Received: from sog-mx-2.v43.ch3.sourceforge.com ([172.29.43.192] helo=mx.sourceforge.net) by sfs-ml-2.v29.ch3.sourceforge.com with esmtp (Exim 4.76) (envelope-from ) id 1X8IvB-0003ZN-7o for bitcoin-development@lists.sourceforge.net; Sat, 19 Jul 2014 00:54:41 +0000 Received-SPF: pass (sog-mx-2.v43.ch3.sourceforge.com: domain of gmail.com designates 209.85.219.42 as permitted sender) client-ip=209.85.219.42; envelope-from=el33th4x0r@gmail.com; helo=mail-oa0-f42.google.com; Received: from mail-oa0-f42.google.com ([209.85.219.42]) by sog-mx-2.v43.ch3.sourceforge.com with esmtps (TLSv1:RC4-SHA:128) (Exim 4.76) id 1X8Iv9-0007BV-I9 for bitcoin-development@lists.sourceforge.net; Sat, 19 Jul 2014 00:54:41 +0000 Received: by mail-oa0-f42.google.com with SMTP id n16so4509147oag.29 for ; Fri, 18 Jul 2014 17:54:34 -0700 (PDT) X-Received: by 10.182.114.131 with SMTP id jg3mr12222931obb.9.1405731274094; Fri, 18 Jul 2014 17:54:34 -0700 (PDT) MIME-Version: 1.0 Received: by 10.76.23.193 with HTTP; Fri, 18 Jul 2014 17:54:14 -0700 (PDT) In-Reply-To: References: From: =?UTF-8?Q?Emin_G=C3=BCn_Sirer?= Date: Fri, 18 Jul 2014 17:54:14 -0700 Message-ID: To: Jeff Garzik Content-Type: multipart/alternative; boundary=001a11c2e3963dff4404fe815190 X-Spam-Score: -0.6 (/) X-Spam-Report: Spam Filtering performed by mx.sourceforge.net. See http://spamassassin.org/tag/ for more details. -1.5 SPF_CHECK_PASS SPF reports sender host as permitted sender for sender-domain 0.0 FREEMAIL_FROM Sender email is commonly abused enduser mail provider (el33th4x0r[at]gmail.com) -0.0 SPF_PASS SPF: sender matches SPF record 1.0 HTML_MESSAGE BODY: HTML included in message -0.1 DKIM_VALID_AU Message has a valid DKIM or DK signature from author's domain 0.1 DKIM_SIGNED Message has a DKIM or DK signature, not necessarily valid -0.1 DKIM_VALID Message has at least one valid DKIM or DK signature X-Headers-End: 1X8Iv9-0007BV-I9 Cc: Bitcoin Dev Subject: Re: [Bitcoin-development] Squashing redundant tx data in blocks on the wire X-BeenThere: bitcoin-development@lists.sourceforge.net X-Mailman-Version: 2.1.9 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 19 Jul 2014 00:54:41 -0000 --001a11c2e3963dff4404fe815190 Content-Type: text/plain; charset=UTF-8 I thought I'd chime in and point out some research results that might help. Even if they don't, there is a cool underlying technique that some of you might find interesting. The problem being tackled here is very similar to "set reconciliation," where peer A thinks that the set of transactions that should be in the block is S_A, and peer B has actually included set S_B, and S_A and S_B are expected to not differ much. Ideally, one would like the communication complexity between A and B to be O(delta), not O(S_B) as it is right now. And ideally, one would like B to send a single message to A, and for A to figure out the difference between the two sets, without any lengthy back and forth communication. In essence, I would like to give you some magical packet that is pretty small and communicates just the delta between what you and I know. This paper from Cornell describes a scheme for achieving this: Yaron Minsky, Ari Trachtenberg, Richard Zippel: Set reconciliation with nearly optimal communication complexity. IEEE Transactions on Information Theory 49(9): 2213-2218 (2003) http://ipsit.bu.edu/documents/ieee-it3-web.pdf Those of you looking for a TL;DR should read the intro and then skip to page 8 for the example. The underlying trick is very cool, comes from the peer-to-peer/gossip literature, and it is underused. It'd be really cool if it could be applied to this problem to reduce the size of the packets. This approach has three benefits over the Bloom filter approach (if I understand the Bloom filter idea correctly): (1) Bloom filters require packets that are still O(S_A), (2) Bloom filters are probabilistic, so require extra complications when there is a hash collision. In the worst case, A might get confused about which transaction B actually included, which would lead to a fork. (I am not sure if I followed the Bloom filter idea fully -- this may not happen with the proposal, but it's a possibility with a naive Bloom filter implementation) (3) Bloom filters are interactive, so when A detects that B has included some transactions that A does not know about, it has to send a message to figure out what those transactions are. Set reconciliation is O(delta), non-probabilistic, and non-interactive. The naive version requires that one have some idea of the size of the delta, but I think the paper has some discussion of how to handle the delta estimate. I have not gone through the full exercise of actually applying this trick to the Bitcoin p2p protocol yet, but wanted to draw your attention to it. If someone is interested in applying this stuff to Bitcoin, I'd be happy to communicate further off list. - egs On Fri, Jul 18, 2014 at 12:55 PM, Jeff Garzik wrote: > Related: We must handle some legitimate miner-privately-mined cases, > such as miner payout TXs (outside coinbase) or side chain conditional > TXs[1]. > > [1] https://bitcointalk.org/index.php?topic=676703.msg7682680#msg7682680 > > On Fri, Jul 18, 2014 at 3:51 PM, Kaz Wesley wrote: > > I've updated the gist, and added an additional proposal that I think > > meshes well: > > > https://gist.github.com/kazcw/43c97d3924326beca87d#ultra-fast-block-validation > > > > sparseblocks + UFBV would tighten the new-block process to this (when > > txes have been received in advance): > > - receive block (~2kB for 1000 tx) > > - check whether block contains txes known to belong to conflict-sets, > > and if so whether more than one tx from a single conflict-set has been > > included (a few operations on very small sets) > > - relay block (~2kB) > > > > The benefits of these changes only occur when the transactions have > > been seen in advance, but incentivizing ahead-of-block transaction > > propogation is a plus, as Jeff mentioned; working on a block without > > first ensuring peers have its transactions would be very expensive > > from a miner's point of view. > > > > > ------------------------------------------------------------------------------ > > Want fast and easy access to all the code in your enterprise? Index and > > search up to 200,000 lines of code with a free copy of Black Duck > > Code Sight - the same software that powers the world's largest code > > search on Ohloh, the Black Duck Open Hub! Try it now. > > http://p.sf.net/sfu/bds > > _______________________________________________ > > Bitcoin-development mailing list > > Bitcoin-development@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > -- > Jeff Garzik > Bitcoin core developer and open source evangelist > BitPay, Inc. https://bitpay.com/ > > > ------------------------------------------------------------------------------ > Want fast and easy access to all the code in your enterprise? Index and > search up to 200,000 lines of code with a free copy of Black Duck > Code Sight - the same software that powers the world's largest code > search on Ohloh, the Black Duck Open Hub! Try it now. > http://p.sf.net/sfu/bds > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > --001a11c2e3963dff4404fe815190 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable
I thought I'd chime in and point out some research re= sults that might help.
Even if they don't, there is a cool underlying technique that some of y= ou
might find interesting.

The problem being tackled = here is very similar to "set reconciliation," where
peer A thinks that the set of transactions that should be in the block= is S_A,
and peer B has actually included set S_B, and S_A and S_= B are expected
to not differ much. Ideally, one would like the co= mmunication complexity
between A and B to be O(delta), not O(S_B) as it is right now. A= nd ideally,
one would like B to send a single message to A, and f= or A to figure out the
difference between the two sets, without a= ny lengthy back and forth=C2=A0
communication. In essence, I would like to give you some magical packe= t
that is pretty small and communicates just the delta between wh= at you and
I know.

This paper from Corne= ll describes a scheme for achieving this:
=C2=A0 =C2=A0Yaron Minsky, Ari Trachtenberg, Richard Zippel: Set recon= ciliation with nearly optimal communication complexity. IEEE Transactions o= n Information Theory 49(9): 2213-2218 (2003)

Those of you looking for a TL;DR should read the = intro and then skip to
page 8 for the example. The underlying tri= ck is very cool, comes from the
peer-to-peer/gossip literature, a= nd it is underused. It'd be really cool if it
could be applied to this problem to reduce the size of the packets.

This approach has three benefits over the Bloom filt= er approach (if I=C2=A0
understand the Bloom filter idea correctl= y):=C2=A0

(1) Bloom filters require packets that are still O(S_A)= ,=C2=A0

(2) Bloom filters are probabilistic, so re= quire extra complications
when there is a hash collision. In the = worst case, A might get confused
about which transaction B actually included, which would lead to a=C2= =A0
fork. (I am not sure if I followed the Bloom filter idea full= y -- this may=C2=A0
not happen with the proposal, but it's a = possibility with a naive Bloom
filter implementation)

(3) Bloom filters are = interactive, so when A detects that B has included
some transacti= ons that A does not know about, it has to send a message
to figur= e out what those transactions are.=C2=A0

Set reconciliation is O(delta), non-probabilistic, and = non-interactive. The
naive version requires that one have some id= ea of the size of the delta,
but I think the paper has some discu= ssion of how to handle the delta=C2=A0
estimate.

I have not gone through the full ex= ercise of actually applying this trick to
the Bitcoin p2p protoco= l yet, but wanted to draw your attention to it.=C2=A0
If someone = is interested in applying this stuff to Bitcoin, I'd be happy=C2=A0
to communicate further off list.

- egs
<= div>


On Fri, Jul 18, 2014 at 12:55 PM, Jeff Garzik <jgarzik@bitpa= y.com> wrote:
Related: =C2=A0We must handle some legitimat= e miner-privately-mined cases,
such as miner payout TXs (outside coinbase) or side chain conditional
TXs[1].

[1] https://bitcointalk.org/index.php?topic=3D676= 703.msg7682680#msg7682680

On Fri, Jul 18, 2014 at 3:51 PM, Kaz Wesley <keziahw@gmail.com> wrote:
> I've updated the gist, and added an additional proposal that I thi= nk
> meshes well:
> https://gist.github.com/kazcw/43c97d= 3924326beca87d#ultra-fast-block-validation
>
> sparseblocks + UFBV would tighten the new-block process to this (when<= br> > txes have been received in advance):
> - receive block (~2kB for 1000 tx)
> - check whether block contains txes known to belong to conflict-sets,<= br> > and if so whether more than one tx from a single conflict-set has been=
> included (a few operations on very small sets)
> - relay block (~2kB)
>
> The benefits of these changes only occur when the transactions have > been seen in advance, but incentivizing ahead-of-block transaction
> propogation is a plus, as Jeff mentioned; working on a block without > first ensuring peers have its transactions would be very expensive
> from a miner's point of view.
>
> ----------------------------------------------------------------------= --------
> Want fast and easy access to all the code in your enterprise? Index an= d
> search up to 200,000 lines of code with a free copy of Black Duck
> Code Sight - the same software that powers the world's largest cod= e
> search on Ohloh, the Black Duck Open Hub! Try it now.
> http://p.sf.net/= sfu/bds
> _______________________________________________
> Bitcoin-development mailing list
> Bitcoin-d= evelopment@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/bitco= in-development



--
Jeff Garzik
Bitcoin core developer and open source evangelist
BitPay, Inc. =C2=A0 =C2=A0 =C2=A0https://bitpay.com/

-----------------------------= -------------------------------------------------
Want fast and easy access to all the code in your enterprise? Index and
search up to 200,000 lines of code with a free copy of Black Duck
Code Sight - the same software that powers the world's largest code
search on Ohloh, the Black Duck Open Hub! Try it now.
http://p.sf.net/sfu/b= ds
_______________________________________________
Bitcoin-development mailing list
Bitcoin-develo= pment@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bitcoin-de= velopment

--001a11c2e3963dff4404fe815190--