New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Is it possible to announce a single /24 prefix at two datacenters?
Here is how it's going to work;
I have a single /24 RIPE prefix (C class). To conserve resources, I wish to use the prefix at two different datacenters. They would be located in two different countries in the RIPE region.
E.g.
1.1.1.1/24 announced at DC1 and DC2
DC 1 assigns a /25 and DC2 assigns another /25 on the remaining IPs.
There would be two gateways IPs, broadcast and such.
On network level, it will be announced as normal /24 (as upstreams require) but only the /25 would be routed by the router at DC1 and then the next /25 in DC2.
No CDN, it shouldn't be anycasting. No load-balancing is needed, and they would be separate without interconnections.
Is this a possible configuration?
Comments
Probably through some sort of GRE tunnel from one DC to the other?
First of all, nothing like A, B or C classes no longer exist for more than a decade.
In regards to the configuration. Technically possible. Both Datacentres would have to tunnel the /25 they don't serve to the other Datacentre, either via GRE or a physical link. You'll likely face the bandwidth costs (double) and you will likely face routing issues
So in practice, can I say /24 is the minimum announcing unit?
IPv4 /24's and IPv6 /48's I believe are the minimums.
I'm certainly no expert but isn't /56 a valid announcing unit for 1 subscriber site?
Usually upstreams filter IPv6 prefixes smaller than a /48.
/24 for IPv4.
Yes, /24 for ipv4, /48 for ipv6.
It appears what OP really wants is to split an ipv4 /24 across two data centers. This just cannot work. If it is announced in both places, that is already anycast. Packets headed for either /25 can "arbitrarily" hit either DC, and there is no way to control that.
This can be "sort of made to work but not really" by tunnelling the traffic that ended up in the wrong DC back to the correct DC for it. Besides BW costs, this increases the points of failure in the system.
Just rent another /24.
I see where you're coming from. Point withdrawn.
Yes, and I would really recommend another /24. : ).
Anything smaller than /48 will be filtered by Tier-1's.