Eliminate TX trickle bypass, sort TX invs for privacy and priority. #7805
Conversation
|
Concept ACK |
|
|
||
| bool operator()(const CInv &a, const CInv &b) | ||
| { | ||
| if (a.type != MSG_TX) { |
There was a problem hiding this comment.
Have you verified that using this comparator does not change the order of blocks?
Alternatively, if both entries are blocks, perhaps they should be sorted by height?
There was a problem hiding this comment.
It probably does change it (the sort isn't stable). I'll make it sort them by height... or I could split transactions off entirely.
|
@gmaxwell As long as we intend to support serving blocks to pre-0.10 (and especially 0.9 with the limited block orphan pools), we should not reorder block invs. |
|
Concept ACK. Nit: looks like most of these functions should be const. |
|
Updated to make the sort stable and use static variables. |
| /** Average delay between trickled inventory transmissions in seconds. | ||
| * Blocks and whitelisted receivers bypass this, outbound peers get half this delay. */ | ||
| static const unsigned int INVENTORY_BROADCAST_INTERVAL = 5; | ||
| /** Maximum number of inventor items to send per transmission. |
|
Concept ACK, going to test. |
Previously Bitcoin would send 1/4 of transactions out to all peers instantly. This causes high overhead because it makes >80% of INVs size 1. Doing so harms privacy, because it limits the amount of source obscurity a transaction can receive. These randomized broadcasts also disobeyed transaction dependencies and required use of the orphan pool. Because the orphan pool is so small this leads to poor propagation for dependent transactions. When the bypass wasn't in effect, transactions were sent in the order they were received. This avoided creating orphans but undermines privacy fairly significantly. This commit: Eliminates the bypass. The bypass is replaced by halving the average delay for outbound peers. Sorts candidate transactions for INV by their topological depth then by their feerate (then hash); removing the information leakage and providing priority service to higher fee transactions. Limits the amount of transactions sent in a single INV to 7tx/sec (and twice that for outbound); this limits the harm of low fee transaction floods, gives faster relay service to higher fee transactions. The 7 sounds lower than it really is because received advertisements need not be sent, and because the aggregate rate is multipled by the number of peers.
|
I think it's updated for all comments now. |
|
Midly-tested ACK using sipa@115157b to monitor the sizes of invs-to-send. 80% of the observed queues have less than 10 elements, 99.3% have less than 100, 99.9% have less than 1000. The highest observed was 2698. The average is 10.05 elements, with standard deviation 16.65. |
|
Closing in favor of #7840. |
Previously Bitcoin would send 1/4 of transactions out to all peers
instantly. This causes high overhead because it makes >80% of
INVs size 1. Doing so harms privacy, because it limits the
amount of source obscurity a transaction can receive.
These randomized broadcasts also disobeyed transaction dependencies
and required use of the orphan pool. Because the orphan pool is
so small this leads to poor propagation for dependent transactions.
When the bypass wasn't in effect, transactions were sent in the
order they were received. This avoided creating orphans but
undermines privacy fairly significantly.
This commit:
Eliminates the bypass. The bypass is replaced by halving the
average delay for outbound peers.
Sorts candidate transactions for INV by their topological
depth then by their feerate (then hash); removing the
information leakage and providing priority service to
higher fee transactions.
Limits the amount of transactions sent in a single INV to
7tx/sec (and twice that for outbound); this limits the
harm of low fee transaction floods, gives faster relay
service to higher fee transactions. The 7 sounds lower
than it really is because received advertisements need
not be sent, and because the aggregate rate is multipled
by the number of peers.