The DataPath System: A Data-Centric Analytic Processing Engine for - - PowerPoint PPT Presentation

▶
the datapath system a data centric analytic processing
SMART_READER_LITE
LIVE PREVIEW

The DataPath System: A Data-Centric Analytic Processing Engine for - - PowerPoint PPT Presentation

Feb 12, 2023 292 likes •544 views

The DataPath System: A Data-Centric Analytic Processing Engine for Large Data Warehouses Subi Arumugam 1 ,Alin Dobra 1 ,Christopher M. Jermaine 2 , Niketan Pansare 2 ,Luis Perez 2 1 University of Florida, 2 Rice University June 9, 2010

slide-1
SLIDE 1

The DataPath System: A Data-Centric Analytic Processing Engine for Large Data Warehouses

Subi Arumugam1,Alin Dobra1,Christopher M. Jermaine2, Niketan Pansare2,Luis Perez2

1University of Florida, 2Rice University

June 9, 2010

slide-2
SLIDE 2

Motivation

  • Storage is cheap: 1TB disk is 80-100$
  • Disks have high throughput
  • 100$ 1TB disk can do 150MB/s reads/writes
  • 4,000$ 1TB SSD (OCZ p88) reads at 1.4GB/s
  • Processors are fast: 6GFLOPs/Core, 24GFrops for 100$
  • TPC-H Q1 ( at 1TB scale factor )
  • 8 Aggregates over 95-97% of lineitem
  • need to read about 160-700GB: 2 P88 scan in 60-250s
  • need to perform 30FLOPs*6 · 109=180GFLOPS; 8s
  • Q1 should be I/O bound; should do 8 in parallel
slide-3
SLIDE 3

Motivation

  • Storage is cheap: 1TB disk is 80-100$
  • Disks have high throughput
  • 100$ 1TB disk can do 150MB/s reads/writes
  • 4,000$ 1TB SSD (OCZ p88) reads at 1.4GB/s
  • Processors are fast: 6GFLOPs/Core, 24GFrops for 100$
  • TPC-H Q1 ( at 1TB scale factor )
  • 8 Aggregates over 95-97% of lineitem
  • need to read about 160-700GB: 2 P88 scan in 60-250s
  • need to perform 30FLOPs*6 · 109=180GFLOPS; 8s
  • Q1 should be I/O bound; should do 8 in parallel
  • Best non-clustered performer: 142s for 1.7M$
  • 64 cores, 512GB memory, 576 disks
slide-4
SLIDE 4

Large Scale Analytics

Goals

  • Deal with analytical queries on large data (1-10TB)
  • Get closer to theoretical CPU performance
  • gap stands at 100-1000 for most databases
  • Sub 100,000$ system with minute response time (1TB)
  • stay I/O bound even with fast disks and multiple queries
  • No or little tuning: no indexing, no tunable partitioning
slide-5
SLIDE 5

Large Scale Analytics

Goals

  • Deal with analytical queries on large data (1-10TB)
  • Get closer to theoretical CPU performance
  • gap stands at 100-1000 for most databases
  • Sub 100,000$ system with minute response time (1TB)
  • stay I/O bound even with fast disks and multiple queries
  • No or little tuning: no indexing, no tunable partitioning

DataPath

  • System designed from ground up to meet these goals.
slide-6
SLIDE 6

Benchmark System

Old System (2008) – 60,000$

  • 8 processors, 32 cores
  • 128GB DDR2 RAM (16 bays)
  • 2 Averatec RAID controlless, 4 12-disk enclosures
  • 47 Velociraptor Disks, 8 Baracuda disks
  • Maximum aggregate throughput 2.2GB/s

New System (2010) – 20,000$

  • 4 processors, 48 cores
  • 128GB DDR3 memory
  • 2 OCZ Z-drive 1TB PCI SSD disks
  • Maximum aggregate throughput 2.8GB/s
slide-7
SLIDE 7

Data-centric Computation

slide-8
SLIDE 8

DataPath Execution Model

  • Tuple-oriented execution model
  • Tuples shared by queries in the system
  • Chunks of tuples pushed into waypoints for processing
  • Waypoints implement operations for multiple queries
  • Tuple processing loops at full CPU speed

for (int i = 0; true; i++) { if (tuple[i].BelongsTo (Q1)) Q1.Process (tuple[i]); if (tuple[i].BelongsTo (Q2)) Q2.Process (tuple[i]); if (tuple[i].BelongsTo (Q3)) Q3.Process (tuple[i]); }

slide-9
SLIDE 9

Query Execution – Example

Q1: SELECT SUM (l quantity)

FROM lineitem WHERE l shipdate > ’1-1-06’;

  • ut

lineitem

  • rders

Q1: l_shipdate > ‘1-1-06’

Q1: SUM(l_quantity)

  • Q1
slide-10
SLIDE 10

Query Execution – Example

Q1: SELECT SUM (l quantity)

FROM lineitem WHERE l shipdate > ’1-1-06’;

Q2: SELECT SUM (l extendedprice)

FROM lineitem, order WHERE l shipmode <> ’rail’ AND o orderdate < ’1-1-08’ AND l orderkey = o orderkey;

  • ut

lineitem

  • rders

Q1: l_shipdate > ‘1-1-06’

Q2: l_shipmode <> ‘rail’

Q2: o_orderdate < ‘1-1-08’

Q2: l_orderkey = o_orderkey Q1: SUM(l_quantity)

  • Q2: SUM(l_extendedprice)
  • ut
  • Q1, Q2

Q2 Q1 Q2

slide-11
SLIDE 11

Query Execution – Example

Q1: SELECT SUM (l quantity)

FROM lineitem WHERE l shipdate > ’1-1-06’;

Q2: SELECT SUM (l extendedprice)

FROM lineitem, order WHERE l shipmode <> ’rail’ AND o orderdate < ’1-1-08’ AND l orderkey = o orderkey;

Q3: SELECT AVG (l discount)

FROM lineitem, orders WHERE

  • custkey = 1234 AND

l orderkey = o orderkey;

  • ut

lineitem

  • rders

Q1: l_shipdate > ‘1-1-06’

Q2: l_shipmode <> ‘rail’

Q2: o_orderdate < ‘1-1-08’

Q2: l_orderkey = o_orderkey Q1: SUM(l_quantity)

  • Q2: SUM(l_extendedprice)
  • ut
  • Q1, Q2

Q2 Q1 Q2

slide-12
SLIDE 12

Tuple Processing Loop

Usual problems:

  • branch mis-prediction
  • instruction cache misses
  • per-tuple overhead

DataPath solution – Use a C++ meta-compiler

  • generate new tuple processing loops for each waypoint

when new queries added

  • code is human-readable (has even comments)
  • compiled as a library with -O3 -msse4.1
  • everything is hardcoded
  • compiler finds sharing, branch-misprediction, SSE
slide-13
SLIDE 13

File Scanner

Staging Area

Chunk 1 Chunk 3 Chunk 4 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 3 2 1 5 4 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-14
SLIDE 14

File Scanner

Staging Area

Chunk 1 Chunk 3 2 Chunk 4 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 3 1 5 4 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-15
SLIDE 15

File Scanner

Staging Area

Chunk 1 Chunk 3 2 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 1 5 4 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-16
SLIDE 16

File Scanner

Staging Area

Chunk 1 Chunk 3 2 5 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 1 4 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-17
SLIDE 17

File Scanner

Staging Area

1 Chunk 1 Chunk 3 2 5 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 4 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-18
SLIDE 18

File Scanner

Staging Area

1 Chunk 1 Chunk 3 2 4 5 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 6 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-19
SLIDE 19

File Scanner

Staging Area

1 Chunk 1 Chunk 3 2 4 5 6 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 10 9 7 8 12 15 11 14 13 20 18 16 17 19 Finished:

slide-20
SLIDE 20

File Scanner

Staging Area

1 Chunk 1 Chunk 3 2 4 5 6 8 Chunk 4 3 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 10 9 7 12 15 11 14 13 20 18 16 17 19 Finished: Chunk 1

slide-21
SLIDE 21

File Scanner

Staging Area

Chunk 5 Chunk 3 5 6 8 Chunk 4 7 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 10 9 12 15 11 14 13 20 18 16 17 19 Finished: Chunk 1

slide-22
SLIDE 22

File Scanner

Staging Area

Chunk 5 Chunk 3 5 6 8 10 Chunk 4 7 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 9 12 15 11 14 13 20 18 16 17 19 Finished: Chunk 1

slide-23
SLIDE 23

File Scanner

Staging Area

Chunk 5 Chunk 3 5 9 6 8 10 Chunk 4 7 Chunk 2

Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 12 15 11 14 13 20 18 16 17 19 Finished: Chunk 1