Hi Oren
as of the traffic rate, do you also have numbers about the avg/peak packets/sec 
rate?

Thank you
Alfredo

> On 5 Sep 2019, at 15:23, Oren N <[email protected]> wrote:
> 
> Hi Alfredo,
> Thanks for the quick resp.
> To answer your questions:
> Ingress: 1x 10Gb / Egress: 3x 10Gb + 1x 1Gb
> <5Gbps
> The 1Gb filter is set by selecting 802.1q VLANs  (alternatively, may be 
> replaced by Segment/IP range filter); 10Gb are unfiltered - namely copy 
> everything
> Thanks,
> Oren
> 
> From: [email protected] 
> <[email protected]> on behalf of Alfredo Cardigliano 
> <[email protected]>
> Sent: Wednesday, August 28, 2019 12:59 PM
> To: [email protected] <[email protected]>
> Subject: Re: [Ntop-misc] 10/40Gb simple Tap
>  
> Hi Oren
> traffic aggregation from multiple ingress ports and duplication/distribution
> to multiple egress interfaces is available with PF_RING ZC, we have a
> tool (zbalance_ipc) which is based on the ZC API and able to do that
> A small subset of the functionalities provided by this tool are described at
> https://www.ntop.org/guides/pf_ring/rss.html#zc-load-balancing-zbalance-ipc 
> <https://www.ntop.org/guides/pf_ring/rss.html#zc-load-balancing-zbalance-ipc>
> please note that it is able to distribute traffic to processes on the same 
> host,
> as well as egress interfaces. What is missing is traffic filterins on the 
> egress
> interfaces, however that is already supported by the ZC API, and the tool
> can support it with small changes.
> The main concern here is about performance, it really depends on a few 
> factors:
> 1. number of ingress/egress links
> 2. traffic rate
> 3. filters
> Do you have some number?
> 
> Alfredo
> 
>> On 28 Aug 2019, at 14:33, Oren N <[email protected] 
>> <mailto:[email protected]>> wrote:
>> 
>> Hi,
>> Is it possible to build an inexpensive 10/40Gb simple Tap using PF_RING + 
>> ZC?  The base requirements is as follows:
>> 1. Monitor a mix of 10Gb and 1Gb interfaces
>> 2. Reduplicate traffic to N x 10/40Gb output interfaces
>> 3. Each output interface may have a network filter
>> 4. Each output interface receives all input packets that match the network 
>> filter
>> 5. Simple CLI/GUI
>> 
>> Is this a feasible/documented solution using PF_RING?
>> What are the limitations? E.g., packet loss if input > output
>> What are the HW specs for such box? E.g.,  8 core CPU, mem, disk, ... for 
>> input < 2x 10Gb+2x 1G input ; Enough PCI slots
>> Are there additional HW/SW costs beyond the PF_RING ZC NICs?
>> 
>> Thanks,
>> Oren
>> _______________________________________________
>> Ntop-misc mailing list
>> [email protected] <mailto:[email protected]>
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc 
>> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
> _______________________________________________
> Ntop-misc mailing list
> [email protected] <mailto:[email protected]>
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc 
> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
_______________________________________________
Ntop-misc mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop-misc

Reply via email to