[Nfd-dev] Maximum FIB size

[David Oran]

On 12 Apr 2017, at 13:54, Junxiao Shi wrote:

> Hi Ilane
>
> NFD as of v0.5 has no limit on PIT size. In response to an Interest
> flooding attack, it will continue to allocate more PIT entries until 
> the
> node is out of memory, after which NFD crashes.
Oops :-)

> It is an open question on how to manage PIT memory usage. Our current
> thinking is to treat this as a congestion control problem.
what else could it be?

> Ideas include
> rejecting new Interests with certain probability, and evicting old PIT
> entries.
In the spirit of generational garbage collectors, one could rank PIT 
entries by Interest Lifetime instead of current age, and timeout PIT 
entries with long Interest Lifetimes in preference to ones with shorter 
lifetimes. This may work better than strict age.

It also seems quite useful, at least in a general-purpose server 
implementation of a forwarder, to size the PIT indirectly using 
bandwidth-delay product of the output faces and FIB entries, rather than 
putting your finger up in the wind and picking a fixed size.

> You may watch https://redmine.named-data.net/issues/1301 to receive 
> updates
> on this issue.
>
> Yours, Junxiao
>
> On Mon, Apr 10, 2017 at 7:42 AM, Ilane Karise <kariseilane at gmail.com> 
> wrote:
>
>> I've been looking for at the NFD settings and found no configuration
>> regarding the size of the PIT table. If there is no configuration 
>> option
>> regarding the PIT in NFD's configuration file, is there a default 
>> size for
>> it?
>>


> _______________________________________________
> Nfd-dev mailing list
> Nfd-dev at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/nfd-dev

DaveO