Top Banner
Analyzing the Energy Efficiency of a Database Server D. Tsirogiannis (U of Toronto), S. Harizopoulos, M. Shah (HP Labs), SIGMOD’10 Shimin Chen Big Data Reading Group
18

Shimin Chen Big Data Reading Group. Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Dec 19, 2015

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Analyzing the Energy Efficiency of a Database

ServerD. Tsirogiannis (U of Toronto),

S. Harizopoulos, M. Shah (HP Labs), SIGMOD’10

Shimin ChenBig Data Reading Group

Page 2: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Energy efficiency of:◦ Single-machine instance of DBMS◦ Standard server-grade hardware components◦ A wide spectrum of database tasks

Assess and explore ways to improve energy efficiency

Goals

Page 3: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

HP xw8600 workstation 64-bit Fedora 4 Linux (kernel 2.6.29) Two Intel Xeon E5430 2.66GHz quad core

CPUs (32K L1, 6MB L2) 16GB RAM 4 HDDs (Seagate Savvio 10K.3) 4 SSDs (Intel X-25E)

Machine Configuration

Page 4: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Total system power: ◦ power meter

Individual components: ◦ clamp meter to measure 5V and 12V lines from

the power supply

Power Measurement

Page 5: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Component Power Range:

Page 6: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Configure 4 disks (SSDs) as RAID-0. Read a 100GB file sequentially, varying disk utilization by increasing CPU computation overhead

Power Proportionality of Disks

Page 7: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Consumes 85% of dynamic power Use four micro-benchmarks to study CPU

power Two scheduling policies: Freq adjusted

by OS

CPUs

Page 8: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

• Big jump when a CPU becomes active• Hash join and row scan consumes more power

Page 9: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

• see higher memory bus utilization

Page 10: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

• CPU power is not a linear function of the number of cores used• For a fixed configuration, different operators may differ significantly (60% in the experiments)

Page 11: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Energy efficiency vs. performance for a large number of DB configurations

DB: algorithm kernels, PostgreSQL, commercial System-X

Knobs: ◦ Execution plan selection◦ Intra-operator parallelism (# of cores for a single operator)◦ Inter-query parallelism (# of independent queries in parallel)◦ Physical layout (row vs. column)◦ Storage layout (striping)◦ Choice of storage medium (HDD vs. SDD)

Energy vs. Performance

Page 12: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Dynamic power range among the points is small, 165W + 19%

Page 13: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Again: 169W+14% Therefore the linear relationship

Page 14: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Linear relationship with less than 10% variance

Page 15: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

For this current server, the best performing DB execution plan is also good enough for energy efficiency

This means

Page 16: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

More variance as idle power is reduced

Page 17: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

Power capping leads to more interesting configurations

Page 18: Shimin Chen Big Data Reading Group.  Energy efficiency of: ◦ Single-machine instance of DBMS ◦ Standard server-grade hardware components ◦ A wide spectrum.

This paper studies a DB server representing the current hardware

Results show that performance and energy efficiency are highly co-related.

As server hardware becomes more energy efficient, idle power may reduce, leading to more variance

Power capping also provides interesting research challenges

Summary