comparing open source BGP stacks performance
Hi all, because I couldn't find many results of performance tests of open source routing stacks, I did some tests and wrote up https://elegantnetwork.github.io/posts/comparing-open-source-bgp-stacks/. It compares BIRD, FRRouting and GoBGP. Justin
On Tue, Jul 27, 2021 at 11:39:52AM -0700, Justin Pietsch wrote:
Hi all, because I couldn't find many results of performance tests of open source routing stacks, I did some tests and wrote up https://elegantnetwork.github.io/posts/comparing-open-source-bgp-stacks/. It compares BIRD, FRRouting and GoBGP.
Hi That are interesting data, thanks. In our tests we noticed that in non-trivial setups the total time is dominated by time spent in processing filters. -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
Hello! I've updated bgperf somehow, see my fork: https://github.com/marenamat/bgperf Notable changes are: * replacing ExaBGP by BIRD to consume less resources by feeding * fixed policy configuration in BIRD (no clue whether it works with other stacks) Thanks a lot for the batch feature, it comes handy. I am also thinking about other features currently missing (or I have just not found them): * running every settings combination several times and counting min-avg-max-stdev-whatever stats * specifying a branch to test in the batch Do you have any recommendation where to put the implementation of these features? Maria On 7/28/21 3:07 PM, Ondrej Zajicek wrote:
On Tue, Jul 27, 2021 at 11:39:52AM -0700, Justin Pietsch wrote:
Hi all, because I couldn't find many results of performance tests of open source routing stacks, I did some tests and wrote up https://elegantnetwork.github.io/posts/comparing-open-source-bgp-stacks/. It compares BIRD, FRRouting and GoBGP.
Hi
That are interesting data, thanks. In our tests we noticed that in non-trivial setups the total time is dominated by time spent in processing filters.
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Tuesday, July 27th, 2021 at 7:39 PM, Justin Pietsch <jpietsch@gmail.com> wrote:
Hi all, because I couldn't find many results of performance tests of open source routing stacks, I did some tests and wrote up https://elegantnetwork.github.io/posts/comparing-open-source-bgp-stacks/. It compares BIRD, FRRouting and GoBGP.
Such a shame you missed out OpenBGPD by the OpenBSD team. ;-(
Hej,
Such a shame you missed out OpenBGPD by the OpenBSD team. ;-(
Justin did a followup including rustybgp and openbgpd: https://elegantnetwork.github.io/posts/followup-measuring-BGP-stacks/ Best, f.
OpenBGPd has always been a dog. Of all that you tested, I’d still stick with bird. Awesome testing though. Chris
On Aug 24, 2021, at 2:56 PM, Fran <fatal@mailbox.org> wrote:
Hej,
Such a shame you missed out OpenBGPD by the OpenBSD team. ;-(
Justin did a followup including rustybgp and openbgpd:
https://elegantnetwork.github.io/posts/followup-measuring-BGP-stacks/
Best, f.
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Tuesday, August 24th, 2021 at 8:03 PM, Chris Malayter <mustang@terahertz.net> wrote:
OpenBGPd has always been a dog.
For those of you who don't monitor the OpenBSD mailing list, explanation from Claudio Jeker (one of the key figures in OpenBGPD dev): The massive amount of memory used in OpenBGPD comes from the fact that unlike BIRD OpenBGPD runs with a full Adj-RIB-Out. The tests result in large amount of prefixes that need to be tracked. If you have 100 peers announcing 10000 random prefixes then you end up with 100 * 100 * 10000 = 100Million elements to manage. This is not a realistic test since in most cases the number of routes in the Adj-RIB-Out is limited (even on route servers). In the end for day to day use OpenBGPD performs well enough for many people. Future releases will focus more on performance and optimizing Adj-RIB-Out is on the list.
Could it be compared with Bird Runing one RIB per participant? Em qua., 25 de ago. de 2021 às 11:16, Laura Smith < n5d9xq3ti233xiyif2vp@protonmail.ch> escreveu:
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Tuesday, August 24th, 2021 at 8:03 PM, Chris Malayter < mustang@terahertz.net> wrote:
OpenBGPd has always been a dog.
For those of you who don't monitor the OpenBSD mailing list, explanation from Claudio Jeker (one of the key figures in OpenBGPD dev):
The massive amount of memory used in OpenBGPD comes from the fact that unlike BIRD OpenBGPD runs with a full Adj-RIB-Out.
The tests result in large amount of prefixes that need to be tracked.
If you have 100 peers announcing 10000 random prefixes then you end up with 100 * 100 * 10000 = 100Million elements to manage.
This is not a realistic test since in most cases the number of routes in the Adj-RIB-Out is limited (even on route servers).
In the end for day to day use OpenBGPD performs well enough for many people. Future releases will focus more on performance and optimizing Adj-RIB-Out is on the list.
-- Douglas Fernando Fischer Engº de Controle e Automação
participants (7)
-
Chris Malayter -
Douglas Fischer -
Fran -
Justin Pietsch -
Laura Smith -
Maria Matejka -
Ondrej Zajicek