Management and Load balancing

We have many management options for Cisco ACE:

  • telnet/ssh, http
  • tacacs/radius for acces
  • XML API – you can create your own gui or tools for sending commands
  • For ACE4710 a build in gui is available
  • For ACE10/20/30 and ACE4710, ANM (application network monitoring, its just GUI) is available

Global server load balancing (GSLB)

GSS (global site selector) Device – Cisco older LB just for DNS.

GSLB utilizes both ACE and GSS working together to make load balancing more efficient. How this works? The GSS utilizes something called KAL-AP to communicated with the ACES. The ACE responds back with a  groupings of VIP and tells for each VIP a variable from 1 to 255 telling GSS how busy these VIPs are. The GSS has possibility to do also some other searchings for clients and based on that he decides where to send the request. Basically ping probe (RTT), if ping is blocked you can send syn to the ACE, how many servers are available behind the VIP, traceroute for amount of hops, GDB (check longitude and latitude). So it is just about to choose the right ACE based on something called proximity.

GSLB also adds another layer of fault tolerance. It is mostly about if one DC where 2 ACEs in failover mode failes you have prepared another DC with another pair of ACES. This is the additional layer GSLB brings to this. (But I think that it doesnt matter, because the weakest particle is of course GSS and when that is down you can have whatever number of ACES)