Redundancy between router and servers

Question

In the past, I was using a NetGear GS724T between my router and my servers. Unfortunatly the switch suddenly died.

Now I've changed it for a GS748T and also bought one more and I am wondering how I could achieve a redundancy using these two switches.

• How do I need to configure the router to communicate with the two switches : two ports in bridge mode?
• How to synchronize switches, do I need to configure STP and LAG ?
• What I need to do on the servers (HP DL380 G5 with two ethernet ports, Linux), can I plug one ethernet on each switch so the cable will be redundant too?
• If I plug the two ethernet ports from servers, will I be able to have same IP addresses on both interfaces? (context in my other question)

I know this is not a one answer question, but I still don't know where to start, more looking for pointers and/or guides.

Answer 1

I agree with @network_ninja but will extend it a bit.

How I'd solve this

Router1--L3--Router2
|              |
|              |
Switch1--L2--Switch2
|    |         |
|    |         |
PC1 PC2--------+

Router1 and Router2 are running VRRP, HSRP, GLBP or CARP to produce virtual default-GW IP address to the LAN.
This protocol will converse over the Switch core to agree which of the routers is owning the default-GW IP address at any given time.

PC2 is redundant linux server, which is using 'bonding' to redundantly connect to the Switches, it should be configured so that if the the virtual default-gw IP address stops responding to ARP WHO HAS, it'll switch to backup connection. IP address itself is not on the physical interfaces, but on the virtual bonding interface.
Equivalent solution is available to other OS, but often not included in base OS package.

PC1 is non-redundant server.

Switches are not running anything special, no spanning tree (as there is no L2 loop) and no LACP. They can be from different vendors and can be taken down for maintenance separately.

Routers are not running any switching, IP addresses are configured directly in the L3 interfaces facing the switches.
If you choose VRRP as your first-hop-redundancy-protocols, routers can be from different vendor. Each router can be taken down for maintenance separately, by gracefully switching VRRP priority before work on the primary.

Answer 2

I'll toss out for consideration what most will consider a bit more unorthodox solution.

Consider solving this with Layer 3 instead of Layer 2.

Put both switches into place, and DON'T interconnect them. Connect the router(s) to both switches. Connect your HP servers to both switches. Use two different IP blocks internally for the servers to talk to the router(s)...one block on each switch (and therefore interface on the router(s) and servers). Put the IP addresses that you actually use to communicate with the servers on a loopback interface. Put quagga on the servers and run OSPF (at your scale, just throw everything in area 0, no biggie)...make sure the loopback addresses/interfaces are included in the OSPF config. Put OSPF on the router(s).

Voila', the router(s) learn about the addresses that you're actually using to talk to the servers via OSPF as host routes...if a switch dies, the relevant adjancies go away and traffic gets rerouted over to the other switch.

As a bonus, if you use different IP address for the various services that you run on your Linux servers, you can move the services and their associated IP addresses around seamlessly and the network adapts cleanly and easily.

No danger in this setup of having bad behavior from a split-brain situation if the link between the two switches fail...no danger of bad behavior from a FHRP like VRRP, HSRP and the like...no danger of the switches falling back to inefficiently flooding traffic if you run into an asymmetric situation.

I use this solution in a much larger sort of environment and its works EXTREMELY well, is amazingly robust and resilient to both equipment failure and human configuration mistakes.

Answer 3

For the server side of things, if you run centos or similar, you would simply create a bond between the nics.

On windows, I believe you can do the same thing.

Then on the switch side, all of the cables going from the server to the switch would be part of the same vlan.

Another thing you can do if your switches support it, is configure VRRP so that your server uses a redundant gateway address.

Answer 4

As others have stated, the redundant server connections will be specific to the platform. Some operating systems have integrated nic redundancy mechanism and some require additional software, but most should be able to achieve active/passive dual nic redundancy.

If the router has an integrated switch (such as a Cisco switch wic) and support for stp (rapid preferred) then you can do what I call a single-attached U. You would connect an L2 port on the router to each switch and make the router the stp root. Connect the switches together (L2) and you have eliminated all single points of failure below the router.

Another solution would require L3 switches. You could connect an L3 port from the router to an L3 port on each switch. Then connect the switches together with an L2 port and an L3 port. Run FHRP between the switches and this will eliminate all single points of failure below the router without the need for a router-integrated switch.

Given two routers, there are some additional possibilities which can eliminate the router as a single point of failure.