94c21e5c05
Single-row reads from large partition issue 64 KiB reads to the data file, which is equal to the default span of the promoted index block in the data file. If users would want to increase selectivity of the index to speed up single-row reads, this won't be effective. The reason is that the reader uses promoted index to look up the start position in the data file of the read, but end position will in practice extend to the next partition, and amount of I/O will be determined by the underlying file input stream implementation and its read-ahead heuristics. By default, that results in at least 2 IOs 32KB each. There is already infrastructure to lookup end position based on upper bound of the read, in anticipation for sharing the promoted index cache, but it's not effective becasue it's a non-populating lookup and the upper bound cursor has its own private cached_promoted_index, which is cold when positions are computed. It's non-populating on purpose, to avoid extra index file IO to read upper bound. In case upper bound is far-enough from the lower bound, this will only increase the cost of the read. The solution employed here is to warm up the lower bound cursor's cache before positions are computed, and use that cursor for non-populating lookup of the upper bound. We use the lower bound cursor and the slice's lower bound so that we read the same blocks as later lower-bound slicing would, so that we don't incur extra IO for cases where looking up upper bound is not worth it, that is when upper bound is far from the lower bound. If upper bound is near lower bound, then warming up using lower bound will populate cached_promoted_index with blocks which will allow us to locate the upper bound block accurately. This is especially important for single-row reads, where the bounds are around the same key. In this case we want to read the data file range which belongs to a single promoted index block. It doesn't matter that the upper bound is not exactly the same. They both will likely lie in the same block, and if not, binary search will bring adjacent blocks into cache. Even if upper bound is not near, the binary search will populate the cache with blocks which can be used to narrow down the data file range somewhat. Fixes #10030. The change was tested with perf-fast-forward. I populated the data set with `column_index_size_in_kb` set to 1 scylla perf-fast-forward --populate --run-tests=large-partition-slicing --column-index-size-in-kb=1 Test run: build/release/scylla perf-fast-forward --run-tests=large-partition-select-few-rows -c1 --keep-cache-across-test-cases --test-case-duration=0 This test issues two reads of subsequent keys from the middle of a large partition (1M rows in total). The first read will miss in the index file page cache, the second read will hit. Notice that before the change, the second read issued 2 aio requests worth of 64KiB in total. After the change, the second read issued 1 aio worth of 2 KiB. That's because promoted index block is larger than 1 KiB. I verified using logging that the data file range matches a single promoted index block. Also, the first read which misses in cache is still faster after the change. Before: ``` running: large-partition-select-few-rows on dataset large-part-ds1 Testing selecting few rows from a large partition: stride rows time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu 500000 1 0.009802 1 1 102 0 102 102 21.0 21 196 2 1 0 1 1 0 0 0 568 269 4716050 53.4% 500001 1 0.000321 1 1 3113 0 3113 3113 2.0 2 64 1 0 1 0 0 0 0 0 116 26 555110 45.0% ``` After: ``` running: large-partition-select-few-rows on dataset large-part-ds1 Testing selecting few rows from a large partition: stride rows time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu 500000 1 0.009609 1 1 104 0 104 104 20.0 20 137 2 1 0 1 1 0 0 0 561 268 4633407 43.1% 500001 1 0.000217 1 1 4602 0 4602 4602 1.0 1 2 1 0 1 0 0 0 0 0 110 26 313882 64.1% ``` Backports: none, not a regression Closes scylladb/scylladb#20522 * github.com:scylladb/scylladb: perf: perf_fast_forward: Add test case for querying missing rows perf-fast-forward: Allow overriding promoted index block size perf-fast-forward: Test subsequent key reads from the middle in test_large_partition_select_few_rows perf-fast-forward: Allow adding key offset in test_large_partition_select_few_rows perf-fast-forward: Use single-partition reads in test_large_partition_select_few_rows sstables: bsearch_clustered_cursor: Add more tracing points sstables: reader: Log data file range sstables: bsearch_clustered_cursor: Unify skip_info logging sstables: bsearch_clustered_cursor: Narrow down range using "end" position of the block sstables: bsearch_clustered_cursor: Skip even to the first block test: sstables: sstable_3_x_test: Improve failure message sstables: mx: writer: Never include partition_end marker in promoted index block width sstables: Reduce amount of I/O for clustering-key-bounded reads from large partitions sstables: clustered_cursor: Track current block
6.2 KiB
File format of the Scylla.db sstable component
The Scylla.db component (present in a file named like mc-223-big-Scylla.db
contains assorted Scylla-only metadata. Its presence indicates the sstable was
created by Scylla (or some Scylla-aware creator). Non-Scylla consumers will ignore it.
The file is small and intended to be processed in-memory.
Main structure
The main structure is that of an unordered set of subcomponents. Each component is prefixed with a be32 tag that indicates its type, and its serialized size (so unknown subcomponents can be skipped).
scylla_db = subcomponent_count (tag serialized_size subcomponent)*
subcomponent_count = be32
serialized_size = be32
tag = be32
Subcomponents and tag values
The following subcomponents are recognized. They are described in more detail in individual sections
subcomponent = sharding_metadata
| features
| extension_attributes
| run_identifier
| large_data_stats
| sstable_origin
| scylla_build_id
| scylla_version
| ext_timestamp_stats
sharding_metadata (tag 1): describes what token sub-ranges are included in this
sstable. This is used, when loading the sstable, to determine which shard(s)
it occupies.
features (tag 2): a set of boolean flags that describe the sstable
extension_attributes (tag 3): a map<string, string> with additional attributes
run_identifier (tag 4): a uuid that is the same for all sstables in the same run
(and different for sstables in different runs).
large_data_stats (tag 5): a map<large_data_type, large_data_stats_entry> with statistics
about large data entities in the sstable.
sstable_origin (tag 6): a string describing the origin of the
sstable ("memtable" for memtable flush, "garbage collection" for
compaction, etc.).
scylla_build_id (tag 7): a string containing the build id of the
Scylla executable that created the sstable.
scylla_version (tag 8): a string containing the version of the
Scylla executable that created the sstable.
ext_timestamp_stats (tag 9): a map<ext_timestamp_stats_type, int64_t> with statistics
about timestamps in the sstable, like: min_live_timestamp, and min_live_row_marker_timestamp.
sstable_identifier (tag 10): a uuid identifying the sstable for its whole lifetime.
It is derived from the sstable uuid generation, upon creation (or uniquely generated
if the sstable has numerical generation). Yet, unlike the sstable that may
change if the sstable is migrated to a different shard or node, the sstable
identifier is stable and copied with the rest of the scylla metadata.
The scylla sstable dump-scylla-metadata tool can be used to dump the scylla metadata in JSON format.
sharding_metadata subcomponent
sharding_metadata = token_range_count token_range*
token_range_count = be32
token_range = left_token_bound right_token_bound
left_token_bound = token_bound
right_token_bound = token_bound
token_bound = exclusive_flag token
exclusive_flag = byte // 0=inclusive, 1=exclusive
token = token_size byte*
token_size = be16
Sharding metadata is a sorted list of disjoint token ranges. Each token range consists of a left bound and a right bound; either bound may be inclusive or exclusive. The tokens are interpreted according to the partitioner.
The sstable contains no partitions whose token is outside the ranges described by sharding_metadata.
features subcomponent
features = be64 // interpreted as a set of bits
bit 0: NonCompoundPIEntries (if set, indicates the sstable was generated by Scylla with issue #2993 fixed)
bit 1: NonCompoundRangeTombstones (if set, indicates the sstable was generated by Scylla with issue #2986 fixed)
bit 2: ShadowableTombstones (if set, indicates the sstable was generated by Scylla with issue #3885 fixed)
bit 3: CorrectStaticCompact (if set, indicates the sstable was generated by Scylla with issue #4139 fixed)
bit 4: CorrectEmptyCounters (if set, indicates the sstable was generated by Scylla with issue #4363 fixed)
bit 5: CorrectUDTsInCollections (if set, indicates that the sstable was generated by Scylla with issue #6130 fixed)
bit 6: CorrectLastPiBlockWidth (if set, indicates that the width of the last promoted index block never includes the partition end marker)
extension_attributes subcomponent
extension_attributes = extension_attribute_count extension_attribute*
extension_attribute_count = be32
extension_attribute = extension_attribute_key extension_attribute_value
extension_attribute_key = string32
extension_attribute_value = string32
string32 = string32_size byte*
string32_size = be32
There are currently no defined attributes.
run_identifier subcomponent
run_identifier = uuid
uuid = uuid_high_bits uuid_low_bits
uuid_high_bits = be64
uuid_low_bits = be64
If the run_identifier subcomponent is present, the sstable is part of a run. All sstables with the same run_identifier belong to the same run. They are guaranteed to be disjoint (non-overlapping) in their partition keys.
large_data_stats subcomponent
large_data_stats = large_data_count large_data_pair*
large_data_count = be32
large_data_pair = large_data_type large_data_stats_entry
large_data_type = partition_size | row_size | cell_size | rows_in_partition | elements_in_collection
partition_size = be32(1) // partition size, in bytes
row_size = be32(2) // row size, in bytes
cell_size = be32(3) // cell size, in bytes
rows_in_partition = be32(4) // number of rows in a partition
elements_in_collection = be32(5) // number of elements in a collection
large_data_stats_entry = max_value threshold above_threshold
max_value = be64
threshold = be64
above_threshold = be32
The large_data_stats component holds statistics about partition, row, and cell sizes and about number of rows in partition. For each entry, it keeps the largest value for the entry type, the respective large_data threshold and the number of entities that are above the threshold.