October 23, 2022 7:00 PM PDT

This document summarizes the key points discussed during the Load Balancer Live Demo, focusing on the implementation and configuration of an L4 load balancer using LVS. The demo included setting up a virtual IP (VIP), configuring real servers, and ensuring high availability through the use of keepalived.

Presenter: Brother M, Tech Lead

System Design Summary

Load Balancer Type: L4 Load Balancer
Components Used:
- VIP
- LVS
- Tomcat servers (Tomcat1, Tomcat2)

Configuration Steps

Kernel Configuration

Change the kernel parameters on the destination machines:

Navigate to /proc/sys/net/ipv4/conf

Configure parameters for ens33, all, lo, and virbr0:

echo 1 > /proc/sys/net/ipv4/conf/ens33/arp_ignore
echo 2 > /proc/sys/net/ipv4/conf/ens33/arp_announce
echo 1 > /proc/sys/net/ipv4/conf/all/arp_ignore
echo 2 > /proc/sys/net/ipv4/conf/all/arp_announce

Map the VIP to the loopback device on Tomcat servers.

Load Balancer Configuration

Bind the VIP on the load balancer:
```
ifconfig
```
- The load balancer will have one network card with two IP addresses.
Start Apache servers on real servers:
- Configure each Apache server to return a static page indicating its identity (server1 or server2).
Confirm accessibility of real servers by visiting their real IP addresses.

LVS Load Balancer Configuration

Configure round robin for incoming packets:
```
ipvsadm -A -t 192.168.1.250:80 -s rr
```

Configure outgoing packets to direct to different real servers:

ipvsadm -a -t 192.168.1.250:80 -r 192.168.1.12 -g
ipvsadm -a -t 192.168.1.250:80 -r 192.168.1.13 -g

Test the load balancer by curling the webpage from the VIP:
- Confirm that requests are distributed to real servers in a round-robin fashion.

High Availability Configuration

Issue: L4 load balancer LVS is a single point of failure.
Solution: Implement master and backup configurations.
- Multiple backups can be set up.
- Backups monitor the health of the master and real machines.
- The master sends heartbeat signals.

Set up keepalived on master and backups:
- Configuration file: /etc/keepalived/keepalived.conf
- Use virtual_server for persistent timeout settings.
Start keepalived:
- The master broadcasts its status.
- Keepalived uses HTTP requests to check the health of real servers and auto-configures the load balancer accordingly.

Testing Failover Mechanisms

Test the backup mechanism by shutting down the network card of the main server.
The backup server should take over upon detection of the main server's failure.
When the original master is restored, it will reclaim the master role.

Additional Testing

Easier to test network configurations using curl instead of a browser.
Test the scenario of taking down a real server; the load balancer configuration should automatically exclude the down server.

HTTPS Configuration

Keepalived can utilize SSL to test the health of real servers.

Summary

The flow of requests is as follows:

User -> DNS -> L4 LB -> L7 LB -> Tomcat Real Servers

Deeper layers provide more features but may introduce latency.
An API gateway is one example of an L7 load balancer.