DPDK patches and discussions
 help / color / Atom feed
From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: "akhil.goyal@nxp.com" <akhil.goyal@nxp.com>,
	"dev@dpdk.org" <dev@dpdk.org>,
	dpdk-techboard <techboard@dpdk.org>
Cc: "Medvedkin, Vladimir" <vladimir.medvedkin@intel.com>,
	Anoob Joseph <anoobj@marvell.com>,
	Thomas Monjalon <thomas@monjalon.net>,
	Ravi Kumar <ravi1.kumar@amd.com>,
	Ruifeng Wang <ruifeng.wang@arm.com>
Subject: Re: [dpdk-dev] [PATCH v4 0/5] integrate librte_ipsec SAD into ipsec-secgw
Date: Thu, 23 Jan 2020 12:52:01 +0000
Message-ID: <SN6PR11MB25587F743337C38DC1F30FEA9A0F0@SN6PR11MB2558.namprd11.prod.outlook.com> (raw)
In-Reply-To: <VE1PR04MB663974FAC62180BDBC2A8B7DE60F0@VE1PR04MB6639.eurprd04.prod.outlook.com>

Hi Akhil,

> > > > Hi Vladimir,
> > > > The SA lookup logic and management is purely requirement based for the
> > > application.
> > > >The application may only cater to <128 SAs which can
> > > > be handled based on the current logic.
> > >
> > > Not always, current implementation can handle < 128 SA,
> > > whose SPI%128 never match (let say it cant't handle SPI=1 and SPI=129).
> > > Yes, what we have right now has nearly zero overhead,
> > > and might be ok for some really simple show-cases.
> > > But for majority of production IPsec implementations,
> > > I believe that definitely wouldn't be enough.
> > >
> > > > –single-sa option cannot handle this.
> > > > Sample applications in DPDK are there to showcase the best a hardware can
> > > deliver.
> > >
> > > My thought was - that's the reason we have single-sa option -
> > > demonstrate best possible HW perf without minimal SW intervention.
> > > For something more serious than that, we use generic SAD implementation.
> > >
> > > > IMO, we cannot allow this logic on NXP hardwares. We
> > > > give performance numbers based on IPSec app to customers and we cannot
> > > allow 15% degradation.
> > >
> > > As Vladimir said, we are looking how to improve current SAD numbers
> > > and minimize the drop.
> > > But with same equals - plain array will always be faster than hash table,
> > > so not sure we will be able to match existing performance.
> > > So two questions:
> > > 1. What exact case you use for perf testing
> > >     (total number of SAs, packets per burst belong to the same/different SAs)?
> > >     Might be there is a way to speedup it.
> > >     Again if 10-15% is not an affordable drop, which one is: zero or ...?
> >
> > We should add features judiciously, we cannot drop the performance of a
> > benchmarking
> > Application in lieu of adding functionality. We should only add features which
> > are not
> > Impacting the performance significantly.
> > Every vendor may have different cases. We cannot tune for everybody.
> > However, I see drop in 64 outbound 64 inbound SAs all with different SPI and IPs.
> > Packets per burst = 32 all with different SAs.
> >
> 
> We can have two modes of lookup similar to l3fwd - EM and LPM.
> LPM is O(1) while EM is more realistic. Similar logic can be added here as well.
> With L3fwd also we showcase performance for best case(lpm) and the worst case(em)
> What Say?

We discussed it off-line with Vladimir and came up with similar idea:
Have a proper/generic SAD implementation and add limited size plain-array
on top of it as 1xway associative cache.
So for the case when all active SAs fit into the cache and no SPI collisions,
we should have same performance as now (with plain array).
From other side, we'll still have generic/scalable/rfc compliant implementation.
Sort of best sides from two words.
Plans are to submit v4 with such approach in next few days.   

> 
> As discussed in the DPDK-status meeting today, this patchset need to be discussed in
> Techboard meeting. Please include this topic in the upcoming meeting on 29th Jan.

As I said above, I think we found a way to deal with it without any perf drop
for existing cases.
Though sure, if you feel some extra discussion is needed, let's request to
put it into agenda.

Konstantin 

> 
> -Akhil
> 
> > > 2. I think there are 2 different directions for ipsec-secgw:
> > >    From one-side there is a desire to use it as a show-case for best-possible HW
> > > IPsec performance
> > >   (which is understandable).
> > >    From other side - attempt to make it as close as real-world generic ipsec
> > > processing app as possible
> > >    (support for ESN, replay window, fragmented packets, generic proper SAD,
> > > etc).
> > >    Obviously these goals contradict and it makes really hard for the same app to
> > > fulfill both.
> > >    Any thoughts how to deal with that?
> > >    One obvious would be to split the app, anything else?
> >
> > We can have a fallback mechanism back to original functionality for whatever
> > feature
> > which has some perf drop.
> > Splitting an app can be thought of but that would be similar to a full fledged
> > IPSec stack
> > like VPP-IPSec.
> >
> > >
> > > Konstantin
> > >
> > > > Other vendors(Marvell, ARM, AMD) please comment?
> > > > Regards,
> > > > Akhil
> > > > From: Medvedkin, Vladimir <mailto:vladimir.medvedkin@intel.com>
> > > > Sent: Friday, January 17, 2020 10:35 PM
> > > > To: Akhil Goyal <mailto:akhil.goyal@nxp.com>; mailto:dev@dpdk.org
> > > > Cc: mailto:konstantin.ananyev@intel.com
> > > > Subject: Re: [PATCH v4 0/5] integrate librte_ipsec SAD into ipsec-secgw
> > > >
> > > > Hi Akhil,
> > > > Indeed with our tests we also seeing ~15% perf drop for small packets (~90B)
> > > and ~3-4% drop for 1KB packets. While I am looking on a ways
> > > > to minimize the drop, I think it would be hard, if possible at all to eliminate it
> > > completely.
> > > > Reason for that: current SAD implementation is completely synthetic (using
> > > plain array structure indexed by SPI value). That provides a very
> > > > low overhead, but doesn't provide expected functionality and can't be used
> > in
> > > proper implementation.
> > > > To measure plain IPsec performance without SAD user can still use '--signle-
> > sa'
> > > option.
> > > > On 15/01/2020 15:45, Akhil Goyal wrote:
> > > > Hi Vladimir,
> > > >
> > > > There is more than 10% drop with this patchset on NXP hardware with both
> > > legacy mode and the ipsec lib mode. This would need some
> > > > debugging.
> > > > Didn't you see any drop on intel?
> > > >
> > > > Regards,
> > > > Akhil
> > > >
> > > > -----Original Message-----
> > > > From: Vladimir Medvedkin mailto:vladimir.medvedkin@intel.com
> > > > Sent: Tuesday, January 14, 2020 7:57 PM
> > > > To: mailto:dev@dpdk.org
> > > > Cc: mailto:konstantin.ananyev@intel.com; Akhil Goyal
> > > mailto:akhil.goyal@nxp.com
> > > > Subject: [PATCH v4 0/5] integrate librte_ipsec SAD into ipsec-secgw
> > > >
> > > > This series integrates SA database (SAD) capabilities from ipsec library.
> > > > The goal is to make ipsec-secgw RFC compliant regarding inbound SAD.
> > > > Also patch series removes hardcoded limitation for maximum number of SA's
> > > > and SP's.
> > > >
> > > > v4:
> > > >  - put tunnel SA's into SAD with SPI_ONLY type for performance reason
> > > >
> > > > v3:
> > > >  - parse SA and SP into sorted array instead of linked list
> > > >
> > > > v2:
> > > >  - get rid of maximum sp limitation
> > > >
> > > > Vladimir Medvedkin (5):
> > > >   ipsec: move ipsec sad name length into .h
> > > >   examples/ipsec-secgw: implement inbound SAD
> > > >   examples/ipsec-secgw: integrate inbound SAD
> > > >   examples/ipsec-secgw: get rid of maximum sa limitation
> > > >   examples/ipsec-secgw: get rid of maximum sp limitation
> > > >
> > > >  examples/ipsec-secgw/Makefile      |   1 +
> > > >  examples/ipsec-secgw/ipsec-secgw.c |   4 +-
> > > >  examples/ipsec-secgw/ipsec.h       |  11 +-
> > > >  examples/ipsec-secgw/meson.build   |   2 +-
> > > >  examples/ipsec-secgw/parser.c      |   4 +
> > > >  examples/ipsec-secgw/parser.h      |   9 ++
> > > >  examples/ipsec-secgw/sa.c          | 256 +++++++++++++++++++++++-----------
> > --
> > > -
> > > >  examples/ipsec-secgw/sad.c         |  90 +++++++++++++
> > > >  examples/ipsec-secgw/sad.h         |  74 +++++++++++
> > > >  examples/ipsec-secgw/sp4.c         | 114 ++++++++++++-----
> > > >  examples/ipsec-secgw/sp6.c         | 112 +++++++++++-----
> > > >  lib/librte_ipsec/ipsec_sad.c       |  20 +--
> > > >  lib/librte_ipsec/rte_ipsec_sad.h   |   2 +
> > > >  13 files changed, 528 insertions(+), 171 deletions(-)
> > > >  create mode 100644 examples/ipsec-secgw/sad.c
> > > >  create mode 100644 examples/ipsec-secgw/sad.h
> > > >
> > > > --
> > > > 2.7.4
> > > >
> > > > --
> > > > Regards,
> > > > Vladimir
> > > > -->

  reply index

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-12-11 16:45 [dpdk-dev] [PATCH 0/4] " Vladimir Medvedkin
2019-12-11 16:45 ` [dpdk-dev] [PATCH 1/4] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2019-12-11 16:45 ` [dpdk-dev] [PATCH 2/4] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2019-12-11 16:45 ` [dpdk-dev] [PATCH 3/4] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2019-12-11 16:45 ` [dpdk-dev] [PATCH 4/4] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 0/5] integrate librte_ipsec SAD into ipsec-secgw Vladimir Medvedkin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 " Vladimir Medvedkin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 " Vladimir Medvedkin
2020-01-15 15:45       ` Akhil Goyal
2020-01-17 12:26         ` Akhil Goyal
2020-01-17 17:05         ` Medvedkin, Vladimir
2020-01-20  6:44           ` Akhil Goyal
2020-01-20 12:44             ` Anoob Joseph
     [not found]             ` <SN6PR11MB25581C7C8F969AA18EE8C1949A320@SN6PR11MB2558.namprd11.prod.outlook.com>
     [not found]               ` <SN6PR11MB25588E3DD326CFC90DD1E3989A320@SN6PR11MB2558.namprd11.prod.outlook.com>
2020-01-20 14:45                 ` [dpdk-dev] FW: " Ananyev, Konstantin
2020-01-21 14:47                   ` [dpdk-dev] " Akhil Goyal
2020-01-23 11:11                     ` Akhil Goyal
2020-01-23 12:52                       ` Ananyev, Konstantin [this message]
2020-01-23 12:56                         ` Akhil Goyal
2020-01-23 13:33                           ` Thomas Monjalon
2020-01-23 15:46                             ` Ananyev, Konstantin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 0/6] " Vladimir Medvedkin
2020-01-31 12:53         ` Akhil Goyal
2020-02-04  4:11           ` Anoob Joseph
2020-02-04 15:22             ` Akhil Goyal
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 0/8] " Vladimir Medvedkin
2020-02-04 15:25           ` Akhil Goyal
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 1/8] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 2/8] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 3/8] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 4/8] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 5/8] examples/ipsec-secgw: get rid of maximum sp limitation Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 6/8] examples/ipsec-secgw: add SAD cache Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 7/8] examples/ipsec-secgw: set/use mbuf ptype Vladimir Medvedkin
2020-01-31 17:39         ` [dpdk-dev] [PATCH v6 8/8] doc: update ipsec-secgw guide Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 1/6] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 2/6] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 3/6] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 4/6] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 5/6] examples/ipsec-secgw: get rid of maximum sp limitation Vladimir Medvedkin
2020-01-29 14:06       ` [dpdk-dev] [PATCH v5 6/6] examples/ipsec-secgw: add SAD cache Vladimir Medvedkin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 1/5] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2020-01-14 15:51       ` Ananyev, Konstantin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 2/5] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2020-01-14 15:53       ` Ananyev, Konstantin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 3/5] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2020-01-14 15:54       ` Ananyev, Konstantin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 4/5] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2020-01-14 15:56       ` Ananyev, Konstantin
2020-01-14 14:27     ` [dpdk-dev] [PATCH v4 5/5] examples/ipsec-secgw: get rid of maximum sp limitation Vladimir Medvedkin
2020-01-14 15:57       ` Ananyev, Konstantin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 1/5] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 2/5] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 3/5] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 4/5] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2020-01-13 12:55   ` [dpdk-dev] [PATCH v3 5/5] examples/ipsec-secgw: get rid of maximum sp limitation Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 1/5] ipsec: move ipsec sad name length into .h Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 2/5] examples/ipsec-secgw: implement inbound SAD Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 3/5] examples/ipsec-secgw: integrate " Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 4/5] examples/ipsec-secgw: get rid of maximum sa limitation Vladimir Medvedkin
2019-12-18 16:00 ` [dpdk-dev] [PATCH v2 5/5] examples/ipsec-secgw: get rid of maximum sp limitation Vladimir Medvedkin

Reply instructions:

You may reply publically 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=SN6PR11MB25587F743337C38DC1F30FEA9A0F0@SN6PR11MB2558.namprd11.prod.outlook.com \
    --to=konstantin.ananyev@intel.com \
    --cc=akhil.goyal@nxp.com \
    --cc=anoobj@marvell.com \
    --cc=dev@dpdk.org \
    --cc=ravi1.kumar@amd.com \
    --cc=ruifeng.wang@arm.com \
    --cc=techboard@dpdk.org \
    --cc=thomas@monjalon.net \
    --cc=vladimir.medvedkin@intel.com \
    /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

DPDK patches and discussions

Archives are clonable:
	git clone --mirror http://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ http://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev


Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/ public-inbox