Using Region-X assigned IP space in Region-Y?
Hello NANOG'ers, This is a question that popped into my mind recently. So far my google-fu has not turned up anything of use. Imagine a company that has an RIR-assigned IPv4/23, and has it subnetted into a pair of /24's so that they can have replicated datacenters in two different geographical locations each running its own BGP infrastructure (using the same ASN). As far as I can see, this is all fine and good assuming that these two sites are in the same RIR's domain. Is it possible/allowable to move one of these datacenters to a different geographical region with a different RIR and keep using the same two subnets, or will a new /24 need to be requested from the new RIR? If there's a book or reference that answers this question, please let me know - I'm willing to do as much research on my own as I can. Thanks!
On Sun, 27 Mar 2011 08:58:29 MDT, Mark Leonard said:
Is it possible/allowable to move one of these datacenters to a different geographical region with a different RIR and keep using the same two subnets, or will a new /24 need to be requested from the new RIR?
There's only one question to be asked - will the (possibly new) upstream of the moved datacenter announce the route for the /24 or not? There's plenty of AS's that announce little chunks on different continents, as long as the upstreams will carry the announcement, you're golden.
On 3/27/11 8:19 AM, "Valdis.Kletnieks@vt.edu" <Valdis.Kletnieks@vt.edu> wrote:
On Sun, 27 Mar 2011 08:58:29 MDT, Mark Leonard said:
Is it possible/allowable to move one of these datacenters to a different geographical region with a different RIR and keep using the same two subnets, or will a new /24 need to be requested from the new RIR?
There's only one question to be asked - will the (possibly new) upstream of the moved datacenter announce the route for the /24 or not?
Why would the new upstream refuse to announce the /24 assuming he has the correct information for his route objects and visible through the RIR database. Zaid
On 3/27/2011 12:10 PM, Zaid Ali wrote:
On 3/27/11 8:19 AM, "Valdis.Kletnieks@vt.edu"<Valdis.Kletnieks@vt.edu> wrote:
There's only one question to be asked - will the (possibly new) upstream of the moved datacenter announce the route for the /24 or not?
Why would the new upstream refuse to announce the /24 assuming he has the correct information for his route objects and visible through the RIR database.
Some transit providers dislike announcing smaller networks, and thus have lower limits. Jima
On 3/27/11 10:54 AM, "Jima" <nanog@jima.tk> wrote:
On 3/27/2011 12:10 PM, Zaid Ali wrote:
On 3/27/11 8:19 AM, "Valdis.Kletnieks@vt.edu"<Valdis.Kletnieks@vt.edu> wrote:
There's only one question to be asked - will the (possibly new) upstream of the moved datacenter announce the route for the /24 or not?
Why would the new upstream refuse to announce the /24 assuming he has the correct information for his route objects and visible through the RIR database.
Some transit providers dislike announcing smaller networks, and thus have lower limits.
Jima
Then the said transit provider customer will turn off the circuit and move to the next transit provider that doesn't have a problem with /24. If you are in a monopolistic ISP environment then it is different and that is a different topicof discussion. Sadly been there done that. Zaid
On Sun, 27 Mar 2011 12:54:52 CDT, Jima said:
On 3/27/2011 12:10 PM, Zaid Ali wrote:
On 3/27/11 8:19 AM, "Valdis.Kletnieks@vt.edu"<Valdis.Kletnieks@vt.edu> wrote:
There's only one question to be asked - will the (possibly new) upstream of the moved datacenter announce the route for the /24 or not?
Why would the new upstream refuse to announce the /24 assuming he has the correct information for his route objects and visible through the RIR database.
Some transit providers dislike announcing smaller networks, and thus have lower limits.
Which is why I said you need to find a willing upstream. And yes, if Mark moving into a monopoly area, he'll probably have to do it the way the monopoly wants it done.
As far as I can see, this is all fine and good assuming that these two sites are in the same RIR's domain.
Is it possible/allowable to move one of these datacenters to a different geographical region with a different RIR and keep using the same two subnets, or will a new /24 need to be requested from the new RIR?
If there's a book or reference that answers this question, please let me know - I'm willing to do as much research on my own as I can.
- RIR Comparative Policy Overview http://www.nro.net/category/documents/rir-comparative-policy-overview and AfriNIC http://www.afrinic.net/policy.htm APNIC http://www.apnic.net/policy/current ARIN http://www.arin.net/policy LACNIC http://lacnic.net/en/politicas RIPE NCC http://www.ripe.net/ripe/policies/ -M
On Mar 27, 2011, at 7:58 AM, Mark Leonard wrote:
Hello NANOG'ers,
This is a question that popped into my mind recently. So far my google-fu has not turned up anything of use.
Imagine a company that has an RIR-assigned IPv4/23, and has it subnetted into a pair of /24's so that they can have replicated datacenters in two different geographical locations each running its own BGP infrastructure (using the same ASN).
As far as I can see, this is all fine and good assuming that these two sites are in the same RIR's domain.
It actually doesn't matter where they are located, but, it is a lot cleaner to use different ASNs as the two networks have different routing policies.
Is it possible/allowable to move one of these datacenters to a different geographical region with a different RIR and keep using the same two subnets, or will a new /24 need to be requested from the new RIR?
Yes, not a problem.
If there's a book or reference that answers this question, please let me know - I'm willing to do as much research on my own as I can.
Thanks!
As a general rule, you can get all your space from the RIR that serves the region where your company has its headquarters, or, you can get space from any region where you have equipment deployed. Owen
participants (6)
-
Jima -
Mark Leonard -
Max Larson Henry -
Owen DeLong -
Valdis.Kletnieks＠vt.edu -
Zaid Ali