3

Suppose there is only one switch within a system and it is connected to a bunch of machines to form a network. If this switch goes down, the entire network goes down. Chances of this happening are unlikely. I would like to know... what are the ways to increase the availability of these machines?

Would I need to have a second switch connected to each machine (so each machine needs 2 network adapters)? Or could I somehow obtain a switch that is able to failover to another backup switch? Is there other recommended ways to do this?

Thank you for your guidance.

O_O
  • 635
  • 3
  • 15
  • 25
  • 3
    Having a switch that failed over to another switch of be a neat trick, you'd have to beam the cable over to the new switch. – Jim B Jan 29 '16 at 01:46
  • 1
    Living near a bestbuy help if you need one in urgency, and most of all keep your old switch when you upgrade, a 100baset from a closet can save your ass even if an antiquity. For the failover idea its more something for the backbone switch. (where all the switch uplink to) – yagmoth555 Jan 29 '16 at 01:53
  • As we say when you sign up serverfault is for those 'managing information technology systems in a business environment' - having a single switch is considered unprofessional and wouldn't fit this site. Perhaps our sister site superuser - which is more for lab/hack environments - would be more appropriate. – Chopper3 Jan 29 '16 at 12:43

3 Answers3

4

Very interesting question. The more I think of an answer to this problem them more road blocks I am hitting.

Here is a few possible solutions that I can think of:

Two NICs in Each Machine + Two Switches

This is obviously what you suggested but you might run into problems with each computer on your network taking 2 IP address's. What I would suggest for this scenario is to have the computers connected to both switches but make sure you only have one switch on at a time. So if one fails you can manually flick the second switch on with minimum down time.

Problems:

  • You'll have to remember to keep a copy of any changes made to the config of the first switch so you can duplicate them on the second switch.
  • Someone in the know will have to be physically present to do the change over.

Keep a Spare Switch

If you are going to have two swtiches, instead of going through the effort of buying a second NIC for each computer and then running the cable, just keep a spare switch that has the exact config. You could easily rack mount it or sit it on top of the production switch. If the switch goes down, just unplug one port at a time from faulty switch to backup switch.

Problems:

  • If you have tons of computers on the network this may take a long time, or not be possible space wise
  • You'll have to remember to keep a copy of any changes made to the config of the first switch so you can duplicate them on the second switch, or make the changes on both
  • Someone will have to be present to do the change over

Failover Switching

This is much more complicated than the above but companies like Cisco have Failover features on some switches. I would images you would have to get the right switches that support it. Have a read of these links:

This stuff is over my head but might give you some ideas.

tyelford
  • 255
  • 2
  • 11
  • A decent backup switch will duplicate all the config and failover automatically. There shouldn't be a need to "remember" a config, or it defeats the point of the exercise. – Ken Sharp Jan 29 '16 at 02:45
  • Can you set up some sort of link aggregation? – user253751 Jan 29 '16 at 08:22
  • Ad problem with two IP adresses - you can setup failover bond across those links, so workstation will have only one IP adress and failover will works automaticly. You should look on STP settings of network. – Ondra Sniper Flidr Jan 29 '16 at 12:59
  • If your software can do automatic failover between IPv4 and IPv6, the solution may be as simple as running IPv4 on one switch and IPv6 on the other. Implementing RFC 6555 or equivalent on the client side of every communication in your system should be sufficient to do the failover. – kasperd Jan 29 '16 at 13:07
4

You will need the multiple NICs in each box and the two switches, with each box plugged into both switches. Then setup NIC Teaming/NIC Bonding.

For most modern operating systems it is a pretty simple setup, and in most basic environments it will just work. In Windows Server 2012 you have to use Switch Independent mode and the OS will take care of everything.

Assuming Windows Server 2012:

  • In Server Manager click Local Server.
  • Click Disabled next to NIC Teaming.
  • In the Teams area, click Tasks - New Team.
  • Provide a name for the NIC Team.
  • Choose the adapters you want to add to the team.
  • Default settings under Additional Settings should be fine. If you have trouble you can try changing the Load Balancing Mode.

You have just created a logical interface that contains the other physical interfaces. Now you can provide the Team interface an IP, Default Gateway, etc. just like you would any network interface.

Dre
  • 1,710
  • 7
  • 12
1

Redundant Network Basic

This image represents an entry level amount of redundancy that can be achieved network wide.

Switch 1 and Switch 2 represent a collapsed Core/Distribution network and Switch 3, 4, and 5 represent an access network.

Switches 1 and 2 should usually be aggregated by combining them into some kind of stack when possible, however you can configure them as 2 separate non stacked switches if need be. The two switches should be connected by multiple aggregated 802.3ad (LACP) connections consisting of at least 2 cables

Switch 3, 4 and 5 should also be aggregated into a stack when possible. Each switch should have a connection to each other switch, preferably at least 2 cables setup to use 802.3ad (LACP).

Each switch in the Core network should be connected to each switch in the access network using at least 2 cables again when possible use 802.3ad (LACP).

Each server/host should have 2+ network ports capable of LACP or NIC Teaming. These should distributed to different switches and cluster members should be distributed so that at least one port of one member of the cluster exists on each switch.

Then you just build your access network again and again and connect it back to your core. You would never connect 2 access networks however because it is the job of the core to handle that part of the switching.

This is a rough outline of what I see most and have adopted myself in the past. It would allow multiple simultaneous failures without network outage.

Nick Young
  • 688
  • 4
  • 14