Presentation is loading. Please wait.

Presentation is loading. Please wait.

@andy_pavlo On Predictive Modeling for D istributed D atabases VLDB - August 28 th, 2012.

Similar presentations


Presentation on theme: "@andy_pavlo On Predictive Modeling for D istributed D atabases VLDB - August 28 th, 2012."— Presentation transcript:

1 @andy_pavlo On Predictive Modeling for D istributed D atabases VLDB - August 28 th, 2012

2 Databases?Evan Jones?

3 Romney has a Swiss bank account! Muammar Gaddafi is in trouble! Putin is going to get re- elected!

4

5 Transact ion Processi ng ?z, High- Volume

6 Main Memory Parallel Shared-Nothing H-Store: A High-Performance, Distributed Main Memory Transaction Processing System Proc. VLDB Endow., vol. 1, iss. 2, pp. 1496-1499, 2008.

7 FastRepetitiveSmall

8 Client Database Cluster Proc. Name Input Params Proc. Name Input Params Transac tion Executi on Transac tion Executi on Database Cluster Transac tion Result Transac tion Result

9 Client Database Cluster P1 P2 P3 P4

10 (txn/s) Magic Mode Assume Single-Part. Assume Distributed TPC-C NewOrder

11 This transactio n will execute 4 queries on partitions 1,3, and 6!

12 Pro Tip: Canadians do not like unnecessar y surgeries.

13 Main Idea: On Predictive Modeling for Optimizing Transaction Execution in Parallel OLTP Systems Proc. VLDB Endow., vol. 5, iss. 2, pp. 85-96, 2011. Use models to predict transactio n behavior before execution.

14 Client Database Cluster

15

16 Step #1: Estimate the path that the transactio n will take.

17 Current State SELECT * FROM WAREHOUSE WHERE W_ID = ? w_id=0 i_w_ids=[0,0] i_ids=[1001,1002] w_id=0 i_w_ids=[0,0] i_ids=[1001,1002] GetWarehouse: Input Parameters:

18 Step #2: Determine which optimizati ons to enable in the DBMS.

19 Optimizations: +1 w_id=0 i_w_ids=[0,0] i_ids=[1001,1002] w_id=0 i_w_ids=[0,0] i_ids=[1001,1002] Best Partition? Touched Partitions? Finished Partitions? Input Parameters:

20 SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID = ?; Current State X w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] CheckStock: Input Parameters: INSERT INTO ORDERS (o_id, o_w_id) VALUES (?, ?); INSERT INTO ORDERS (o_id, o_w_id) VALUES (?, ?); InsertOrder:

21 November 9, 2011

22 =2 w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] =1 =2 ArrayLength(i_ w_ids) =1 =0 HashValue(w_id )

23 SELECT S_QTY FROM STOCK WHERE S_W_ID = ? AND S_I_ID = ?; w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] w_id=0 i_w_ids=[0,1] i_ids=[1001,1002] CheckStock: Input Parameters:

24 Evaluat ion Experimen tal

25 AccuracyOverhead TATP TPC-C AuctionM 94.9% 95.0% 90.2% +1.86% +1.17% +8.15%

26 TATPTPC-CAuctionM (txn/s) +57%+126%+117% HoudiniAssume Single-Partitioned

27 Scaling your OLTP DBMS must come from within. Conclusio n:

28 https://github.com/apavlo/h-store http://hstore.cs.brown.edu

29 November 9, 2011


Download ppt "@andy_pavlo On Predictive Modeling for D istributed D atabases VLDB - August 28 th, 2012."

Similar presentations


Ads by Google