Return-Path: Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id 8E366D00 for ; Thu, 24 May 2018 01:04:47 +0000 (UTC) X-Greylist: whitelisted by SQLgrey-1.7.6 Received: from mail-ua0-f173.google.com (mail-ua0-f173.google.com [209.85.217.173]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 8BD59180 for ; Thu, 24 May 2018 01:04:46 +0000 (UTC) Received: by mail-ua0-f173.google.com with SMTP id b25-v6so16067257uak.3 for ; Wed, 23 May 2018 18:04:46 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=lightning-engineering.20150623.gappssmtp.com; s=20150623; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc; bh=uR1/NR6x6swPHU1N1JiAjUctDbWuf+3vn+WVNocYG9w=; b=Me70NCjpLsLNvZJAsy4NxgVgMp3WGuIXpuZ4E5io2Yd5QitXcXcFtT4BzOQzPSilXD MCWX98KCQv6AAvhC1d4n+Nq4EjQNPjkbrhGM/DV5TTnaXXVuVC12oWooWDTZqqlQ6bPi 8oPt4y73M+7iT5m7LQ39J3K8rK9z44VurQ3jT5Aw6U6PHw71+IiT2XdEr8seGIbf2oxh NuaAKEVmb5GjdexfXDcJc2BSgV3H7RuXQtTVOys27ce2RJnXusLBYPN+D6cS1gL2T/kn LAc6sghx0VIMAjMAddoDF1uIbvgSKEU4y8hevVQQiiVWmFRNj4s909vkrCKvxX6rNkuJ izow== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=uR1/NR6x6swPHU1N1JiAjUctDbWuf+3vn+WVNocYG9w=; b=FbNoxA9j6UJtGzOzPWqKqz+wDZLKcjqfeNuDDjOQae8e63HGSDgpRWmrSontCViiKQ 9ACXXXNpV8bPqC6gQ0RYFEhjtoY1tvF1hp6uLEEc1DM9L6/75XBYMDjaj6qPCqq7xY5+ TSzs7sJd8F5IUKhzsGh0WnzL7j5DXpv7zWUsRGMCg3/L/pSJqCscKyg3f4k1stZOmjzc tjBuVUI9usclGAubmPitQ9SqB121A7C4Sbk5/JmaBDLelv7rR8n8ADCbua9lMmej+xQe 6klNTywGFi6HWSpyP3KHflY9zFycjWEJKeUvTpPZEqZiMJSeYkcUCQnswGNMKY0qGtqH CqSQ== X-Gm-Message-State: ALKqPwfR+CdSWCCMdVnzwDgHXwaAj9mqFSFE2U6gjhAoJdbn1TYlxW04 /J+Fe4v/bgwJXMiUPVp1I8dT6YnTCjRnepzlMmO+eg== X-Google-Smtp-Source: AB8JxZrCxE70GvSZBqriJj3q0M+CHwG20R2Vh8YysEAOqQapBJW6YEH6I6zhludmHuF+KmRDnfESf6tSiEvciE66Mog= X-Received: by 2002:ab0:30d5:: with SMTP id c21-v6mr3566808uam.69.1527123885604; Wed, 23 May 2018 18:04:45 -0700 (PDT) MIME-Version: 1.0 References: <22d375c7-a032-8691-98dc-0e6ee87a4b08@mattcorallo.com> In-Reply-To: From: Conner Fromknecht Date: Wed, 23 May 2018 18:04:34 -0700 Message-ID: To: Gregory Maxwell , Bitcoin Protocol Discussion Content-Type: multipart/alternative; boundary="000000000000bad74f056ce93cbc" X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID, HTML_MESSAGE, RCVD_IN_DNSWL_NONE autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org X-Mailman-Approved-At: Thu, 24 May 2018 01:12:46 +0000 Subject: Re: [bitcoin-dev] BIP 158 Flexibility and Filter Size X-BeenThere: bitcoin-dev@lists.linuxfoundation.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Bitcoin Protocol Discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 24 May 2018 01:04:47 -0000 --000000000000bad74f056ce93cbc Content-Type: text/plain; charset="UTF-8" Hi all, Jimpo, thanks for looking into those stats! I had always imagined that there would be a more significant savings in having all filters in one bundle, as opposed to separate. These results are interesting, to say the least, and definitely offer us some flexibility in options for filter sharding. So far, the bulk of this discussion has centered around bandwidth. I am concerned, however, that splitting up the filters is at odds with the other goal of the proposal in offering improved privacy. Allowing clients to choose individual filter sets trivially exposes the type of data that client is interested in. This alone might be enough to fingerprint the function of a peer and reduce anonymity set justifying their potential behavior. Furthermore, if a match is encountered, and block requested, full nodes have more targeted insight into what caused a particular match. They could infer that the client received funds in a particular block, e.g., if they are only requesting output scripts. This is above and beyond the additional complexity of now syncing, validating, and managing five or six distinct header/filter-header/filter/block chains. I agree that saving on bandwidth is an important goal, but bandwidth and privacy are always seemingly at odds. Strictly comparing the bandwidth requirements of a system that heavily weighs privacy to existing ones, e.g. BIP39, that don't is a losing battle IMO. I'm not fundamentally opposed to splitting the filters, I certainly see the arguments for flexibility. However, I also want to ensure we are considering the second order effects that fall out of optimizing for one metric when others exist. Cheers, Conner On Wed, May 23, 2018 at 10:29 Gregory Maxwell via bitcoin-dev < bitcoin-dev@lists.linuxfoundation.org> wrote: > Any chance you could add a graph of input-scripts (instead of input > outpoints)? > > On Wed, May 23, 2018 at 7:38 AM, Jim Posen via bitcoin-dev > wrote: > > So I checked filter sizes (as a proportion of block size) for each of the > > sub-filters. The graph is attached. > > > > As interpretation, the first ~120,000 blocks are so small that the > > Golomb-Rice coding can't compress the filters that well, which is why the > > filter sizes are so high proportional to the block size. Except for the > > input filter, because the coinbase input is skipped, so many of them > have 0 > > elements. But after block 120,000 or so, the filter compression converges > > pretty quickly to near the optimal value. The encouraging thing here is > that > > if you look at the ratio of the combined size of the separated filters vs > > the size of a filter containing all of them (currently known as the basic > > filter), they are pretty much the same size. The mean of the ratio > between > > them after block 150,000 is 99.4%. So basically, not much compression > > efficiently is lost by separating the basic filter into sub-filters. > > > > On Tue, May 22, 2018 at 5:42 PM, Jim Posen wrote: > >>> > >>> My suggestion was to advertise a bitfield for each filter type the node > >>> serves, > >>> where the bitfield indicates what elements are part of the filters. > This > >>> essentially > >>> removes the notion of decided filter types and instead leaves the > >>> decision to > >>> full-nodes. > >> > >> > >> I think it makes more sense to construct entirely separate filters for > the > >> different types of elements and allow clients to download only the ones > they > >> care about. If there are enough elements per filter, the compression > ratio > >> shouldn't be much worse by splitting them up. This prevents the > exponential > >> blowup in the number of filters that you mention, Johan, and it works > nicely > >> with service bits for advertising different filter types independently. > >> > >> So if we created three separate filter types, one for output scripts, > one > >> for input outpoints, and one for TXIDs, each signaled with a separate > >> service bit, are people good with that? Or do you think there shouldn't > be a > >> TXID filter at all, Matt? I didn't include the option of a prev output > >> script filter or rolling that into the block output script filter > because it > >> changes the security model (cannot be proven to be correct/incorrect > >> succinctly). > >> > >> Then there's the question of whether to separate or combine the headers. > >> I'd lean towards keeping them separate because it's simpler that way. > > > > > > > > _______________________________________________ > > bitcoin-dev mailing list > > bitcoin-dev@lists.linuxfoundation.org > > https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev > > > _______________________________________________ > bitcoin-dev mailing list > bitcoin-dev@lists.linuxfoundation.org > https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev > --000000000000bad74f056ce93cbc Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

Hi all,

Jimpo, thanks for looking into those stats! I had always imagined that ther= e
would be a more significant savings in having all filters in on= e bundle, as
opposed to separate. These results are interesting, to say = the least, and
= definitely offer us some flexibility in opti= ons for filter sharding.

So far,= the bulk of this discussion has centered around bandwidth. I am
concern= ed, however, that splitting up the filters is at odds with the other
go= al of the proposal in offering improved privacy.

Allowing clients to= choose individual filter sets trivially exposes the type of
data that c= lient is interested in. This alone might be enough to fingerprint the
function of a peer a= nd reduce anonymity set justifying their potential
behavior.

Furt= hermore, if a match is encountered, and block requested, full nodes havemore targeted insight into what caused a particular match. They could infe= r that
the client received funds in a particular block, e.g., if they ar= e only requesting
output scripts.

This is above and beyond the ad= ditional complexity of now syncing, validating,
and managing five or six= distinct header/filter-header/filter/block chains.

I agree that sav= ing on bandwidth is an important goal, but bandwidth and privacy
are alw= ays seemingly at odds. Strictly comparing the bandwidth requirements of
= a system that heavily weighs privacy to existing ones, e.g. BIP39, that don= 't is a
losing battle IMO.

I'm n= ot fundamentally opposed to splitting the filters, I certainly see the
a= rguments for flexibility. However, I also want to ensure we are considering= the
second order effects that fall out of optimizing for one=C2=A0metric when others exis= t.

Cheers,
Conner

On Wed, May 23, 2018 at 10:29 Gregory Maxwell via b= itcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
Any chance you could add a graph of input-= scripts=C2=A0 (instead of input outpoints)?

On Wed, May 23, 2018 at 7:38 AM, Jim Posen via bitcoin-dev
<bitcoin-dev@lists.linuxfoundation.org> wrote:
> So I checked filter sizes (as a proportion of block size) for each of = the
> sub-filters. The graph is attached.
>
> As interpretation, the first ~120,000 blocks are so small that the
> Golomb-Rice coding can't compress the filters that well, which is = why the
> filter sizes are so high proportional to the block size. Except for th= e
> input filter, because the coinbase input is skipped, so many of them h= ave 0
> elements. But after block 120,000 or so, the filter compression conver= ges
> pretty quickly to near the optimal value. The encouraging thing here i= s that
> if you look at the ratio of the combined size of the separated filters= vs
> the size of a filter containing all of them (currently known as the ba= sic
> filter), they are pretty much the same size. The mean of the ratio bet= ween
> them after block 150,000 is 99.4%. So basically, not much compression<= br> > efficiently is lost by separating the basic filter into sub-filters. >
> On Tue, May 22, 2018 at 5:42 PM, Jim Posen <jim.posen@gmail.com> wrote:
>>>
>>> My suggestion was to advertise a bitfield for each filter type= the node
>>> serves,
>>> where the bitfield indicates what elements are part of the fil= ters. This
>>> essentially
>>> removes the notion of decided filter types and instead leaves = the
>>> decision to
>>> full-nodes.
>>
>>
>> I think it makes more sense to construct entirely separate filters= for the
>> different types of elements and allow clients to download only the= ones they
>> care about. If there are enough elements per filter, the compressi= on ratio
>> shouldn't be much worse by splitting them up. This prevents th= e exponential
>> blowup in the number of filters that you mention, Johan, and it wo= rks nicely
>> with service bits for advertising different filter types independe= ntly.
>>
>> So if we created three separate filter types, one for output scrip= ts, one
>> for input outpoints, and one for TXIDs, each signaled with a separ= ate
>> service bit, are people good with that? Or do you think there shou= ldn't be a
>> TXID filter at all, Matt? I didn't include the option of a pre= v output
>> script filter or rolling that into the block output script filter = because it
>> changes the security model (cannot be proven to be correct/incorre= ct
>> succinctly).
>>
>> Then there's the question of whether to separate or combine th= e headers.
>> I'd lean towards keeping them separate because it's simple= r that way.
>
>
>
> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists.linuxfoundation.org
> https://lists.linuxfoundation.org= /mailman/listinfo/bitcoin-dev
>
_______________________________________________
bitcoin-dev mailing list
= bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mail= man/listinfo/bitcoin-dev
--000000000000bad74f056ce93cbc--