LACP of 10G and 40G ports

Ok. please tell me one thing, here i have 2 40G ports which i am using for wan and lan side. also 8 10G ports. so can i bundle 4 10G and one 40G in LACP for interface redundancy?

A LACP bundle of 5 units, 4x 10G and 1x 40G? The answer is no. Your best bet is to re-arrange cabling and use both of the 40G ports for the client-side (or server-side) LACP bundle, and then bundle as many 10G interfaces as you need for server-side (or client-side) LACP bundle.

Keep in mind with a LACP bundle no single session can utilize more bandwidth than the bandwidth supported by a single interface. So with 4x 10G bundle, a single session is not able to utilize 40G bandwidth, it can still utilize 10G max. The LACP bandwidth advantage comes from session distribution across multiple links but the speed limits as defined by media still apply for individual sessions.

Thanks..Hannes...

can we add three bigip appliances 12000 series in one cluster like Bigip A will be active and rest two will be standby..connect back to back interface like: BigipA---BigipB BigipA---BigipC

will it create any loop?

can we add three bigip appliances 12000 series in one cluster like Bigip A will be active and rest two will be standby..connect back to back interface like: BigipA---BigipB BigipA---BigipC

To answer the first part of question, yes you can have a cluster of 3 appliances where 1 is Active and 2 are Standby. I actually recommend Active-Active-Standby to make better use of your hardware. At first sight, it seems it will make make most sense to go with 2 traffic groups where Unit A serves traffic-group-1 as primary Active, Unit B serves traffic-group-2 as primary Active, and Unit C is Standby for traffic-group-1 and traffic-group-2. Unit C is not under substantial risk of overloading because it's unlikely for Unit A and Unit B to suffer from a failure at the same time.

The way Active-Active(-Standby) works in F5 is that you distribute services across traffic groups. A traffic group takes ownership of some IP addresses so you need to make a rough guess (or use available data) to allocate services to traffic groups so that they are more or less balanced in terms of load. It's a one-time effort, but traffic groups can be adjusted at a later time if some new services need to be added, or old ones removed.

• To give you an example, if you have one popular service that uses 40% of bandwidth and 6 services that use 10% each, it would make sense to put that 40% bandwidth service and one 10% bandwidth service in traffic-group-1 (two IP addresses total), and the remaining 5x of 10% bandwidth services in traffic-group-2 (five IP addresses total).

I cannot answer second part of the question, how exactly you want to cable these together? Possible to make a drawing and upload somewhere? FYI, only HA cables should be directly connected appliance-to-appliance (bypassing network switch).

If these answers help you, would appreciate an upvote on the initial answer.

Regards,

ok.in case of active -standby -standby or active-active-standby, just wanted to understand the back to back I.e.ha cable connectivity.i don’t have any bypass switch.should I connect bigip1 one port to bigip2 and bigip1 second port to one of the port in bigip3?

Created an image. https://imgur.com/a/aIyqv

This gives full HA where all 3 units are able to fail over, sync config, or mirror traffic to any unit. All links to be configured as aggregated links (LACP). Uses 4x interfaces per appliance. If you think an aggregate link for insurance against a single link failure is not needed, remove one of the cables of each color.

If you observe your physical appliance, it may or may not have its own dedicated HA interface. This is more likely to do harm than any good, should be left unpopulated. It will not work with active-active and has other limitations.