MemoryStore is used in a MemCachedStore as a persistent layer in tests.
Further commits suppose that persistent storage returns sorted values
from Seek, so sort the result of MemoryStore.Seek.
Benchmark results for 10000 matching items in MemoryStore compared to
master:
name old time/op new time/op delta
MemorySeek-8 712µs ± 0% 3850µs ± 0% +440.52% (p=0.000 n=8+8)
name old alloc/op new alloc/op delta
MemorySeek-8 160kB ± 0% 2724kB ± 0% +1602.61% (p=0.000 n=10+8)
name old allocs/op new allocs/op delta
MemorySeek-8 10.0k ± 0% 10.0k ± 0% +0.24% (p=0.000 n=10+10)
For details on implementation efficiency see the
https://github.com/nspcc-dev/neo-go/pull/2193#discussion_r722993358.
We're using batches in wrong way during persist, we already have all changes
accumulated in two maps and then we move them to batch and then this is
applied. For some DBs like BoltDB this batch is just another MemoryStore, so
we essentially just shuffle the changeset from one map to another, for others
like LevelDB batch is just a serialized set of KV pairs, it doesn't help much
on subsequent PutBatch, we just duplicate the changeset again.
So introduce PutChangeSet that allows to take two maps with sets and deletes
directly. It also allows to simplify MemCachedStore logic.
neo-bench for single node with 10 workers, LevelDB:
Reference:
RPS 30189.132 30556.448 30390.482 ≈ 30379 ± 0.61%
TPS 29427.344 29418.687 29434.273 ≈ 29427 ± 0.03%
CPU % 33.304 27.179 33.860 ≈ 31.45 ± 11.79%
Mem MB 800.677 798.389 715.042 ≈ 771 ± 6.33%
Patched:
RPS 30264.326 30386.364 30166.231 ≈ 30272 ± 0.36% ⇅
TPS 29444.673 29407.440 29452.478 ≈ 29435 ± 0.08% ⇅
CPU % 34.012 32.597 33.467 ≈ 33.36 ± 2.14% ⇅
Mem MB 549.126 523.656 517.684 ≈ 530 ± 3.15% ↓ 31.26%
BoltDB:
Reference:
RPS 31937.647 31551.684 31850.408 ≈ 31780 ± 0.64%
TPS 31292.049 30368.368 31307.724 ≈ 30989 ± 1.74%
CPU % 33.792 22.339 35.887 ≈ 30.67 ± 23.78%
Mem MB 1271.687 1254.472 1215.639 ≈ 1247 ± 2.30%
Patched:
RPS 31746.818 30859.485 31689.761 ≈ 31432 ± 1.58% ⇅
TPS 31271.499 30340.726 30342.568 ≈ 30652 ± 1.75% ⇅
CPU % 34.611 34.414 31.553 ≈ 33.53 ± 5.11% ⇅
Mem MB 1262.960 1231.389 1335.569 ≈ 1277 ± 4.18% ⇅
It's used a lot and it looks a lot like MemoryStore, it just needs not to
return errors from Put and Delete, so make it use MemoryStore internally with
adjusted interface.
Make it look more like a real transaction, put/delete things with a single
lock. Make a copy of value in Put also, just for safety purposes, no one knows
how this value slice can be used after the Put.
Using pointers is just plain wrong here, because the batch can be updated with
newer values for the same keys.
Fixes Seek() to use HasPrefix also because this is the intended behavior.
BoltDB doesn't have internal batching mechanism, thus we have a substitute for
it, but this substitute is absolutely identical to MemoryBatch, so it's better
to unify them and import ac5d2f94d3 fix into the
MemoryBatch.
Commit 578ac414d4 was wrong in that it saved
only a part of the block, so depending on how you use blockchain, you may
still see that the block was not really processed properly. To really fix it
this commit introduces intermediate storage layer in form of memStore, which
actually is a MemoryStore that supports full Store API (thus easily fitting
into the existing code) and one extension that allows it to flush its data to
some other Store.
It also changes AddBlock() semantics in that it only accepts now successive
blocks, but when it does it guarantees that they're properly added into the
Blockchain and can be referred to in any way. Pending block queing is now
moved into the server (see 8c0c055ac657813fe3ed10257bce199e9527d5ed).
So the only thing done with persist() now is just a move from memStore to
Store which probably should've always been the case (notice also that
previously headers and some other metadata was written into the Store
bypassing caching/batching mechanism thus leading to some inefficiency).
add close function to storage interface
add common defer function call which will close db connection
remove context as soon as it's not needed anymore
updated unit tests
* added account_state + changed ECPoint to PublicKey
* account state persist
* in depth test for existing accounts.
* implemented GetTransaction.
* added enrollment TX
* added persist of accounts and unspent coins
* bumped version -> 0.32.0
* Created test_data folder with block json files for testing + create separate file for block base.
* Fixed bug in WriteVarUint + Trim logic + unit tests
* Refactored store and add more tests for it.
* restore headerList from chain file
* Fix tx decode bug + lots of housekeeping.
* Implemented Node restore state from chain file.
* Created standalone package for storage. Added couple more methods to Batch and Store interfaces.
* Block persisting + tests
* bumped version -> 0.31.0