[PPT] - Dynamic Resource Allocation for Database Servers Running on Virtual PowerPoint Presentation

SLIDE 1

Dynamic Resource Allocation for Database Servers Running on Virtual Storage

Gokul Soundararajan, Daniel Lupei, Saeed Ghanbari, Adrian Daniel Popescu, Jin Chen, Cristiana Amza University of Toronto

1

SLIDE 2

Multi-tier Resource Allocation

Consolidated Environment

Web Server Application Server Database Server

2

Composed of several tiers Application-B Application-A Share resources in each tier Can lead to interference

SLIDE 3

Storage Database

Our Focus: Storage Hierarchy

Application-A Application-B

3

Buffer Pool Storage Cache Disk Bandwidth Want to use all resources efficiently Disk is a bottleneck for Database Apps

Network

SLIDE 4

State of the Art

Previous work studied resources in isolation
Memory Partitioning: MRC [ASPLOS’04]
Disk Bandwidth: Facade [FAST’03], Argon [FAST’07], etc.
... and many more
Want to use the storage hierarchy efficiently
However, performance depends on all layers
Interdependency between resources
E.g., Increasing buffer pool reduces number of storage accesses

4

SLIDE 5

Motivating Scenario

Small Large

5

Buffer Pool Storage Cache Disk Bandwidth Cache Friendly 1 Outstanding I/O Cache Un-Friendly 10 Outstanding I/O Using Oracle ORION I/O tool

SLIDE 6

Motivating Scenario

2.5 5.0 7.5 Shared Cache Disk Cache & Disk Small Large

6

Normalized Latency

Benefits cache- friendly workload Avoids disk interference Has best performance

SLIDE 7

Contributions

Build performance models dynamically
Account for interdependencies between resources
Lightweight but still accurate
Multi-level Resource Allocator
Uses performance models to guide resource allocation
Corrects model errors through runtime sampling
Uses global utility (SLOs) to partition resources
Minimize sum of application latencies

7

SLIDE 8

Approach

8

Build performance models
One per application
Derive function to predict application latency given configuration
Find resource partitioning setting
Minimize sum of application latencies
Find best setting using hill climbing

Lavg = f(ρc, ρs, ρd)

SLIDE 9

Outline

Online Performance Models
What are they?
Why are they hard to build?
Multi-level Resource Allocator
Prototype Implementation
Experimental Results
Conclusions

9

SLIDE 10

One-Level Cache Model

10

Rd(A) Cache size

Avg. Latency

Allocate in 32MB chunks

32 64 512

m=CacheSize/ChunkSize =1GB/32MB=32 choices

1G 32 choices Choose 512MB ... ...

SLIDE 11

MRC Cache Model

11

Rd(A) Cache size Miss-Ratio

32 64 512 1G ... ...

Computes miss-ratio given an I/O trace Multiply by I/O latency gets

Avg. Latency

SLIDE 12

Two-Level Cache Model

Performance affected by
DB Buffer Pool Size (m choices)
Storage Cache (n choices)
Performance model
Needs to consider all parameters (m*n choices)
1GB caches allocated in 32MB chunks
m = 1GB/32MB = 32 settings
m*n = 1024 distinct settings

12

Changes the I/O trace at storage

SLIDE 13

Two-Level Cache Model

13

256 512 768 256 512 768 1024 Buffer Pool Size (MB) Storage Cache Size (MB)

2 caches create a 3D surface

Buffer Pool Size Storage Cache Size Avg. Latency

32x32=1024 data points! 32 data points

High Latency Low Latency

SLIDE 14

Overall Performance Model

14

Application

ρc ρs ρd

Buffer Pool Storage Cache Disk Bandwidth Needs 32x32x10=10240 samples 15 mins/sample takes 3 months!

SLIDE 15

Outline

Online Performance Models
Multi-level Resource Allocator
Building performance models
Allocating resources using models
Prototype Implementation
Experimental Results
Conclusions

15

SLIDE 16

Key Observations

Known cache replacement policies
Most cache replacement algorithms are LRU
Only as effective as the largest cache (cache inclusiveness)
Disk is a closed loop system
Rate of responses is same as rate of requests
Performance proportional to the disk bandwidth fraction

16

SLIDE 17

Cache Inclusiveness

17

LRU LRU

8 8 1 6 8 4 5 6 3 4 8 8

I/Os: 0 I/Os: 1

8

SLIDE 18

Cache Inclusiveness

18

LRU LRU

8 8 1 6 8 4 5 6 3 4 8 8 1 6

I/Os: 6

8 4 5 6 3 4 3 4 3 4 5 6 8

Storage cache includes data in the buffer pool

SLIDE 19

Cache Inclusiveness

19

LRU LRU

8 8 1 6 8 4 5 6 3 4 8 8 1 6

I/Os: 6

8 4 5 6 3 4 3 5 3 4 5 6 8

Buffer pool includes data in the storage cache

SLIDE 20

Approximate Single Cache Model (LRU)

20

8 8 1 6 8 4 5 6 3 4 8 8 1 6 8 4 5 6 3 4

I/Os: 6

Same Number of I/Os

LRU LRU

3 4 3 4 5 6 8

Mc(max[ρc, ρs])

3 4 5 6 8

LRU

SLIDE 21

Cache Model (DEMOTE)

21

Maintain cache exclusiveness
E.g., using DEMOTEs [USENIX’02]
Every block brought into buffer pool is not cached below
Only evictions from buffer pool cached in storage cache
Approximate performance using single cache
Mc(ρc + ρs)

SLIDE 22

Find Best Partitioning Setting

22

Find Best Resource Allocation Setting

Buffer Pool Size

Storage Cache Size

Latency

App-1 App-2

Buffer Pool Size

Storage Cache Size

Latency

Minimize sum of application latencies

SLIDE 23

Observation: Closed loop system
Rate of responses same as rate of requests
Use interactive response time law
Performance proportional to disk bandwidth fraction
Measure base disk latency:
Predict latency for smaller bandwidth fractions

Disk Model

23

Ld(ρd) = Ld(1) ρd Ld(1)

SLIDE 24

Putting it All Together

24

Application

Mc(ρc)Ms(ρc, ρs)N

Storage Cache Buffer Pool

Approximate Single-Level Cache

Can now be solved using MRC

= Mc(max[ρc, ρs])N

SLIDE 25

Putting it all Together

Application

Hc(ρc)Lc Mc(ρc)Hs(ρc, ρs)Lnet Mc(ρc)Ms(ρc, ρs)Ld(ρd)

25

SLIDE 26

Inaccuracies in the Model

Cache Model
Approximations to LRU, i.e., CLOCK
Large fraction of writes in the workload
Disk Model
Using Quanta-based scheduler [Wachs et. al, FAST’07]
Interference due to disk seeks at small quanta
Inaccuracies localized in known regions
E.g., Small disk quanta

26

SLIDE 27

Iterative Refinement

Build model
Use trace collected at the database buffer pool
Refine the model
Use cross-validation to measure quality
Selectively sample where error is high
Interpolate computed and measured samples
Using regression (SVM)

27

SLIDE 28

Virtual Storage Prototype

28

Storage MySQL Linux

NBD

CLIENT

Block Layer SCSI Buffer Pool Disk

SERVER

NBD

Linux

Block Layer SCSI Disk Network Disk Disk Cache Quanta

SLIDE 29

Experimental Setup

Benchmarks
UNIFORM (microbenchmark), TPC-W and TPC-C
LAMP Architecture
Linux, Apache 1.3, MySQL/InnoDB 5.0, and PHP 5.0
Cache Configuration
MySQL buffer pool = 1GB
Storage cache = 1GB
Using InnoDB cache replacement in MySQL, CLOCK in storage

cache

29

SLIDE 30

Our Algorithms

30

GLOBAL
Gather trace at the buffer pool
Measure base disk latency
Compute performance using performance model
GLOBAL+
Run GLOBAL
Evaluate model accuracy
Refine model using runtime samples

SLIDE 31

Algorithms for Comparison

31

MRC
Partition cache (independently) using miss-ratio curves
DISK
Partition caches equally, determine best disk quanta
MRC+DISK
Run MRC then DISK
IDEAL*
Build model with SVM using 16*16*5=1280 sampled configurations

SLIDE 32

Roadmap of Results

32

Multi-level cache allocator
Using LRU and DEMOTE cache replacement policies
Multi-level cache and disk
Accuracy of computed models

SLIDE 33

Miss-Ratio Curves

33

25 50 75 100 128 256 384 512 640 768 896 1024 Miss Ratio (%) Buffer Pool Size (MB) TPC-W TPC-C UNIFORM

SLIDE 34

Multi-Level Caching (LRU)

34

Optimal HeatMap Lighter: Better Darker: Worse Optimal

Storage Cache Size (A) Buffer Pool Size (A) 1G 1G

2 TPC-W Instances

SLIDE 35

Multi-Level Caching (DEMOTE)

35

Optimal

Storage Cache Size (A) Buffer Pool Size (A) 1G 1G

2 TPC-W Instances

SLIDE 36

Roadmap of Results

36

Multi-level cache allocator
Multi-level cache and disk
Using two identical applications
Using different applications
Accuracy of computed models

SLIDE 37

UNIFORM/UNIFORM

1.682 3.364 5.046 6.728 8.410

GLOBAL GLOBAL+ MRC DISK MRC+DISK IDEAL*

UNIFORM UNIFORM

37

Average Latency (ms)

Allocate caches to 50/50 Matches GLOBAL

SLIDE 38

TPC-W/UNIFORM

1.5 3.0 4.5 6.0 7.5

GLOBAL GLOBAL+ MRC DISK MRC+DISK IDEAL*

TPC-W UNIFORM

38

Average Latency (ms)

Allocate caches to 50/50 Not enough buffer pool to UNIFORM Compensate for MRC settings

SLIDE 39

TPC-W/TPC-C

0.16 0.32 0.48 0.64 0.80

GLOBAL GLOBAL+ MRC DISK MRC+DISK IDEAL*

TPC-W TPC-C

39

Average Latency (ms)

Corrects model at runtime Corrects imbalance in MRC TPC-C allocated more in both

SLIDE 40

Roadmap of Results

40

Multi-level cache allocator
Multi-level cache and disk
Accuracy of computed models
Cache model
Disk model

SLIDE 41

Cache Model Accuracy (TPC-W)

41

128 256 384 512 640 768 896 1024 0 128 256 384 512 640 768 896 1024 Storage Cache Size (MB) Buffer Pool Size (MB) 10 20 30 40 50 Error (%)

Localized in the middle

SLIDE 42

Disk Model Accuracy (TPC-W)

42

10 20 30 40 50 0.2 0.4 0.6 0.8 1 Latency (ms) Disk Quota Measured Computed

SLIDE 43

Conclusions

Problem
Need to consider resources on multiple tiers
Independent cache/disk allocators are not sufficient
Dynamic allocation of cache hierarchy and disk
Build performance models dynamically
Iteratively refine (if necessary)
Use models for global resource partitioning
Performance up to 2.9 better than single resource allocators

43

SLIDE 44

Thank you.

44