DPDK usage discussions
 help / color / mirror / Atom feed
From: "Дмитрий Степанов" <stepanov.dmit@gmail.com>
To: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
Cc: Steffen Weise <stweise.linux@gmail.com>, users@dpdk.org
Subject: Re: Find all matches with DPDK ACL
Date: Fri, 26 Nov 2021 17:56:27 +0300	[thread overview]
Message-ID: <CA+-SuJ3pYHcsNn_4ZhVDERe8FdWM8Qaia+My+Qx8OEkA90kLuQ@mail.gmail.com> (raw)
In-Reply-To: <20211126171232.401fefca@sovereign>

[-- Attachment #1: Type: text/plain, Size: 1653 bytes --]

I have approx 5K-10K (5 000 - 10 000) rules.
On average I have 10-20 matches (60 max).
I don't need to insert/delete/update rules frequently - you can consider
rules being permanent which are loaded once on startup.

пт, 26 нояб. 2021 г. в 17:12, Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>:

> 2021-11-26 16:53 (UTC+0300), Дмитрий Степанов:
> > Hi!
> > I have a big number of IPv4 5-tuple rules, every rule corresponds to some
> > action. I need to find all matched rules and perform all tied actions.
>
> I rather meant the subject field,
> like splitting the flows or access control is a typical application of ACL.
> I'm asking partially out of curiosity,
> but also because there may be a better solution then DPDK ACL.
>
> > The search time greatly affects overall system performance, so I can't
> just
> > scan all rules. ACL is based on multi-bit tries and provides great
> > performance, so I'm looking for nearly the same performance with the
> > ability to find all matches within a single request.
>
> Some regex libraries, Hyperscan or DPDK regexdev in particular,
> take a database of rules, compile it to an efficient form
> (Hyperscan generates vector instructions, regexdev may use HW
> acceleration),
> and then allow to match input to the entire database in a single request,
> yielding every match for every expression.
>
> From my experience, performance is decent,
> but of course it depends on the number or rules and their complexity.
> How many rules do you have?
> How many rules are expected to match (avg/max)?
> How often do you need to insert/delete/update rules?
>

[-- Attachment #2: Type: text/html, Size: 2040 bytes --]

  reply	other threads:[~2021-11-28 12:57 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-18 16:55 Дмитрий Степанов
2021-11-24 10:06 ` Steffen Weise
2021-11-24 15:19   ` Dmitry Kozlyuk
2021-11-26 13:53     ` Дмитрий Степанов
2021-11-26 14:12       ` Dmitry Kozlyuk
2021-11-26 14:56         ` Дмитрий Степанов [this message]
2021-11-26 23:56           ` Dmitry Kozlyuk

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CA+-SuJ3pYHcsNn_4ZhVDERe8FdWM8Qaia+My+Qx8OEkA90kLuQ@mail.gmail.com \
    --to=stepanov.dmit@gmail.com \
    --cc=dmitry.kozliuk@gmail.com \
    --cc=stweise.linux@gmail.com \
    --cc=users@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).