC address and its corresponding IP address and perform address
conversion in both directions. CAM contains the Internet protocol addresses, that point to corresponding locations in
RAM, which contains the MAC or Ethernet address.
Address resolution protocol (ARP) is used to map an Internet protocol address to a recognized MAC address in the
local network. When a host machine receives a packet, ARP runs a search to verify that the packet’s destination
address is on its particular network. The CAM and RAM combination, implemented in the ARP, compares the
Internet protocol address against the content of the CAM table to find the appropriate MAC address in RAM. If CAM
contains the Internet protocol address, the packet can be converted to the appropriate length and format, then sent to
the destination device. If no match is found, ARP broadcasts a request packet to all the devices on the LAN in search
of the corresponding address. If a device recognizes the Internet protocol addresses, it sends a message to the ARP
claiming the Internet protocol address. ARP updates the contents of CAM and RAM for future reference and sends
the packet to the destination machine. Figure 6 shows this Internet protocol address resolution process. This process
can be implemented in software; however, implementing the search in hardware improves system performance.
Figure 6. Internet Protocol Address Resolution
Internet
Protocol
Address
Address
CAM
MAC
Address
RAM
Data
Address
Address
192.2.41.53
0
0
92:ad:26:7e:81:ae
192.63.12.3
1
1
12:84:ea:72:9e:bd
192.21.42.3
2
2
ae:b3:f4:87:2e:81
Data
Multi-Protocol Label Switching
Multi-protocol label switching provides a solution that combines the control of IP routing with the simplicity of
layer 1 switching. Figure 3 shows an example of layer 1 switching. Multi-protocol label switching is a new
routing/forwarding technique for the core of the Internet and supports the use of advanced routing services. This
solves problematic issues such as the quality of services and delivery of new routing capabilities. Figures 7 and 8
show multi-protocol label switching by indicating that routers on the edge or in the core of the network have seperate
functions.
4
Altera Corporation
Altera Corporation
Designing Switches & Routers with APEX CAM White Paper
Figure 7. Multi-Protocol Label Switching Network
In Core:
-Forwards Packet Using Labels
-Label indicates service
class and destination
At Edge:
Classify and Packets
Router
Router
Router
Switching
Router
Router
Edge LSR:
Edge Label
Switching Router
Switching
Router
Switching
Router
Switching
Router
Router
Core LSR:
Label Switching Router
Router
Router
A multi-protocol label switching system is composed of two functional components: control and forwarding. The
control component uses a standard routing protocol to exchange information and maintain a forwarding table with
other routers along the network. When a router receives a packet, the forwarding component searches the forwarding
table, which is maintained by the control component, to make routing decisions for each packet. The forwarding
component is based on a label-swapping forwarding algorithm. Each control component is responsible for assigning
and distributing a set of labels, as well as maintaining other relevant control information. Because multi-protocol
label switching allows different modules to assign labels to packets, it decouples the forwarding of a packet from the
contents of the packet’s IP header.
At the ingress edges (entry) of the network, each incoming packet is classified and an initial label is assigned. The
label switch performs a “longest-match routing” table lookup, assigns a label to the packet, then forwards the packet
to the next hop on the label-switched path. In the core of the network, when a labeled packet reaches a switch, the
forwarding component uses the input port number and label to perform an exact match search from its forwarding
table. If a match is found, the forwarding component retrieves the outgoing label, the outgoing interface, and the next
hop address from the forwarding table. The forwarding component then replaces the incoming label with the outgoing
label, then directs the packet to the outbound interface for transmission to the next hop in the label-switched path.
When the packet reaches the egress edge, the forwarding component searches its forwarding table. If the next hop is
not a label switch, the egress switch discards the label and forwards the packet using conventional longest-match IP
forwarding. Figure 8 shows the basic multi-protocol label switching operation.
CAM can implement the required table and perform a fast search operation in the multi-protocol label switching
operation. A CAM block finds the appropriate label for the incoming packet by searching the lookup table in the
ingress label switch. The label is used as an index of a table, which specifies the next hop and the new label. The
packet is forwarded to its next hop with the label attached. At the last edge of the network or egress label switch
section, a CAM block can efficiently implement the table to remove the label from the incoming packet and forward
the packet using IP forwarding.
Altera Corporation
5
Designing Switches & Routers with APEX CAM
Altera Corporation
A combination of CAM and RAM can be used to implement the table, which routes the incoming labeled packet to its
next hop and finds the new label. Typical multi-protocol label switches store up to 1,024 labels at a time, requiring a
1,024 × 32 CAM block. This CAM block requires only 32 embedded system blocks (ESBs) and can be efficiently
implemented within APEX devices. Because the out-label is another label, RAM consisting of 1,024 × 32 bits
consuming 16 ESBs can assign the appropriate label to the incoming label. A total of 48 ESBs can perform the
multi-protocol label switching operation (see Figure 8).
Figure 8. Multi-Protocol Label Switching Basic Operation
Edge
Core
Core
Ingress Label Switch
192.3.4.1
IP
Address
Out
Label
CAM
192.4.16
Core
Label Switching
18
In Label
Label Switching
Out Label
23
In Label
CAM
RAM
CAM
Address Data
Data Address
18
18
5
5
23
23
Egress Label Switch
Out Label
Data Address
Edge
4
In
Label
Next
Hop
192
RAM
0
Address Data
CAM
192.2.3.1
4
0
4
Assign Initial Label
Remove Label
Label Switched Path
IP
Header
Data
MPLS
Label
IP
Header
Data
MPLS
Label
IP
Header
Data
IP
Header
Data
Conclusion
Many networking applications require a fast search operation. APEX CAM can provide a solution for search
operations in critical applications such as data compression, network switching, Internet protocol filtering, Internet
protocol address resolution, and multi-protocol label switching. Using CAM simplifies and accelerates these
functions and applications by increasing the performance of table look-up and translation implementation.
®
101 Innovation Drive
San Jose, CA 95134
(408) 544-7000
http://www.altera.com
6
Copyright  2000 Altera Corporation. Altera and APEX are trademarks and/or service marks of Altera Corporation in the United States and
other countries. Other brands or products are trademarks of their respective holders. The specifications contained herein are subject to change
without notice. Altera assumes no responsibility or liability arising out of the application or use of any information, product, or service
described herein except as expressly agreed to in writing by Altera Corporation. Altera customers are advised to obtain the latest version of
device specifications before relying on any published information and before placing orders for products or services. All rights reserved.
Altera Corporation