Avoiding Single point of failure nodes

[networking.stud]

On our Head Quarter buildings we need to upgrade to 10Gbps connections and avoiding single point of failure nodes also.

Refering to attached topo please, what equipments do you suggest invoking?

-must have at least 2x 10Gbps interfaces

-must be hotstandby (redundancy/failover) protected but with one single Fo connection per path.

Any suggest will be very helpful.

Thank you for your time.

[Alessio Andreoli]

Hi Net,

in your pic there ae three switches and it is not really clear the role of each of them. By he way, with three devices connected in this way you are not achieving a full redundancy. If you want some suggestion about the devices anyway, you need to provide more information about:

- bandwidth used (or packet per second)

- number of users

- features and budget (do you need particular features? PFR, Particular high computational power etc.)

- how you want to connect your enterprise to the ISP

and how much you predict to scale your core in the future(MPLS? BGP?).

By now i would say you should go for the ASR series maybe 1002 or 1004. But DO NOT choose your devices only based on this few info you gave.

HTH

Alessio

[sean_evershed]

Hi,

Another alternative is to deploy small 4500 or 6500 switches in these locations. They have several advantages:

- Support for dual supervisors.

- Support for dual power supplies.

- Support for multiple blades.

- Highly reliable.

If you had the budget you could deploy Nexus switches and run vPC across the links to help protect against failures.

[networking.stud]

Thank you guys for your help.

We need to build a L2 connection between buildings.

Our budget is limmited to lets say 4500 series switches

Is it any technology that 2 switches will serve as one with hotstandby of each-other. So, if primary fails the 2nd one will take care of routing traffic. Lets say like HSRP but with single (fo) connection.

On Stackwise tech (as I know) if primary sw goes down you need to manuallly put connections of dead sw to operating one, I am correct.

10Gbps connection is for prespective pourpose. Our actual traffic is max 1Gbps (average 300-600Mbps)

Thanks again!

[networking.stud]

Please I need help regarding Stuckwise tech.

So, if primary switch goes down, all Layer 1 connections of dead switch must be transfered to other alive switch manuallly, is this correct?

Does exist any other technology to double switches per node (2 sw serve as one with just one WAN/MAN connection)?

Thank you for help!

[Leo Laohoo]

Your picture has one vague data.

What do you mean your fibre optic run is "more than 10 Km"???

How far?  11 Km?  100 Km?  10000 Km?

Our budget is limmited to lets say 4500 series switches

Will you be doing MPLS?  Layer 3 routing per site?  Layer 3 routing per floor?

On Stackwise tech (as I know) if primary sw goes down you need to manuallly put connections of dead sw to operating one, I am correct.

Don't worry about StackWise because it's not applicable to your solution.  Cisco StackWise cable can go up to 3 metres only.

[stephenshaw · Accepted solution]

Hi,

don't neglect the physical elements if you need to avoid single points of failure. You can design the network to have redundancy and no single points of failure, but if a backhoe takes out both circuits down the street .....

i.e. separate physical entrances into the buildings, perhaps CO diversity for each circuit, etc. You need to work with your Telco(s) to ensure that, they too, are not a single point of failure. Even selecting two different Telcos may not help if they are using the same CO. Something to consider.

Steve

[networking.stud]

Thank you Steve.

Our target is to secure our border equipments, regarding Telco conn. we are trusting coverage of SLA :-)

We think going for 2 devices with 2 different Fo conn. per node, after them HSRP configured routers.

We will welcome any other idea/way that you think may be helpful.

Thanks all for help!

Best Regards!