I'd like to ask for your help in order to validate a design of mine concerning the deployment of two Cisco CSS 11503 switches for load balancing, failover, and high availability.
Currently I am at the very first phase of this deployment and the layer 2 parameters of both CSS switches have already been configured and it is fully functional. The next step will be the load balancing and failover scenarios and I've a question about whether it can be done or not.
Please review the attachments in order to understand what the scenario deployment looks like.
The scenario description is as follows:
- There are two Cisco CSS 11503 switches
- There are three backend Servers
- These servers run the end customer application + an Oracle (Timesten) database
- Each server has the same content and the Oracle databases are fully synchronized
So how things are supposed to work? Let's see:
1) The high availability between the two Cisco CSS will be Active/Standby. This includes VIP redundancy and Virtual Interface Redundancy
2) The end users will communicate with the VIP that represents the application, initiating the connections to that VIP.
3) The active CSS will handle the request from the end user and will load balance (round-robin) the requests to one of the three available servers.
4) The server on the other hand will initiate a connection to the database, which might be its own copy of Oracle/Timesten or not (it depends on which of the three servers is holding the âmasterâ role).
5) So, therefore, the connection from the application to the database will be made through a second VIP and the Cisco CSS 11503 switches must NOT load balance in this case, since it could corrupt the entire database.
The following summarizes what we need to accomplish our goals: there will be two VIP addresses (one for the application and the second for the database), loadbalancing for the application VIP only, failover for the database VIP, and virtual redundancy to provide the servers with default gateway high availability.
I'm enclosing the initial configuration. I still need to insert the rest of the configuration parameters to fulfill my project's objectives. In order to do that, my questions are:
- How can I prevent the CSS switches from load balancing the requests from the application to the database? Load balancing must NOT occur as per the design's requirements.
- Since load balancing cannot be done, how would the CSS switches provide failover services in case of issues with the database on the master server?
Thanks for your reply. Fortunately I've read - for the most part - the manuals and several of the examples available at Cisco.com, that's how I managed to come up with my current config.
I am enclosing the up-to-date configs.
Basically, what I need to accomplish is:
1) End users will communicate with VIP 10.125.174.7 and the CSS will load balance these requests to the backend servers accordingly.
2) In addition to the VIP, there is a Virtual Interface (10.125.174.8) which will offer higher availability. Networking devices in front of these two CSS boxes will be configured to send packets to that Virtual Interface IP address.
3) One of the backend servers will handle the end user's request, and the server itself will initiate a connection to a database through another VIP (10.125.174.41).
4) As a matter of fact, this database is present in all of the backend servers (it is the same database spanned throughout three different servers (Oracle/Timesten thingâ¦), but there must be ONLY Master database. In other words, one of the servers will sync up with the others and they will establish a master/backup relationship.
So this is my major concern. The connections from the end users to the applications (the 1st VIP) can and will be load balanced. After that, the server will initiate a secondary connection to its database, which might not be its own (let's say that this particular server isn't the Master database server). The (master server) database will be reached through that second VIP (10.125.174.41) and the CSS switches must NOT load balance. The reason why is that it can and eventually will corrupt the databases.
I know it sounds like weird but this is how things are supposed to work. Application = load balancing and failover. Database = failover only.
Honestly I am not sure whether my proposed configuration will work and I can't test it right now. That's why I am asking for some advice!
I am seeking now for a true failover method and a way to implement VIP without load balancing, because apparently the only feature I need to the second VIP is the failover capability.
Moquery is the command line cousin of Vizore, it's very helpful and efficient sometimes during the troubleshooting. This article aims to provide moquery cheat sheet to the users for some most common seen scenarios.
Here is the checklist before customers/partners contact Cisco TAC:
Firmware Version of APIC and Switch
Download Switch and APIC techsupport logs
Problem description (Symptoms with details)
Business impact (eg, what kind of services...
moquery usageAPIC moquerySwitchmoquery
This document discuss a common issue observed during the VMM integration & VM workload migration to ACI fabric.
VMware Virtual machines are hosted in Cisco UCS-B seri...