Active-Active Configuration for MX-250

[Abdullah_Pracha]

I want to setup my MX-250 hubs in active-active mode. Currently the throughput on one hub maxes out occasionally, causing issues. Is it possible I can split my spokes on each hub?

[RaphaelL]

From what I'm reading your MX250s are running in HA ( warm spare ) ? if so , they can't be in active-active. You would need to either split the HA and run them in standalone mode or buy 2 more appliance and 1 licence.

[Abdullah_Pracha]

Yes, they are running in HA (warm spare). So if I split and run them in standalone mode I wouldn't be able to get redundancy.
Also how can I configure them in standalone mode?

[RaphaelL]

Exactly you will lose the redundancy. Before going that way , are you running MX 18.2++ ?

[Abdullah_Pracha]

Thanks for telling that, and yes its running 18.107.x. 

[RaphaelL]

MX18.2 offers quite a boost in performance for your MX250. That might save you until you make you choice about your design

[Abdullah_Pracha]

is there any guide where I can find how to configure it in standalone mode, I think a separate network would be created for each of them

[RaphaelL]

Edited reply since not related to OP problems

[PhilipDAth]

Wrong thread?

[RaphaelL]

I got an error that I had never seen before. When I replied on the correct thread , it said that my post was deleted due to a spam protection feature ( too many replies with the same message ). I never posted this here... strange. Will edit

[Ryan_Miles]

*Edited to remove HA terms that might be confusing

Keep in mind there are two common ways to deploy redundant MX hubs. The most common way is active / spare. Two MXs, one license and only the primary is doing all the work.

Another way to deploy hubs is in a active / active config. Both are licensed, in their own dashboard networks, both are active. This type of deployment allows you to use both hubs and split the spoke load. You can still plan for redundancy depending on what type of failure you think is most likely. Meaning, if you think the most likely scenario is losing one hub then make sure the remaining hub can handle all the spokes while the other hub is offline.

Many large customers deploy hubs all as active and use horizontal scaling for size and resiliency.

[Abdullah_Pracha]

Is there any guide which I can follow for the 1+1 configuration?

[PhilipDAth]

If the MX250s are in pure VPN concentrator mode then you can buy an additional licence, and put them into separate networks, and pull it all together using BGP.  You will need a network core than can talk BGP as well.

https://documentation.meraki.com/MX/Networks_and_Routing/Border_Gateway_Protocol_(BGP)

If you need to buy an additional licence I would consider how long is left on your current licence, and riding it out till that expires and put the money into bigger MXs at that time.

This is a sizing guide to help you choose an appropriate model.
https://documentation.meraki.com/MX/MX_Sizing_Information/MX_Sizing_Principles

MX450s would be the next model up at the moment.

https://meraki.cisco.com/product/security-sd-wan/large-branch-campus-concentrator/mx450/

Cisco's end of financial year is July 31st.  If you could commit to placing an order by then Cisco Meraki would be all over you with discounts on an MX450 upgrade ...

[LUX-Merakifyer]

Hi there. has somebody already deployed 2xMX250 in active/active mode ?

I understood it has to work with eBGP on the WAN sides, but it's not clear on the LAN, especially how the 2 MX can 'monitor' eachother if they are up and still have route to Internet.

I would expect a king of VRRP discussion between these, but is it based only via pure routing with iBGP?

I cannot find any documentation with the 'basic scenario' of 2 MXs in routed mode for just load balancing internal clients to Internet. Please do advise. Thanks

[PhilipDAth]

>MX can 'monitor' eachother if they are up and still have route to Internet.

They don't monitor each other when active/active.  They are simply two standalone units.  They need to be in seperate Meraki networks, and you need to use some kind of layer 3 device (such as a layer 3 switch) to manage the failover between each other.