we are using a glassfiber full duplex 1Gbps line, and we used one of the internet routable ip address to configure Mikrotik router, and created a GRE tunnel to route more ip’s to current network.
When connecting with current ISP’s router we are getting 1Gbps, when connecting same host to Mikrotik router we are getting about 250-300 Mbps. So the question is what is happening with the rest of the +/- 700 Mbps which is in capacity? So why we are getting only 1/3 of the total speed with a GRE tunnel? is this a normal behaviour? or is there a way to finetune configurations?
Can anybody advise/suggest what to do to be able to achieve to use full capacity of the line?
First thing I would recommend is to check is the MTU of the GRE Interface(s). MTU mismatches can cause repacketing ouf outgoing GRE traffic substancialy incerasing the CPU load.
If I remember right the MTU for GRE IPv4 interfaces should be 1436
1.436 byte (payload) + 20 byte (TCP header) + 20 byte (inner IP header) + 24 byte (GRE header and outer IP header]) = 1.500 bytes = Ethernet MTU
Yep, those 1436 would be MSS, not MTU. Many people keep mixing up the two as the most suggested workaround for issues with path MTU discovery is MSS adjustment.
Also, the description in the OP suggests that maybe a packet for the public IP arrives via WAN, gets encapsulated into GRE and leaves via the same WAN, so in case of bi-directional testing, each direction of the WAN is used by both A->B traffic and B->A traffic?
While theoretically GRE (as protocol) can carry many different types of payload (as we see for example in case of EoIP), but on Mikrotik interface type “GRE tunnel” can carry only IP or IPv6, so there is no L2 there, and existence of L2MTU field for it is quite odd and doesn’t really make a lot of sense.
And you can’t change it anyway.
Probably unrelated but I had to set the MTU on my wireguard interfaces to 1500 to get the tunnel to work properly for my clients access to sites via my internet connection.