Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

RLValueEncodingSynExpScripts.sh
压缩方式
编码方式
GZIP
GORILLA
LZ4
PLAIN
SNAPPY
RLE
UNCOMPRESSED
TS_2DIFF
datasetsyntheticsyntheticsyntheticsynthetic
pagePointNum(ppn)10000100001000010000
numOfPagesInChunk(pic)1000100010001000
chunksWritten(cw)10101010
timeEncoding(te)TS_2DIFFTS_2DIFFTS_2DIFFTS_2DIFF
valueDataType(vt)INT64INT64INT64INT64
valueEncoding(ve)
compression
GORILLAPLAINRLETS_2DIFF
compression(co)UNCOMPRESSEDUNCOMPRESSEDUNCOMPRESSEDUNCOMPRESSED
totalPointNum100000000100000000100000000100000000
tsfileSize(MB)805.3812895781.4226151781.8422318793.3244705
chunkDataSize_stats_mean(MB)80.5380362478.1421661478.1841278179.33235168
compressedPageSize_stats_mean(B)84386.25189818748191883122
uncompressedPageSize_stats_mean(B)
timeBufferSize
84386.25189818748191883122
timeBufferSize_stats_mean(B)1872187218721872
valueBufferSize_stats_mean(B)82512.25189800008004481248
[2] category: (A)get ChunkStatistic->(B)load on-disk Chunk->(C)get PageStatistics->(D)load in-memory PageData
[Avg&Per] (A)get_chunkMetadatas91331.98490000001 us - 0.8580518676474486%100944.7581 us - 1.2939556377951902%88098.20449999999 us - 0.9671805828234409%88231.2157 us - 0.9257222461823116%
[Avg&Per] (B)load_on_disk_chunk
[Avg&
5552645.935400001 us - 52.16637107440095%5170158.3812 us - 66.27343223726832%5270914.364100001 us - 57.866400973957255%5526099.6186 us - 57.97985793317826%
[Avg&Per] (C)get_pageHeader8185.805399999992 us - 0.07690455451459878%7712.402700000001 us - 0.09886107156476356%7813.90999999998 us - 0.08578451820695045%7725.986500000003 us - 0.08106107934716146%
[Avg&Per] (D_1)
decompress_pageData
decompress_pageData548160.3352000009 us - 5.149893543905802%525441.0348000005 us - 6.7353412114264035%585036.2351000007 us - 6.422783415941812%632154.6457000006 us - 6.632568914631739%
[Avg&Per] (D_2)decode_pageData4443785.968300003 us - 41.748778959531215%1996996.816400002 us - 25.598409841945312%3156902.088299994 us - 34.65785050907054%3276856.417400001 us - 34.38078982666053%
[3] D_1 compare each step inside
[Avg&Per] (D-1)7_1_data_ByteBuffer_to_ByteArray(us)



[Avg&Per] (D-1)7_2_data_decompress_PageDataByteArray(us)



[Avg&Per] (D-1)7_3_data_ByteArray_to_ByteBuffer(us)



[Avg&Per] (D-1)7_4_data_split_time_value_Buffer(us)



[3] D_2 compare each step inside
[Avg&Per] (D-2)8_1_createBatchData(us)



[Avg&Per] (D-2)8_2_timeDecoder_hasNext(us)



[Avg&Per] (D-2)8_3_timeDecoder_readLong(us)



[Avg&Per] (D-2)8_4_valueDecoder_read(us)



[Avg&Per] (D-2)8_5_checkValueSatisfyOrNot(us)



[Avg&Per] (D-2)8_6_putIntoBatchData(us)




  • 总的来说,使用的人工数据集不太好,其随机生成的取值,编码压缩效率都不高


中车数据实验结果

RLValueEncodingRealExpScripts.sh

ZT11529传感器数据如下图所示,共12,780,287个点。

Image Added

编码方式GORILLAPLAINRLETS_2DIFFGORILLAPLAINRLETS_2DIFF
压缩方式GZIPLZ4SNAPPYUNCOMPRESSED
dataset



pagePointNum(ppn)



numOfPagesInChunk(pic)



chunksWritten(cw)



timeEncoding(te)



valueDataType(vt)



valueEncoding(ve)



compression(co)



totalPointNum



tsfileSize(MB)



chunkDataSize_stats_mean(MB)



compressedPageSize_stats_mean(B)



uncompressedPageSize_stats_mean(B)



timeBufferSize_stats_mean(B)



valueBufferSize_stats_mean(B)



[2] category: (A)get ChunkStatistic->(B)load on-disk Chunk->(C)get PageStatistics->(D)load in-memory PageData
[Avg&Per] (A)get_chunkMetadatas



[Avg&Per] (B)load_on_disk_chunk



[Avg&Per] (C)get_pageHeader



[Avg&Per] (D_1)decompress_pageData



[Avg&Per] (D_2)decode_pageData



[3] D_1 compare each step inside
[Avg&Per] (D-1)7_1_data_ByteBuffer_to_ByteArray(us)



[Avg&Per] (D-1)7_2_data_decompress_PageDataByteArray(us)



[Avg&Per] (D-1)7_3_data_ByteArray_to_ByteBuffer(us)



[Avg&Per] (D-1)7_4_data_split_time_value_Buffer(us)



[3] D_2 compare each step inside
[Avg&Per] (D-2)8_1_createBatchData(us)



[Avg&Per] (D-2)8_2_timeDecoder_hasNext(us)



[Avg&Per] (D-2)8_3_timeDecoder_readLong(us)



[Avg&Per] (D-2)8_4_valueDecoder_read(us)



[Avg&Per] (D-2)8_5_checkValueSatisfyOrNot(us)



[Avg&Per] (D-2)8_6_putIntoBatchData(us)