zips/zip-0208.rst

410 lines
18 KiB
ReStructuredText
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

::
ZIP: 208
Title: Shorter Block Target Spacing
Owner: Daira Hopwood <daira@electriccoin.co>
Original-Authors: Daira Hopwood
Simon Liu
Status: Final
Category: Consensus
Created: 2019-01-10
License: MIT
Pull-Request: <https://github.com/zcash/zips/pull/237>
Terminology
===========
The key words "MUST" and "SHOULD" in this document are to be interpreted as
described in RFC 2119. [#RFC2119]_
The terms "block chain", "consensus rule change", "consensus branch", and
"network upgrade" are to be interpreted as defined in [#zip-0200]_.
The term "block target spacing" means the time interval between blocks targeted
by the difficulty adjustment algorithm in a given consensus branch. It is normally
measured in seconds. (This is also sometimes called the "target block time",
but "block target spacing" is the term used in the Zcash Protocol Specification
[#protocol-diffadjustment]_.)
The terms "Testnet" and "Mainnet" are to be interpreted as described in
section 3.12 of the Zcash Protocol Specification [#protocol-networks]_.
Abstract
========
This proposal specifies a change in the block target spacing, to take effect in
the Blossom network upgrade [#zip-0206]_.
The emission schedule of mined ZEC will be approximately the same in terms of
time, but this requires the emission per block to be adjusted to take account
of the changed block target spacing.
Motivation
==========
The motivations for decreasing the block target spacing are:
- Reduced latency for considering transactions to be sufficiently confirmed;
- Greater throughput of transactions in unit time.
The latter goal could alternatively be achieved by increasing the block size
limit, but that would not also achieve the former goal.
Note that, for a given security requirement (in terms of the expected cost
distribution of a rollback attack), the number of confirmations needed
increases more slowly than the decrease in block time, and so, up to a point,
decreasing the block target spacing can provide a better trade-off between
latency and security. This argument assumes that the validation and
propagation time for a block remain small compared to the block target spacing.
See [#slowfastblocks]_ for further analysis in various attack models.
Specification
=============
The changes described in this section are to be made in the Zcash Protocol Specification
[#protocol]_, relative to the pre-Blossom specification in [#preblossom-protocol].
Consensus changes
-----------------
Throughout the specification, rename HalvingInterval to PreBlossomHalvingInterval,
and rename PoWTargetSpacing to PreBlossomTargetSpacing. These constants retain
their values from [#preblossom-protocol]_ of 840000 (blocks) and 150 (seconds)
respectively.
In section 2 (Notation), add BlossomActivationHeight and PostBlossomPoWTargetSpacing
to the list of integer constants.
In section 5.3 (Constants), define PostBlossomPoWTargetSpacing := 75 seconds.
For a given network (production or test), define BlossomActivationHeight as the
height at which Blossom activates on that network, as specified in [#zip-0206]_.
In section 7.6.3 (Difficulty adjustment) [later moved to section 7.7.3], make the
following changes:
Define IsBlossomActivated(*height*) to return true if *height* ≥ BlossomActivationHeight,
otherwise false.
This specification assumes that BlossomActivationHeight ≥ SlowStartInterval.
Define:
- BlossomPoWTargetSpacingRatio := PreBlossomPoWTargetSpacing / PostBlossomPoWTargetSpacing
- PostBlossomHalvingInterval := floor(PreBlossomHalvingInterval · BlossomPoWTargetSpacingRatio).
In the same section, redefine PoWTargetSpacing as a function taking a *height*
parameter, as follows:
- PoWTargetSpacing(*height*) :=
- PreBlossomPoWTargetSpacing, if not IsBlossomActivated(*height*)
- PostBlossomPoWTargetSpacing, otherwise.
Also redefine AveragingWindowTimespan, MinActualTimespan, MaxActualTimespan,
ActualTimespanDamped, ActualTimespanBounded, and Threshold as follows:
- add a *height* parameter to each of these functions that does not already
have one;
- ensure that each reference to any of these values, or to PoWTargetSpacing,
are replaced with a function call passing the *height* parameter.
In section 7.7 (Calculation of Block Subsidy and Founders Reward) [later moved
to section 7.8], redefine the Halving and BlockSubsidy functions as follows:
- Halving(*height*) :=
- floor((*height* - SlowStartShift) / PreBlossomHalvingInterval), if not IsBlossomActivated(*height*)
- floor((BlossomActivationHeight - SlowStartShift) / PreBlossomHalvingInterval + (*height* - BlossomActivationHeight) / PostBlossomHalvingInterval), otherwise
- BlockSubsidy(*height*) :=
- SlowStartRate · *height*, if *height* < SlowStartInterval / 2
- SlowStartRate · (*height* + 1), if SlowStartInterval / 2 ≤ *height* and *height* < SlowStartInterval
- floor(MaxBlockSubsidy / 2\ :sup:`Halving(*height*)`\ ), if SlowStartInterval ≤ *height* and not IsBlossomActivated(*height*)
- floor(MaxBlockSubsidy / (BlossomPoWTargetSpacingRatio · 2\ :sup:`Halving(*height*)`\ )), otherwise
Note: BlossomActivationHeight, PostBlossomHalvingInterval, and PostBlossomTargetSpacing are chosen so that:
- (BlossomActivationHeight - SlowStartShift) / PreBlossomHalvingInterval + (*height* - BlossomActivationHeight) / PostBlossomHalvingInterval)
is exactly 1 for some integer *height*.
- MaxBlockSubsidy / (BlossomPoWTargetSpacingRatio · 2\ :sup:`Halving(*height*)`\ )
is an integer for the next few periods.
In section 7.8 (Payment of Founders Reward) [later moved to section 7.9], define:
- FounderAddressAdjustedHeight(*height*) :=
- *height*, if not IsBlossomActivated(*height*)
- BlossomActivationHeight + floor((*height* - BlossomActivationHeight) / BlossomPoWTargetSpacingRatio), otherwise
and in the definition of FounderAddressIndex, replace the use of *height* with FounderAddressAdjustedHeight(*height*).
Also define:
- FoundersRewardLastBlockHeight := max({ *height* ⦂ N | Halving(*height*) < 1 })
Replace the first note in that section with:
- No Founders Reward is required to be paid for *height* > FoundersRewardLastBlockHeight
(i.e. after the first halving), or for *height* = 0 (i.e. the genesis block).
and in the second note, replace SlowStartShift + PreBlossomHalvingInterval - 1 with
FoundersRewardLastBlockHeight.
Effect on difficulty adjustment
-------------------------------
The difficulty adjustment parameters PoWAveragingWindow and PoWMedianBlockSpan
refer to numbers of blocks, but do *not* change at Blossom activation. This is
because the amount of damping/averaging required is expected to be roughly the
same, in terms of the number of blocks, after the change in block target
spacing.
The change in the effective value of PoWTargetSpacing will cause the block
spacing to adjust to the new target, at the normal rate for a difficulty
adjustment. The results of simulations are consistent with this expected
behaviour.
Note that the change in AveragingWindowTimespan(height) takes effect
immediately when calculating the target difficulty starting from the block at
the Blossom activation height, even though the difficulty of the preceding
PoWAveragingWindow blocks will have been adjusted using the pre-Blossom target
spacing. Therefore it is likely that the difficulty adjustment for the first
few blocks after activation will be limited by PoWMaxAdjustDown. This is not
anticipated to cause any problem.
Minimum difficulty blocks on Testnet
''''''''''''''''''''''''''''''''''''
On Testnet from block height 299188 onward, the difficulty adjustment algorithm
[#protocol-diffadjustment]_ allows minimum-difficulty blocks, as described in
[#zip-0205]_, when the block time is greater than a given threshold. This
specification changes this threshold to be proportional to the block target
spacing.
That is, if the block time of a block at height *height* ≥ 299188 is greater than
6 · PoWTargetSpacing(*height*) seconds after that of the preceding block,
then the block is a minimum-difficulty block. In that case its ``nBits`` field
MUST be set to ToCompact(PoWLimit), where PoWLimit is the value defined for Testnet
in section 5.3 of the Zcash Protocol Specification [#protocol-constants]_, and
ToCompact is as defined in section 7.7.4 of that specification [#protocol-nbits]_.
Note: a previous revision of this ZIP (and [#zip-0205]_) incorrectly said that
only the target threshold of minimum-difficulty blocks is affected. In fact
the ``nBits`` field is modified as well, and this affects difficulty adjustment
for subsequent blocks.
This change does not affect Mainnet.
Non-consensus node behaviour
----------------------------
End-of-Service halt
'''''''''''''''''''
`zcashd` implements an "End-of-Service halt" behaviour that halts the node at a
block height that corresponds approximately to a given time after release. This
interval SHOULD be adjusted in releases where the End-of-Service halt time will
follow Blossom activation.
Default expiry delta
''''''''''''''''''''
When not overridden by the ``-txexpirydelta`` option, `zcashd` RPC calls that
create transactions use a default value for the number of blocks after which a
transaction will expire. The default in recent versions of `zcashd` is
20 blocks, which at the pre-Blossom block target spacing corresponds to roughly
50 minutes.
This default SHOULD change to BlossomPoWTargetSpacingRatio · 20 blocks after
Blossom activation, to maintain the approximate expiry time of 50 minutes.
If the ``-txexpirydelta`` option is set, then the set value SHOULD be used both
before and after Blossom activation.
Sprout to Sapling migration
'''''''''''''''''''''''''''
ZIP 308 [#zip-0308]_ defines a procedure for migrating funds from Sprout to
Sapling z-addresses. In that procedure, migration transactions are sent every
500 blocks, which corresponded to roughly 20.83 hours before Blossom.
The 500-block constant has not been changed. Therefore, migration transactions
are now sent roughly every 10.42 hours after Blossom activation. This has been
noted in the ZIP, and a table showing the expected time to complete migration
has been updated accordingly.
Fingerprinting mitigation
'''''''''''''''''''''''''
A "fingerprinting attack" is a network analysis technique in which nodes are
identified across network sessions, for example using information about which
blocks they request or send.
`zcashd` inherits from Bitcoin Core the following behaviour, described in a
comment in ``main.cpp``, intended as a fingerprinting mitigation::
// To prevent fingerprinting attacks, only send blocks outside of the active
// chain if they are valid, and no more than a month older (both in time, and in
// best equivalent proof of work) than the best header chain we know about.
We make no assertion about the significance of fingerprinting for Zcash,
and (despite the word "prevent" in the above comment) no claim about the
effectiveness of this mitigation.
In any case, to estimate the "best equivalent proof of work" of a given block
chain (measured in units of time), we take the total work of the chain as
defined in section 7.7.5 of the Zcash Protocol Specification [#protocol-workdef]_,
divide by the work of the block at the active tip, and multiply by the target
block spacing of that block.
It is not a requirement of the Zcash protocol that this fingerprinting
mitigation is used; however, if it is used, then it SHOULD use the target
block spacing at the same block height that is used for the current work
estimate.
Monitoring for quicker- or slower-than-expected blocks
''''''''''''''''''''''''''''''''''''''''''''''''''''''
`zcashd` previously did this monitoring every 150 seconds; it is now done
every 60 seconds.
Block timeout
'''''''''''''
The timeout for a requested block is calculated as the target block time,
multiplied by 2 + (the number of queued validated headers)/2.
Latency optimization when requesting blocks
'''''''''''''''''''''''''''''''''''''''''''
When `zcashd` sees an announced block that chains from headers that it does
not already have, it will first ask for the headers, and then the block itself.
A latency optimization is performed only if the chain is "nearly synced"::
// First request the headers preceding the announced block. In the normal fully-synced
// case where a new block is announced that succeeds the current tip (no reorganization),
// there are no such headers.
// Secondly, and only when we are close to being synced, we request the announced block directly,
// to avoid an extra round-trip. Note that we must *first* ask for the headers, so by the
// time the block arrives, the header chain leading up to it is already validated. Not
// doing this will result in the received block being rejected as an orphan in case it is
// not a direct successor.
The heuristic for "nearly synced" is that the timestamp of the block at the active tip
is no more than 20 block times before the current "adjusted time". In `zcashd` this
calculation uses the block target spacing as of the best known header. Around Blossom
activation when the block target spacing changes, this could cause the heuristic to be
based on the pre-Blossom block target spacing until the node has synced headers past the
activation block, but this is not anticipated to cause any problem.
Response to getblocks message when pruning
''''''''''''''''''''''''''''''''''''''''''
If pruning is enabled, when `zcashd` responds to an "getblocks" peer-to-peer message,
it will only include blocks that it has on disk, and is likely to still have on disk
an hour after responding to the message::
// If pruning, don't inv blocks unless we have on disk and are likely to still have
// for some reasonable time window (1 hour) that block relay might require.
For each block, when estimating whether it will still be on disk after an hour, we
take MIN_BLOCKS_TO_KEEP = 288 blocks, minus approximately the number of blocks expected
in one hour at the target block spacing as of that block. Around Blossom activation,
this might underestimate the number of blocks in the next hour, but given the value
of MIN_BLOCKS_TO_KEEP, this is not anticipated to cause any problem.
Estimation of fully synced chain height
'''''''''''''''''''''''''''''''''''''''
`zcashd` uses the ``EstimateNetHeight`` function to estimate the approximate height
of the fully synced chain, so that the progress of block download can be displayed to
the node operator. This function has been rewritten, simplified, and changed to take
account of cases where the time period that needs to be estimated crosses Blossom
activation.
Other block-related constants
'''''''''''''''''''''''''''''
The following constants, measured in number of blocks, were reviewed and a
decision was made not to change them::
/** The number of blocks within expiry height when a tx is considered to be expiring soon */
TX_EXPIRING_SOON_THRESHOLD = 3
/** Maximum reorg length we will accept before we shut down and alert the user. */
MAX_REORG_LENGTH = COINBASE_MATURITY - 1;
static const int COINBASE_MATURITY = 100;
/** Number of blocks that can be requested at any given time from a single peer. */
static const int MAX_BLOCKS_IN_TRANSIT_PER_PEER = 16;
static const unsigned int BLOCK_DOWNLOAD_WINDOW = 1024;
/** Block files containing a block-height within MIN_BLOCKS_TO_KEEP of chainActive.Tip() will not be pruned. */
static const unsigned int MIN_BLOCKS_TO_KEEP = 288;
/**
* The period before a network upgrade activates, where connections to upgrading peers are preferred (in blocks).
* This was three days for upgrades up to and including Blossom, and is 1.5 days from Heartwood onward.
*/
static const int NETWORK_UPGRADE_PEER_PREFERENCE_BLOCK_PERIOD = 1728;
Deployment
==========
This proposal will be deployed with the Blossom network upgrade. [#zip-0206]_
Backward compatibility
======================
This proposal intentionally creates what is known as a "bilateral consensus
rule change". Use of this mechanism requires that all network participants
upgrade their software to a compatible version within the upgrade window.
Older software will treat post-upgrade blocks as invalid, and will follow any
pre-upgrade consensus branch that persists.
Reference Implementation
========================
https://github.com/zcash/zcash/pull/4025
References
==========
.. [#RFC2119] `RFC 2119: Key words for use in RFCs to Indicate Requirement Levels <https://www.rfc-editor.org/rfc/rfc2119.html>`_
.. [#preblossom-protocol] `Zcash Protocol Specification, Version 2018.0-beta-37 (exactly) <https://github.com/zcash/zips/blob/9515d73aac0aea3494f77bcd634e1e4fbd744b97/protocol/protocol.pdf>`_
.. [#protocol] `Zcash Protocol Specification, Version 2021.2.16 or later <protocol/protocol.pdf>`_
.. [#protocol-networks] `Zcash Protocol Specification, Version 2021.2.16. Section 3.12: Mainnet and Testnet <protocol/protocol.pdf#networks>`_
.. [#protocol-constants] `Zcash Protocol Specification, Version 2021.2.16. Section 5.3: Constants <protocol/protocol.pdf#constants>`_
.. [#protocol-diffadjustment] `Zcash Protocol Specification, Version 2021.2.16. Section 7.7.3: Difficulty adjustment <protocol/protocol.pdf#diffadjustment>`_
.. [#protocol-nbits] `Zcash Protocol Specification, Version 2021.2.16. Section 7.7.4: nBits conversion <protocol/protocol.pdf#nbits>`_
.. [#protocol-workdef] `Zcash Protocol Specification, Version 2021.2.16. Section 7.7.5: Definition of Work <protocol/protocol.pdf#workdef>`_
.. [#zip-0200] `ZIP 200: Network Upgrade Mechanism <zip-0200.rst>`_
.. [#zip-0205] `ZIP 205: Deployment of the Sapling Network Upgrade <zip-0205.rst>`_
.. [#zip-0206] `ZIP 206: Deployment of the Blossom Network Upgrade <zip-0206.rst>`_
.. [#zip-0308] `ZIP 308: Sprout to Sapling Migration <zip-0308.rst>`_
.. [#slowfastblocks] `On Slow and Fast Block Times <https://blog.ethereum.org/2015/09/14/on-slow-and-fast-block-times/>`_