Ask Your Question
1

Do virtual routers ever fail?

asked 2014-09-20 01:42:50 -0500

fhussain gravatar image

updated 2014-09-20 07:37:58 -0500

Hi,

Do virtual routers ever fail or stop functioning, like the physical routers do?

What I am trying to say is that is it possible that a Virtual Network(i.e. a Project or a Container) which is functionally well and is provisioned to a tenant, later on fails due to a router failure or due to the failure of one or some of its interfaces?

Do elements in virtual networks e.g. Virtual routers, virtual switches, virtual interfaces and virtual machines fail, even if the underlying physical infrastructure works fine?

Has this ever happened to someone?

Thanks.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2014-09-20 14:22:07 -0500

larsks gravatar image

Do virtual routers ever fail or stop functioning, like the physical routers do?

Well...sure. A "virtual router" is a network namespace running on your network controller. If that network controller stops working, then all of your virtual routers will stop working. Aside from that sort of failure, then virtual routers are fairly resilient -- they don't actually run any software; they simply connected virtual interfaces with a set of iptables rules and routing tables inside a network namespace. While it is possible that a kernel bug could stop such a router from forwarding traffic, it is fairly unlikely.

So yes, a virtual router can fail, but typically that failure is going to manifest on the entire host and will not effect an individual router while leaving other functioning correctly. But that brings us to your next question:

Do elements in virtual networks e.g. Virtual routers, virtual switches, virtual interfaces and virtual machines fail, even if the underlying physical infrastructure works fine?

You have asked a number of distinct questions here.

VIrtual machines can certainly fail, even if the underlying physical infrastructure is sound. A virtual machine can fail for the same software-related reasons that can cause problems in a physical machine: kernel bugs, application bugs, out-of-memory errors, resource contention problems, and so forth.

I'm sure there are situations in which the other components can also fail, even if the underlying physical infrastructure has not failed. You typically design around these sorts of issues by ensuring that you have sufficient redundancy in your environment. At the application level, this means multiple application servers running on multiple physical host, with a load balancer in front of the service (ensuring that the load balancer itself has a failover partner). For the virtual network infrastructure, you make use of something like the Neutron Distributed Virtual Router feature to provide redundancy in the event of virtual router failures. At the OpenStack level, you provide redundancy similar to how you achieved it at the application level -- by running multiple instances of key services and using load balancers where necessary.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2014-09-20 01:42:50 -0500

Seen: 185 times

Last updated: Sep 20 '14