In Cisco ACE configuration Real servers are dedicated physical servers that you typically configure in groups called server farms. Server farms are groups of networked real servers that contain the same content and that typically reside in the same physical location in a data center. Web sites often comprise groups of servers configured in a server farm. Load-balancing software distributes client requests for content or services among the real servers based on the configured policy and traffic classification, server availability and load, and other factors. If one server goes down, another server can take its place and continue to provide the same content to the clients who requested it. This document describes steps to use to add an extra device to the server farm and in case of failure fall back to the original server farm.
The existing server farm has 3 operational servers configured and working properly. You need to add an additional device, like cache server, so that if the cache server (let's call this vc130) is up and running, it should send all traffic to this server. If the cache server is down, on not replying on port 80, it should fall back to this current scenario.
probe tcp PROBE_TCP port 80 interval 5
rserver host ns123 ip address 172.16.0.50 conn-limit max 50000 min 40000 inservice rserver host ns124 ip address 172.16.0.40 conn-limit max 50000 min 40000 inservice rserver host ns125 ip address 172.16.0.30 conn-limit max 50000 min 40000 inservice
Configure a new serverfarm, add real servers etc and put that serverfarm as backup serverfarm under L7 policy map. So, for example, you have serverfarm "test" which you want as back up.
policy-map type loadbalance http first-match WEB_L7_POLICY class class-default sticky-serverfarm web-sticky backup test -----> This is where you will specify the backup serverfarm. insert-http x-forward header-value "%is"
You will have probes etc in the same way as you have for regular serverfarm. You can also use backup command under sticky group and specify the backup serverfarm there with sticky keyword. But remember if you have sticky configured for backup serverfarm, even when the primary serverfarm comes back as "operational", the connections which have sticky entry in DB for servers in backup serverfarm, will continue to be served by the backup serverfarm.
For example if your primary serverfarm is APP1-SF1 & Backup serverfarm is APP1-SF2 you need to configure the following.
sticky ip-netmask 255.255.255.255 address source APP1-sticky timeout 60 serverfarm APP1-SF1 backup App1-SF2 sticky ---> This will ensure that sticky is applied to backup serverfarm too
Dear Experts, One of our customer is evaluating DCI to connect 2 sites across L3 mpls circuit. For DCI they are considering Vxlan (only open standard no OTV or juniper specific etc). From design perspective what i dont understand is, its easy t...
Hi all, we've been using a hardware version of the ACI Simulator for quite some time and really grew fond of it for testing automation solutions we developed for our production fabric. Recently we decided to switch to VM. Our Production-Fabric runs o...
Hello, We would like to change the airflow direction (and thus the fan module) on a CISCO C3850-24-XS-S switch.Is it possible on this type of switch ? We have also seen that the CISCO C3850-48-XS-S had two fan modules (FAN-T3-R= & FAN-T3-F=)...
Hello ACI Gurus. I am currently migrating a two sets of Palo Alto Physical firewalls directly counted to old Cisco 6509 switches to ACI. The way current environment is communicating between ACI and legacy 6509 switches is via a L2 link with a S...