High Availability (HA)

1
High Availability (HA)

• Goal
• a solution for reliability, availability, and
  serviceability in Linux.
• Focus on the services in the network.
• Use the Piranha in the experiment.

2
Piranha - Red Hat Inc.-

• FailOver Service (FOS)
• 2-node clustering with a floating IP.
• Linux Virtual Service (LVS)
• Include FOS.
• More than 2 nodes in a cluster.
• Scheduling.
• Load-balancing.

3
An Overview of FOS

• 2-node clustering active and standby.
• FailOver
• the service becomes unresponsive to FOS
  monitoring.

4
Features

• Possible service
• http, ftp, inetd, telnet, smtp, ssh, LDAP,
  firewall, other IP services.
• Not supported service nfs, ntp.
• pop, imap, smtp need a NFS-mounted system.

5
FOS Architecture
Client PC Accessing web server at 140.113.88.9080
Eth0 140.113.88.93
Eth0 140.113.88.87
Eth0 140.113.88.84
Eth01 140.113.88.90
Active Node Web service running On port 80
Inactive Node Monitors 140.113.88.8780
PRIMARY
BACKUP
6
FOS Architecture
Client PC Accessing web server at 140.113.88.9080
Eth0 140.113.88.93
Eth0 140.113.88.84
Eth0 140.113.88.87
Eth01 140.113.88.90
Active Node Monitors 140.113.88.8480
Inactive Node Web service running On port 80
PRIMARY
BACKUP
7
Primary and Backup Nodes

• An FOS cluster always have a primary node.
• Both nodes declare themselves as active primary
  node will always win.

8
FOS States
Heartbeat
pulse
pulse
fos -- active
fos -- monitor
Monitor
ftp service port 21
nanny
Monitor
Web httpd port 80
nanny
Active System
Inactive System
9
Conventional High Availability Concept
Administration/ Configuration
Application
HA Software
Operating System
I/O Drivers
HW