SYSTEM IS THE ONLY THING I CAN TRUST
![SYSTEM IS THE ONLY THING I CAN TRUST @ANDY_PAVLO SYSTEM IS THE ONLY THING I CAN TRUST @ANDY_PAVLO](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-1.jpg)
![Thirty Years Ago… 2 Thirty Years Ago… 2](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-2.jpg)
![INTERACTIVE TRANSACTIONS SMALL # OF CPU CORES SMALL MEMORY SIZES INTERACTIVE TRANSACTIONS SMALL # OF CPU CORES SMALL MEMORY SIZES](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-3.jpg)
![TPC-C BENCHMARK Warehouse Order Processing APPLICATION New. Order Transaction 1. Check item stock level. TPC-C BENCHMARK Warehouse Order Processing APPLICATION New. Order Transaction 1. Check item stock level.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-4.jpg)
![TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres 15, 000 10, 000 TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres 15, 000 10, 000](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-5.jpg)
![TRADITIONAL DBMS Measured CPU Cycles 30% 28% 12% OLTP THROUGH THE LOOKING GLASS, AND TRADITIONAL DBMS Measured CPU Cycles 30% 28% 12% OLTP THROUGH THE LOOKING GLASS, AND](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-6.jpg)
![HARDWARE UPGRADE REPLICATION DISTRIBUTED CACHE SHARDING MIDDLEWARE NOSQL HARDWARE UPGRADE REPLICATION DISTRIBUTED CACHE SHARDING MIDDLEWARE NOSQL](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-7.jpg)
![HOW TO SCALE UP WITHOUT GIVING UP TRANSACTIONS? HOW TO SCALE UP WITHOUT GIVING UP TRANSACTIONS?](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-8.jpg)
![Distributed Main Memory Transaction Processing System H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING Distributed Main Memory Transaction Processing System H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-9.jpg)
![x. MAIN MEMORY STORAGE DISK ORIENTED i. SERIAL EXECUTION CONCURRENT EXECUTION HEAVYWEIGHT RECOVERY / x. MAIN MEMORY STORAGE DISK ORIENTED i. SERIAL EXECUTION CONCURRENT EXECUTION HEAVYWEIGHT RECOVERY /](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-10.jpg)
![Application STORED PROCEDURE PARTITIONS Transaction Procedure Name Vote. Count: Transaction Input Parameters. Execution SELECT Application STORED PROCEDURE PARTITIONS Transaction Procedure Name Vote. Count: Transaction Input Parameters. Execution SELECT](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-11.jpg)
![TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres H-Store 15, 000 40 TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres H-Store 15, 000 40](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-12.jpg)
![DISTRIBUTED TRANSACTIO NS DISTRIBUTED TRANSACTIO NS](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-13.jpg)
![TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 40, 000 H-Store 30, 000 TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 40, 000 H-Store 30, 000](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-14.jpg)
![DISTRIBUTED TRANSACTIONS Query Count Application P 1 P 2 P 3 P 4 15 DISTRIBUTED TRANSACTIONS Query Count Application P 1 P 2 P 3 P 4 15](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-15.jpg)
![KNOW WHAT TRANSACTIONS WILL DO BEFORE THEY START KNOW WHAT TRANSACTIONS WILL DO BEFORE THEY START](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-16.jpg)
![BUT PEOPLE ALWAYS GIVE ME BAD ADVICE BUT PEOPLE ALWAYS GIVE ME BAD ADVICE](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-17.jpg)
![DON’T GET INVOLVED WITH COMPUTERS. YOU’LL NEVER MAKE ANY MONEY. DON’T GET INVOLVED WITH COMPUTERS. YOU’LL NEVER MAKE ANY MONEY.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-18.jpg)
![DON’T GET A PHD. EVERYONE WILL THINK YOU ARE A JERK. DON’T GET A PHD. EVERYONE WILL THINK YOU ARE A JERK.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-19.jpg)
![THE DATABASE SYSTEM ALWAYS HAS MORE INFORMAT THE DATABASE SYSTEM ALWAYS HAS MORE INFORMAT](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-20.jpg)
![DO USE MACHINE LEARNING TO PREDICT TRANSACTION BEHAVIOR. ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION DO USE MACHINE LEARNING TO PREDICT TRANSACTION BEHAVIOR. ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-21.jpg)
![PREDICTIVE MODELS 22 PREDICTIVE MODELS 22](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-22.jpg)
![L K R WO SELECT AD* FROM O SELECT * FROM WAREHOUSE WHERE W_ID L K R WO SELECT AD* FROM O SELECT * FROM WAREHOUSE WHERE W_ID](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-23.jpg)
![Application DISTRIBUTED TRANSACTIONS 24 Application DISTRIBUTED TRANSACTIONS 24](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-24.jpg)
![TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 25, 000 60, 000 Naïve TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 25, 000 60, 000 Naïve](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-25.jpg)
![DISTRIBUTED TRANSACTIONS … Application Zzzz… SP 1 - Waiting for Query Result SP 2 DISTRIBUTED TRANSACTIONS … Application Zzzz… SP 1 - Waiting for Query Result SP 2](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-26.jpg)
![TRANSACTION STALL POINTS BASE PARTITION 18% 45% 37% SP 1 - Waiting for Query TRANSACTION STALL POINTS BASE PARTITION 18% 45% 37% SP 1 - Waiting for Query](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-27.jpg)
![DO SOMETHING USEFUL WHEN STALLED DO SOMETHING USEFUL WHEN STALLED](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-28.jpg)
![DON’T BE SURPRISED IF YOU & KB DON’T LAST THROUGH GRAD SCHOOL. DON’T BE SURPRISED IF YOU & KB DON’T LAST THROUGH GRAD SCHOOL.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-29.jpg)
![DON’T BE STAN’S STUDENT IF YOU GO TO BROWN. DON’T BE STAN’S STUDENT IF YOU GO TO BROWN.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-30.jpg)
![DO USE MACHINE LEARNING TO SCHEDULE SPECULATIVE TASKS. THE ART OF SPECULATIVE EXECUTION In DO USE MACHINE LEARNING TO SCHEDULE SPECULATIVE TASKS. THE ART OF SPECULATIVE EXECUTION In](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-31.jpg)
![SERIALIZABLE SCHEDULE Distributed Transaction Zzzz… VERIFY C Single-Partition Transaction Speculative Transaction C C Speculative SERIALIZABLE SCHEDULE Distributed Transaction Zzzz… VERIFY C Single-Partition Transaction Speculative Transaction C C Speculative](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-32.jpg)
![Zzzz… SPECULATIVE TRANSACTIONS … Transaction Queue Speculation Candidate: WRITE X Distributed Transaction: READ X Zzzz… SPECULATIVE TRANSACTIONS … Transaction Queue Speculation Candidate: WRITE X Distributed Transaction: READ X](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-33.jpg)
![SPECULATIVE QUERIES Query. Y: SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID SPECULATIVE QUERIES Query. Y: SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-34.jpg)
![Transaction Parameters: w_id=0 i_w_ids=[1, 0] i_ids=[1001, 1002] Get. Warehouse: Check. Stock: SELECT WHERE AND Transaction Parameters: w_id=0 i_w_ids=[1, 0] i_ids=[1001, 1002] Get. Warehouse: Check. Stock: SELECT WHERE AND](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-35.jpg)
![VERIFICATION Distributed Transaction Speculative Transactions Query 1 Query 2 Query 3 Query 1 Query VERIFICATION Distributed Transaction Speculative Transactions Query 1 Query 2 Query 3 Query 1 Query](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-36.jpg)
![TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 50, 000 None Spec Queries TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 50, 000 None Spec Queries](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-37.jpg)
![Optimize Single-Partition Execution H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc. VLDB Optimize Single-Partition Execution H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc. VLDB](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-38.jpg)
![FUTURE WORK FUTURE WORK](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-39.jpg)
![One Size Almost Fits All One Size Almost Fits All](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-40.jpg)
![NN- HSSTORE 41 NN- HSSTORE 41](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-41.jpg)
![One Size Almost ™ Fits All One Size Almost ™ Fits All](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-42.jpg)
![NN- HSSTORE 43 NN- HSSTORE 43](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-43.jpg)
![Escape From Planet Zdonik (i. e. , Andy Needs to Get Tenure) Escape From Planet Zdonik (i. e. , Andy Needs to Get Tenure)](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-44.jpg)
![Beyond the ‘Stores • Non-Partitionable Workloads. • The Poor Man’s Spanner. • Scientific Databases. Beyond the ‘Stores • Non-Partitionable Workloads. • The Poor Man’s Spanner. • Scientific Databases.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-45.jpg)
![DON’T MESS IT UP WITH KB. DON’T MESS IT UP WITH KB.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-46.jpg)
![Stan Zdonik Evan Jones “The Thrill” Stonebraker Saurya Velagapudi Xin Jia Sam Madden Ugur Stan Zdonik Evan Jones “The Thrill” Stonebraker Saurya Velagapudi Xin Jia Sam Madden Ugur](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-47.jpg)
- Slides: 47
![SYSTEM IS THE ONLY THING I CAN TRUST ANDYPAVLO SYSTEM IS THE ONLY THING I CAN TRUST @ANDY_PAVLO](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-1.jpg)
SYSTEM IS THE ONLY THING I CAN TRUST @ANDY_PAVLO
![Thirty Years Ago 2 Thirty Years Ago… 2](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-2.jpg)
Thirty Years Ago… 2
![INTERACTIVE TRANSACTIONS SMALL OF CPU CORES SMALL MEMORY SIZES INTERACTIVE TRANSACTIONS SMALL # OF CPU CORES SMALL MEMORY SIZES](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-3.jpg)
INTERACTIVE TRANSACTIONS SMALL # OF CPU CORES SMALL MEMORY SIZES
![TPCC BENCHMARK Warehouse Order Processing APPLICATION New Order Transaction 1 Check item stock level TPC-C BENCHMARK Warehouse Order Processing APPLICATION New. Order Transaction 1. Check item stock level.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-4.jpg)
TPC-C BENCHMARK Warehouse Order Processing APPLICATION New. Order Transaction 1. Check item stock level. 2. Create new order information. 3. Update item stock levels. 4
![TPCC BENCHMARK Warehouse Order Processing 20 000 My SQL Postgres 15 000 10 000 TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres 15, 000 10, 000](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-5.jpg)
TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres 15, 000 10, 000 5, 000 0 TXN/SEC 1 2 3 4 5 6 7 8 CPU CORES 9 10 11 12 5
![TRADITIONAL DBMS Measured CPU Cycles 30 28 12 OLTP THROUGH THE LOOKING GLASS AND TRADITIONAL DBMS Measured CPU Cycles 30% 28% 12% OLTP THROUGH THE LOOKING GLASS, AND](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-6.jpg)
TRADITIONAL DBMS Measured CPU Cycles 30% 28% 12% OLTP THROUGH THE LOOKING GLASS, AND WHAT WE FOUND THERE SIGMOD, pp. 981 -992, 2008. Buffer Pool Locking Recovery 6
![HARDWARE UPGRADE REPLICATION DISTRIBUTED CACHE SHARDING MIDDLEWARE NOSQL HARDWARE UPGRADE REPLICATION DISTRIBUTED CACHE SHARDING MIDDLEWARE NOSQL](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-7.jpg)
HARDWARE UPGRADE REPLICATION DISTRIBUTED CACHE SHARDING MIDDLEWARE NOSQL
![HOW TO SCALE UP WITHOUT GIVING UP TRANSACTIONS HOW TO SCALE UP WITHOUT GIVING UP TRANSACTIONS?](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-8.jpg)
HOW TO SCALE UP WITHOUT GIVING UP TRANSACTIONS?
![Distributed Main Memory Transaction Processing System HSTORE A HIGHPERFORMANCE DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING Distributed Main Memory Transaction Processing System H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-9.jpg)
Distributed Main Memory Transaction Processing System H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc. VLDB Endow. , vol. 1, iss. 2, pp. 1496 -1499, 2008.
![x MAIN MEMORY STORAGE DISK ORIENTED i SERIAL EXECUTION CONCURRENT EXECUTION HEAVYWEIGHT RECOVERY x. MAIN MEMORY STORAGE DISK ORIENTED i. SERIAL EXECUTION CONCURRENT EXECUTION HEAVYWEIGHT RECOVERY /](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-10.jpg)
x. MAIN MEMORY STORAGE DISK ORIENTED i. SERIAL EXECUTION CONCURRENT EXECUTION HEAVYWEIGHT RECOVERY / COMPACT LOGGING
![Application STORED PROCEDURE PARTITIONS Transaction Procedure Name Vote Count Transaction Input Parameters Execution SELECT Application STORED PROCEDURE PARTITIONS Transaction Procedure Name Vote. Count: Transaction Input Parameters. Execution SELECT](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-11.jpg)
Application STORED PROCEDURE PARTITIONS Transaction Procedure Name Vote. Count: Transaction Input Parameters. Execution SELECT COUNT(*) FROM votes WHERE phone_num = ? ; Result Insert. Vote: INSERT INTO votes VALUES (? , ? ); run(phone. Num, contestant. Id, current. Time) { result = execute(Vote. Count, phone. Num); if (result > MAX_VOTES) { return (ERROR); } execute(Insert. Vote, phone. Num, contestant. Id, current. Time); return (SUCCESS); } SINGLE-THREADED CMD SNAPSHOTS LOG EXECUTION ENGINES 11
![TPCC BENCHMARK Warehouse Order Processing 20 000 My SQL Postgres HStore 15 000 40 TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres H-Store 15, 000 40](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-12.jpg)
TPC-C BENCHMARK Warehouse Order Processing 20, 000 My. SQL Postgres H-Store 15, 000 40 x 10, 000 5, 000 0 TXN/SEC 1 2 3 4 5 6 7 8 CPU CORES 9 10 11 12 12
![DISTRIBUTED TRANSACTIO NS DISTRIBUTED TRANSACTIO NS](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-13.jpg)
DISTRIBUTED TRANSACTIO NS
![TPCC BENCHMARK 8 Cores per Node 10 Distributed Transactions 40 000 HStore 30 000 TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 40, 000 H-Store 30, 000](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-14.jpg)
TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 40, 000 H-Store 30, 000 20, 000 10, 000 0 TXN/SEC 1 2 NODES 3 4 14
![DISTRIBUTED TRANSACTIONS Query Count Application P 1 P 2 P 3 P 4 15 DISTRIBUTED TRANSACTIONS Query Count Application P 1 P 2 P 3 P 4 15](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-15.jpg)
DISTRIBUTED TRANSACTIONS Query Count Application P 1 P 2 P 3 P 4 15
![KNOW WHAT TRANSACTIONS WILL DO BEFORE THEY START KNOW WHAT TRANSACTIONS WILL DO BEFORE THEY START](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-16.jpg)
KNOW WHAT TRANSACTIONS WILL DO BEFORE THEY START
![BUT PEOPLE ALWAYS GIVE ME BAD ADVICE BUT PEOPLE ALWAYS GIVE ME BAD ADVICE](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-17.jpg)
BUT PEOPLE ALWAYS GIVE ME BAD ADVICE
![DONT GET INVOLVED WITH COMPUTERS YOULL NEVER MAKE ANY MONEY DON’T GET INVOLVED WITH COMPUTERS. YOU’LL NEVER MAKE ANY MONEY.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-18.jpg)
DON’T GET INVOLVED WITH COMPUTERS. YOU’LL NEVER MAKE ANY MONEY.
![DONT GET A PHD EVERYONE WILL THINK YOU ARE A JERK DON’T GET A PHD. EVERYONE WILL THINK YOU ARE A JERK.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-19.jpg)
DON’T GET A PHD. EVERYONE WILL THINK YOU ARE A JERK.
![THE DATABASE SYSTEM ALWAYS HAS MORE INFORMAT THE DATABASE SYSTEM ALWAYS HAS MORE INFORMAT](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-20.jpg)
THE DATABASE SYSTEM ALWAYS HAS MORE INFORMAT
![DO USE MACHINE LEARNING TO PREDICT TRANSACTION BEHAVIOR ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION DO USE MACHINE LEARNING TO PREDICT TRANSACTION BEHAVIOR. ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-21.jpg)
DO USE MACHINE LEARNING TO PREDICT TRANSACTION BEHAVIOR. ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION EXECUTION IN PARALLEL OLTP SYSTEMS Proc. VLDB Endow. , Vol 5, Iss. 2, pp. 85 -96, 2011
![PREDICTIVE MODELS 22 PREDICTIVE MODELS 22](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-22.jpg)
PREDICTIVE MODELS 22
![L K R WO SELECT AD FROM O SELECT FROM WAREHOUSE WHERE WID L K R WO SELECT AD* FROM O SELECT * FROM WAREHOUSE WHERE W_ID](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-23.jpg)
L K R WO SELECT AD* FROM O SELECT * FROM WAREHOUSE WHERE W_ID = 10; * FROM SELECT WAREHOUSE WHERE W_ID = 10; SELECT * FROM DISTRICT SELECT W_ID* FROM = 10; DISTRICT WHERE D_W_ID = 10 D_ID AND INSERT INTO ORDERS D_W_ID = 10 AND INSERT INTO ORDERS D_ID =9; O_D_ID, =9; (O_W_ID, O_D_ID, O_C_ID) VALUES (10, 9, INSERT INTO O_C_ID) VALUES INSERT INTOORDERS(10, 9, 12345); (O_W_ID, 12345); O_D_ID, (O_W_ID, O_D_ID, ⋮ O_C_ID, …) O_C_ID) VALUES ⋮ (10, 9, 12345); (10, 9, 12345, …); Feature Clusterer ⋮⋮ Decision Tree Feature 2 ______ ______ ______ ______ ______ ______ ______ ______ ______ Model Generator Classifier Feature 1 Feature 2 Markov Models 23
![Application DISTRIBUTED TRANSACTIONS 24 Application DISTRIBUTED TRANSACTIONS 24](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-24.jpg)
Application DISTRIBUTED TRANSACTIONS 24
![TPCC BENCHMARK 8 Cores per Node 10 Distributed Transactions 25 000 60 000 Naïve TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 25, 000 60, 000 Naïve](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-25.jpg)
TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 25, 000 60, 000 Naïve OPTIMAL Houdini 20, 000 45, 000 2 x 15, 000 30, 000 15, 000 0 TXN/SEC 1 2 NODES 3 4 25
![DISTRIBUTED TRANSACTIONS Application Zzzz SP 1 Waiting for Query Result SP 2 DISTRIBUTED TRANSACTIONS … Application Zzzz… SP 1 - Waiting for Query Result SP 2](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-26.jpg)
DISTRIBUTED TRANSACTIONS … Application Zzzz… SP 1 - Waiting for Query Result SP 2 - Waiting for Query Request SP 3 - Two-Phase Commit 26
![TRANSACTION STALL POINTS BASE PARTITION 18 45 37 SP 1 Waiting for Query TRANSACTION STALL POINTS BASE PARTITION 18% 45% 37% SP 1 - Waiting for Query](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-27.jpg)
TRANSACTION STALL POINTS BASE PARTITION 18% 45% 37% SP 1 - Waiting for Query Result SP 3 - Two-Phase Commit REMOTE PARTITION 73% 5% 22% SP 2 - Waiting for Query Request Real Work 27
![DO SOMETHING USEFUL WHEN STALLED DO SOMETHING USEFUL WHEN STALLED](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-28.jpg)
DO SOMETHING USEFUL WHEN STALLED
![DONT BE SURPRISED IF YOU KB DONT LAST THROUGH GRAD SCHOOL DON’T BE SURPRISED IF YOU & KB DON’T LAST THROUGH GRAD SCHOOL.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-29.jpg)
DON’T BE SURPRISED IF YOU & KB DON’T LAST THROUGH GRAD SCHOOL.
![DONT BE STANS STUDENT IF YOU GO TO BROWN DON’T BE STAN’S STUDENT IF YOU GO TO BROWN.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-30.jpg)
DON’T BE STAN’S STUDENT IF YOU GO TO BROWN.
![DO USE MACHINE LEARNING TO SCHEDULE SPECULATIVE TASKS THE ART OF SPECULATIVE EXECUTION In DO USE MACHINE LEARNING TO SCHEDULE SPECULATIVE TASKS. THE ART OF SPECULATIVE EXECUTION In](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-31.jpg)
DO USE MACHINE LEARNING TO SCHEDULE SPECULATIVE TASKS. THE ART OF SPECULATIVE EXECUTION In Progress (August 2013)
![SERIALIZABLE SCHEDULE Distributed Transaction Zzzz VERIFY C SinglePartition Transaction Speculative Transaction C C Speculative SERIALIZABLE SCHEDULE Distributed Transaction Zzzz… VERIFY C Single-Partition Transaction Speculative Transaction C C Speculative](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-32.jpg)
SERIALIZABLE SCHEDULE Distributed Transaction Zzzz… VERIFY C Single-Partition Transaction Speculative Transaction C C Speculative Single-Partition Transaction C C 32
![Zzzz SPECULATIVE TRANSACTIONS Transaction Queue Speculation Candidate WRITE X Distributed Transaction READ X Zzzz… SPECULATIVE TRANSACTIONS … Transaction Queue Speculation Candidate: WRITE X Distributed Transaction: READ X](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-33.jpg)
Zzzz… SPECULATIVE TRANSACTIONS … Transaction Queue Speculation Candidate: WRITE X Distributed Transaction: READ X 33
![SPECULATIVE QUERIES Query Y SELECT SQTY FROM STOCK WHERE SWID AND SIID SPECULATIVE QUERIES Query. Y: SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-34.jpg)
SPECULATIVE QUERIES Query. Y: SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID = ? ; Distributed Transaction: 34
![Transaction Parameters wid0 iwids1 0 iids1001 1002 Get Warehouse Check Stock SELECT WHERE AND Transaction Parameters: w_id=0 i_w_ids=[1, 0] i_ids=[1001, 1002] Get. Warehouse: Check. Stock: SELECT WHERE AND](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-35.jpg)
Transaction Parameters: w_id=0 i_w_ids=[1, 0] i_ids=[1001, 1002] Get. Warehouse: Check. Stock: SELECT WHERE AND * FROM WAREHOUSE S_QTY STOCK W_ID = ? S_I_ID = ? ; 35
![VERIFICATION Distributed Transaction Speculative Transactions Query 1 Query 2 Query 3 Query 1 Query VERIFICATION Distributed Transaction Speculative Transactions Query 1 Query 2 Query 3 Query 1 Query](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-36.jpg)
VERIFICATION Distributed Transaction Speculative Transactions Query 1 Query 2 Query 3 Query 1 Query 3 Query 1 Query 2 Query 3 36
![TPCC BENCHMARK 8 Cores per Node 10 Distributed Transactions 50 000 None Spec Queries TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 50, 000 None Spec Queries](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-37.jpg)
TPC-C BENCHMARK 8 Cores per Node 10% Distributed Transactions 50, 000 None Spec Queries Spec Txns All 40, 000 30, 000 20, 000 10, 000 0 TXN/SEC 1 2 NODES 3 4 37
![Optimize SinglePartition Execution HSTORE A HIGHPERFORMANCE DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc VLDB Optimize Single-Partition Execution H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc. VLDB](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-38.jpg)
Optimize Single-Partition Execution H-STORE: A HIGH-PERFORMANCE, DISTRIBUTED MAIN MEMORY TRANSACTION PROCESSING SYSTEM Proc. VLDB Endow. , vol. 1, iss. 2, pp. 1496 -1499, 2008. Minimize Distributed Transactions SKEW-AWARE AUTOMATIC DATABASE PARTITIONING IN SHARED-NOTHING, PARALLEL OLTP SYSTEMS Proceedings of SIGMOD, 2012. Identify Distributed Transactions ON PREDICTIVE MODELING FOR OPTIMIZING TRANSACTION EXECUTION IN PARALLEL OLTP SYSTEMS Proc. VLDB Endow. , vol. 5, pp. 85 -96, 2011. Utilize Transaction Stalls THE ART OF SPECULATIVE EXECUTION In Progress (August 2013)
![FUTURE WORK FUTURE WORK](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-39.jpg)
FUTURE WORK
![One Size Almost Fits All One Size Almost Fits All](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-40.jpg)
One Size Almost Fits All
![NN HSSTORE 41 NN- HSSTORE 41](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-41.jpg)
NN- HSSTORE 41
![One Size Almost Fits All One Size Almost ™ Fits All](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-42.jpg)
One Size Almost ™ Fits All
![NN HSSTORE 43 NN- HSSTORE 43](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-43.jpg)
NN- HSSTORE 43
![Escape From Planet Zdonik i e Andy Needs to Get Tenure Escape From Planet Zdonik (i. e. , Andy Needs to Get Tenure)](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-44.jpg)
Escape From Planet Zdonik (i. e. , Andy Needs to Get Tenure)
![Beyond the Stores NonPartitionable Workloads The Poor Mans Spanner Scientific Databases Beyond the ‘Stores • Non-Partitionable Workloads. • The Poor Man’s Spanner. • Scientific Databases.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-45.jpg)
Beyond the ‘Stores • Non-Partitionable Workloads. • The Poor Man’s Spanner. • Scientific Databases.
![DONT MESS IT UP WITH KB DON’T MESS IT UP WITH KB.](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-46.jpg)
DON’T MESS IT UP WITH KB.
![Stan Zdonik Evan Jones The Thrill Stonebraker Saurya Velagapudi Xin Jia Sam Madden Ugur Stan Zdonik Evan Jones “The Thrill” Stonebraker Saurya Velagapudi Xin Jia Sam Madden Ugur](https://slidetodoc.com/presentation_image_h/078b792ad57f14dd5de7858343438938/image-47.jpg)
Stan Zdonik Evan Jones “The Thrill” Stonebraker Saurya Velagapudi Xin Jia Sam Madden Ugur Cetintemel Carlo Justin Curino De. Brabant Yang Zou David De. Witt Visawee Angkana. Dan Abadi Ning Shi John Meehan
Trust is the only thing
If i could only teach you one thing
Melodia impossible
Charitable work
A material thing that can be seen or touched
Leave only footprints take only photos
Gives you wings slogan
Value you can trust
Hình ảnh bộ gõ cơ thể búng tay
Ng-html
Bổ thể
Tỉ lệ cơ thể trẻ em
Voi kéo gỗ như thế nào
Tư thế worm breton là gì
Chúa yêu trần thế alleluia
Các môn thể thao bắt đầu bằng từ đua
Thế nào là hệ số cao nhất
Các châu lục và đại dương trên thế giới
Công thức tính thế năng
Trời xanh đây là của chúng ta thể thơ
Cách giải mật thư tọa độ
Phép trừ bù
Phản ứng thế ankan
Các châu lục và đại dương trên thế giới
Thể thơ truyền thống
Quá trình desamine hóa có thể tạo ra
Một số thể thơ truyền thống
Cái miệng nó xinh thế
Vẽ hình chiếu vuông góc của vật thể sau
Biện pháp chống mỏi cơ
đặc điểm cơ thể của người tối cổ
Ví dụ về giọng cùng tên
Vẽ hình chiếu đứng bằng cạnh của vật thể
Phối cảnh
Thẻ vin
đại từ thay thế
điện thế nghỉ
Tư thế ngồi viết
Diễn thế sinh thái là
Dạng đột biến một nhiễm là
Số.nguyên tố
Tư thế ngồi viết
Lời thề hippocrates
Thiếu nhi thế giới liên hoan
ưu thế lai là gì
Khi nào hổ mẹ dạy hổ con săn mồi
Sự nuôi và dạy con của hươu
Sơ đồ cơ thể người