How off-the-shelf open source software (i.e. Vmware ESXi, CentOS, nginx, haproxy, and keepalived) was used to deploy an effective web load balancing solution to deliver redundant and available web services.
2. Why I needed web load balancing Single web server Hosting 60+ websites Redundant subsystems (disk, power) Spare mobo, anyone?
3. Project Scope Website availability Website availability Websi…er…uptime! Redundant web servers Redundant load balancers Session state persistence for failed web backends not required (yet)
8. Lessons Learned Whoa! TSO HTTP 500 errors Web servers “offline” for 20-30 minutes IIS 7 request tracing did not help Disable TSO Windows: NIC properties (on by default for Windows guest VMs) Linux: ethtool –k <interface>