SlideShare a Scribd company logo

Stuff I Learned About Performance

M
Michael Barker

The document discusses several topics related to software performance including: - The importance of understanding system behavior in detail rather than oversimplifying - How design must consider trade-offs - Techniques for improving performance such as using event sourcing architectures and leveraging vector instructions The key lessons are to understand problems fully before proposing solutions, design systems that allow flexibility in trade-offs, and measure performance carefully considering all relevant details.

Software
1 of 56
Downloaded 24 times
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
learned Stuff I’ve learnt about software performance @mikeb2701
• What is a financial exchange? • How to lie with numbers • Design is (almost) everything • Everything else is about making trade-offs
Stuff I Learned About Performance
TL;DR 1. Low Latency 2. ??? 3. Profit
Lies, Damed Lies and Performance Numbers
320μs @ 10,000 tx/s
Customer Fix Gateway Load Balancer Execution Venue Record Latency Parse/Encode FIX Order Matching Replication Journalling
320μs @ 10,000 tx/s
Stuff I Learned About Performance
320μs @ 10,000 tx/s
http://www.infoq.com/presentations/latency-pitfalls
What did I learn • Detail Matters • Don’t oversimplify the behaviour of your system
If we can really understand the problem, the answer will come out of it, because the answer is not separate from the problem. ! - Jiddu Krishnamurti
Customer Fix Gateway Load Balancer Execution Venue Record Latency Parse/Encode FIX Order Matching Replication Journalling
Instruction Execution Report Execution Venue OrderBook Order Execution
Execution Venue CreateOrderBook <Admin Instruction> instrument: Carrots
Execution Venue OrderBook <Entity> instrument: Carrots CreateOrderBook <Admin Instruction> instrument: Carrots
PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots
PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0
PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 Accept <ExecutionReport> id : ABC LiquidityAdded <ExecutionReport> instrument: Carrots price : 110 quantity : 5
Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 CancelOrder <Instruction> originalId: ABC
Execution Venue OrderBook <Entity> instrument: Carrots CancelOrder <Instruction> originalId: ABC Cancelled <ExecutionReport> id : ABC instrument: Carrots price : 110 quantity : 5 Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 5
Execution Venue OrderBook <Entity> instrument: Carrots CancelOrder <Instruction> originalId: ABC Cancelled <ExecutionReport> id : ABC instrument: Carrots price : 110 quantity : 5
Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK
Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK Accept <ExecutionReport> id : DEF Trade <ExecutionReport> price : 110 quantity : 5 aggressive: DEF passive : ABC Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 5 cancelled : 0
Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK Accept <ExecutionReport> id : DEF Trade <ExecutionReport> price : 110 quantity : 5 aggressive: DEF passive : ABC
Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : ABC instrument: Bananas price : 110 quantity : 5 side : BID
Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : ABC instrument: Bananas price : 110 quantity : 5 side : BID Reject <ExecutionReport> id : ABC
Interesting Properties • Stable working set • Fits in memory • Doesn’t not grow monotonically with time • All decisions are made from local information
Event-Sourced Service Journalling Replication Business Logic Message Message http://www.infoq.com/presentations/Event-Sourced-Architectures-for-High-Availability
What did I learn • Enough understanding of the problem and solutions will appear • Pay attention to: • Entity Life-cycles • Number of entities - in real world scenarios • Try to approach the problem without bias
If you can’t talk about the downside and the trade offs then you’re just a fan person waiving a coffee mug around. That type of behaviour really doesn’t benefit anyone. ! - Some guy on the Internet (charmalloc) Source: http://allthingshadoop.com/2013/12/06/technology-decisions-are-about-trade-offs-and-solving-problems/
Customer Fix Gateway Load Balancer 10K msg/sec Execution Venue Market Data Service 160K msg/sec
Design for trade-offs
TIMESTAMP,BID_PRICE_1,BID_QTY_1,ASK_PRICE_1,ASK_QTY_1 1405890300827,1.35263,50,1.35296,50 1405890300908,1.35263,50,1.35296,49.9 1405890300909,1.35263,50,1.35296,47.9 1405890301090,1.35263,48.2,1.35296,47.9 1405890301193,1.35263,48.2,1.35296,46.1 1405890301194,1.35263,48.2,1.35296,44.6 1405890301280,1.35263,48.2,1.35296,42.6 1405890301571,1.35263,48.1,1.35296,42.6 1405890303037,1.35263,47.1,1.35296,42.6
List<String[]> lines = ! parseAsList("data.csv");! for (String[] line : lines)! {! double bid = parseDouble(line[1]);! double ask = parseDouble(line[3]);! ! // Stuff...! }!
Iterable<String[]> lines = ! parseAsIterable("data.csv");! for (String[] line : lines)! {! double bid = parseDouble(line[1]);! double ask = parseDouble(line[3]);! ! // Stuff...! }!
public interface CsvObservable! {! default void onError(Exception e)! {! }! ! default void onComplete()! {! }! ! void onEvent(! int row, int column, ! CharSequence value, boolean endOfLine);! }
parse("data.csv", ! (row, column, value, endOfLine) ->! {! switch (column)! {! case 1:! bid = parseDouble(value);! case 3:! ask = parseDouble(value);! }! ! // Stuff...! });
Ops/sec GC Count GC Time (ms) Iterable 8.20 104 154 Observable 16.75 0 0
public class CsvObservableDelegate ! implements CsvObservable<CharSequence>! {! private CsvObservable<String> delegate;! ! public CsvObservableDelegate(CsvObservable<String> delegate)! {! this.delegate = delegate;! }! ! public void onEvent(! int row, int column, ! CharSequence value, boolean endOfLine)! {! delegate.onEvent(! row, column, value.toString(), endOfLine);! }! ! // Other delegate methods! }!
Garbage Collection Stallin’ your app since 1959
Stuff I Learned About Performance
Stuff I Learned About Performance
mean 95% 99% 99.99% JDK6 1.386 1.82 5.75 15.72 JDK7 1.44 1.79 11.75 20.66
Solutions •Don’t Generate Any • Use Zing (Azul Systems)
Why Use Java? (why not C/C++/..)
Source: https://software.intel.com/en-us/articles/introduction-to-intel-advanced-vector-extensions
public class Doubles ! { ! public static void sum(! double[] sum, ! double[] addendA, ! double[] addendB)! {! for (int i = 0; i < 0 + sum.length; i++)! {! sum[i] = addendA[i] + addendB[i];! }! }! }!
#include <immintrin.h>! ! void sum(double* c, double* a, double* b, int len)! {! __m256d rA_AVX, rB_AVX, rC_AVX;! ! for (int i = 0; i < len; i += 4)! {! rA_AVX = _mm256_load_pd(&a[i]);! rB_AVX = _mm256_load_pd(&b[i]);! rC_AVX = _mm256_add_pd(rA_AVX, rB_AVX);! _mm256_store_pd(&c[i], rC_AVX);! }! }!
Stuff I Learned About Performance
Stuff I Learned About Performance
More Java vs. C++ https://github.com/real-logic/simple-binary-encoding
What did I learn? • Details matter (performance measurement) • Let the problem define the solution • Design without bias • Understand the costs/benefits of trade-offs • Design to let others make trade-offs • Some trade-offs are not technical
Q&A
Links • Measuring Latency - http://goo.gl/6bllUP • Event Sourcing - http://goo.gl/VcZI7l • LMAX Architecture - http://goo.gl/UI5z0T • Source Code • http://github.com/mikeb01/qconsf2014

More Related Content

PPTX
SLEM - Status and Location of Equipment and Material
Erennio Iannotta
 
PDF
Test First
sivalsk
 
KEY
Deploying Next Gen Systems with Zero Downtime
Twilio Inc
 
PDF
Disruptor yow2013 v2
Michael Barker
 
PDF
Transactional Streaming: If you can compute it, you can probably stream it.
jhugg
 
PDF
Do more with less code in a serverless world
jeromevdl
 
ODP
Beyond php - it's not (just) about the code
Wim Godden
 
PDF
Practical unit testing 2014
Andrew Fray
 
SLEM - Status and Location of Equipment and Material
Erennio Iannotta
 
Test First
sivalsk
 
Deploying Next Gen Systems with Zero Downtime
Twilio Inc
 
Disruptor yow2013 v2
Michael Barker
 
Transactional Streaming: If you can compute it, you can probably stream it.
jhugg
 
Do more with less code in a serverless world
jeromevdl
 
Beyond php - it's not (just) about the code
Wim Godden
 
Practical unit testing 2014
Andrew Fray
 

Similar to Stuff I Learned About Performance (20)

PPT
Mistake and proofing Poka yokae slide presentation
RmzExe
 
ODP
Beyond php - it's not (just) about the code
Wim Godden
 
PDF
Brains & Brawn: the Logic and Implementation of a Redesigned Advertising Mark...
PyData
 
KEY
There and Back Again
Joshua Ballanco
 
PDF
Practical Machine Learning
David Jones
 
PDF
Moving away from legacy code with BDD
Konstantin Kudryashov
 
PDF
High Performance Managed Languages
J On The Beach
 
PDF
Why Agile Works But Isn't Working For You
David Harvey
 
PPTX
Natural born conversion killers - Conversion Jam
Craig Sullivan
 
PDF
Serverless in production, an experience report (London DevOps)
Yan Cui
 
PPTX
Mutation Testing
10Pines
 
PPTX
I don’t understand it fido
Erle Howard
 
PDF
Voicecon - Mashups with Tropo.com
Voxeo Corp
 
ODP
Scalingprezo
fdschoeneman
 
PDF
Building the "right" regression suite using Behavior Driven Testing (BDT)
Anand Bagmar
 
PDF
Contoh Desain Slide Presentasi Ilmiah Kreatif dan Menarik #1
Arry Rahmawan
 
PDF
Acceptance Test Driven Development at StarWest 2014
jaredrrichardson
 
PPTX
Reducing Bugs With Static Code Analysis php tek 2025
Scott Keck-Warren
 
PDF
Road to A/B testing - Alexey Vasiliev (ENG) | Ruby Meditation 25
Ruby Meditation
 
KEY
Effective Testing with Ruby
Akira Sosa
 
Mistake and proofing Poka yokae slide presentation
RmzExe
 
Beyond php - it's not (just) about the code
Wim Godden
 
Brains & Brawn: the Logic and Implementation of a Redesigned Advertising Mark...
PyData
 
There and Back Again
Joshua Ballanco
 
Practical Machine Learning
David Jones
 
Moving away from legacy code with BDD
Konstantin Kudryashov
 
High Performance Managed Languages
J On The Beach
 
Why Agile Works But Isn't Working For You
David Harvey
 
Natural born conversion killers - Conversion Jam
Craig Sullivan
 
Serverless in production, an experience report (London DevOps)
Yan Cui
 
Mutation Testing
10Pines
 
I don’t understand it fido
Erle Howard
 
Voicecon - Mashups with Tropo.com
Voxeo Corp
 
Scalingprezo
fdschoeneman
 
Building the "right" regression suite using Behavior Driven Testing (BDT)
Anand Bagmar
 
Contoh Desain Slide Presentasi Ilmiah Kreatif dan Menarik #1
Arry Rahmawan
 
Acceptance Test Driven Development at StarWest 2014
jaredrrichardson
 
Reducing Bugs With Static Code Analysis php tek 2025
Scott Keck-Warren
 
Road to A/B testing - Alexey Vasiliev (ENG) | Ruby Meditation 25
Ruby Meditation
 
Effective Testing with Ruby
Akira Sosa
 
Ad

More from Michael Barker (6)

PDF
Value Types
Michael Barker
 
PDF
Test automation 3
Michael Barker
 
PDF
Concurrecy techdrop
Michael Barker
 
KEY
Lock? We don't need no stinkin' locks!
Michael Barker
 
PDF
Beginners guide-concurrency
Michael Barker
 
PDF
Disruptor tools in action
Michael Barker
 
Value Types
Michael Barker
 
Test automation 3
Michael Barker
 
Concurrecy techdrop
Michael Barker
 
Lock? We don't need no stinkin' locks!
Michael Barker
 
Beginners guide-concurrency
Michael Barker
 
Disruptor tools in action
Michael Barker
 
Ad

Recently uploaded (20)

PDF
AI in Product Development-omnex systems
omnex systems
 
PDF
How to Choose the Right IT Partner for Your Business in Malaysia
Asian Business Services & Technologies
 
PDF
2025 Textile ERP Trends: SAP, Odoo & Oracle
Aetos Technologies
 
PDF
Raksha Bandhan Grocery Pricing Trends in India 2025.pdf
Actowiz Solustions
 
PDF
Odoo Companies in India – Driving Business Transformation.pdf
azhuhardeen1968
 
PDF
Internet Downloader Manager (IDM) Crack 6.42 Build 41
ghrom2211g
 
PPTX
Agentic AI Use Case- Contract Lifecycle Management (CLM).pptx
mhxyzzp
 
PDF
How Creative Agencies Leverage Project Management Software.pdf
Orangescrum
 
PDF
Addressing The Cult of Project Management Tools-Why Disconnected Work is Hold...
OnePlan Solutions
 
PPTX
Operating system designcfffgfgggggggvggggggggg
rmskumara09
 
PPTX
Introduction to Artificial Intelligence
lohedi9363
 
PDF
Softaken Excel to vCard Converter Software.pdf
markwillsonmw004
 
PDF
Audit Checklist Design Aligning with ISO, IATF, and Industry Standards — Omne...
rohnjim07
 
PDF
Flood Susceptibility Mapping Using Image-Based 2D-CNN Deep Learnin. Overview ...
lawrenceomai2
 
PPTX
history of c programming in notes for students .pptx
Vijayakumari829030
 
PPTX
Odoo POS Development Services by CandidRoot Solutions
CandidRoot Solutions Private Limited
 
PDF
T3DD25 TYPO3 Content Blocks - Deep Dive by André Kraus
Andre Kraus
 
PPTX
ai tools demonstartion for schools and inter college
harshavardhanreddyka11
 
PPTX
Transform Your Business with a Software ERP System
Canada Software Company
 
PDF
Internet Downloader Manager (IDM) Crack 6.42 Build 42 Updates Latest 2025
itskinga12
 
AI in Product Development-omnex systems
omnex systems
 
How to Choose the Right IT Partner for Your Business in Malaysia
Asian Business Services & Technologies
 
2025 Textile ERP Trends: SAP, Odoo & Oracle
Aetos Technologies
 
Raksha Bandhan Grocery Pricing Trends in India 2025.pdf
Actowiz Solustions
 
Odoo Companies in India – Driving Business Transformation.pdf
azhuhardeen1968
 
Internet Downloader Manager (IDM) Crack 6.42 Build 41
ghrom2211g
 
Agentic AI Use Case- Contract Lifecycle Management (CLM).pptx
mhxyzzp
 
How Creative Agencies Leverage Project Management Software.pdf
Orangescrum
 
Addressing The Cult of Project Management Tools-Why Disconnected Work is Hold...
OnePlan Solutions
 
Operating system designcfffgfgggggggvggggggggg
rmskumara09
 
Introduction to Artificial Intelligence
lohedi9363
 
Softaken Excel to vCard Converter Software.pdf
markwillsonmw004
 
Audit Checklist Design Aligning with ISO, IATF, and Industry Standards — Omne...
rohnjim07
 
Flood Susceptibility Mapping Using Image-Based 2D-CNN Deep Learnin. Overview ...
lawrenceomai2
 
history of c programming in notes for students .pptx
Vijayakumari829030
 
Odoo POS Development Services by CandidRoot Solutions
CandidRoot Solutions Private Limited
 
T3DD25 TYPO3 Content Blocks - Deep Dive by André Kraus
Andre Kraus
 
ai tools demonstartion for schools and inter college
harshavardhanreddyka11
 
Transform Your Business with a Software ERP System
Canada Software Company
 
Internet Downloader Manager (IDM) Crack 6.42 Build 42 Updates Latest 2025
itskinga12
 

Stuff I Learned About Performance

  • 1. learned Stuff I’ve learnt about software performance @mikeb2701
  • 2. • What is a financial exchange? • How to lie with numbers • Design is (almost) everything • Everything else is about making trade-offs
  • 4. TL;DR 1. Low Latency 2. ??? 3. Profit
  • 5. Lies, Damed Lies and Performance Numbers
  • 7. Customer Fix Gateway Load Balancer Execution Venue Record Latency Parse/Encode FIX Order Matching Replication Journalling
  • 12. What did I learn • Detail Matters • Don’t oversimplify the behaviour of your system
  • 13. If we can really understand the problem, the answer will come out of it, because the answer is not separate from the problem. ! - Jiddu Krishnamurti
  • 14. Customer Fix Gateway Load Balancer Execution Venue Record Latency Parse/Encode FIX Order Matching Replication Journalling
  • 15. Instruction Execution Report Execution Venue OrderBook Order Execution
  • 16. Execution Venue CreateOrderBook <Admin Instruction> instrument: Carrots
  • 17. Execution Venue OrderBook <Entity> instrument: Carrots CreateOrderBook <Admin Instruction> instrument: Carrots
  • 18. PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots
  • 19. PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0
  • 20. PlaceOrder <Instruction> id : ABC instrument: Carrots price : 110 quantity : 5 side : BID Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 Accept <ExecutionReport> id : ABC LiquidityAdded <ExecutionReport> instrument: Carrots price : 110 quantity : 5
  • 21. Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 CancelOrder <Instruction> originalId: ABC
  • 22. Execution Venue OrderBook <Entity> instrument: Carrots CancelOrder <Instruction> originalId: ABC Cancelled <ExecutionReport> id : ABC instrument: Carrots price : 110 quantity : 5 Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 5
  • 23. Execution Venue OrderBook <Entity> instrument: Carrots CancelOrder <Instruction> originalId: ABC Cancelled <ExecutionReport> id : ABC instrument: Carrots price : 110 quantity : 5
  • 24. Execution Venue OrderBook <Entity> instrument: Carrots Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 0 cancelled : 0 PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK
  • 25. Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK Accept <ExecutionReport> id : DEF Trade <ExecutionReport> price : 110 quantity : 5 aggressive: DEF passive : ABC Order <Entity> orderId : 1 id : ABC price : 110 quantity : 5 side : BID filled : 5 cancelled : 0
  • 26. Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : DEF instrument: Carrots price : 110 quantity : 5 side : ASK Accept <ExecutionReport> id : DEF Trade <ExecutionReport> price : 110 quantity : 5 aggressive: DEF passive : ABC
  • 27. Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : ABC instrument: Bananas price : 110 quantity : 5 side : BID
  • 28. Execution Venue OrderBook <Entity> instrument: Carrots PlaceOrder <Instruction> id : ABC instrument: Bananas price : 110 quantity : 5 side : BID Reject <ExecutionReport> id : ABC
  • 29. Interesting Properties • Stable working set • Fits in memory • Doesn’t not grow monotonically with time • All decisions are made from local information
  • 30. Event-Sourced Service Journalling Replication Business Logic Message Message http://www.infoq.com/presentations/Event-Sourced-Architectures-for-High-Availability
  • 31. What did I learn • Enough understanding of the problem and solutions will appear • Pay attention to: • Entity Life-cycles • Number of entities - in real world scenarios • Try to approach the problem without bias
  • 32. If you can’t talk about the downside and the trade offs then you’re just a fan person waiving a coffee mug around. That type of behaviour really doesn’t benefit anyone. ! - Some guy on the Internet (charmalloc) Source: http://allthingshadoop.com/2013/12/06/technology-decisions-are-about-trade-offs-and-solving-problems/
  • 33. Customer Fix Gateway Load Balancer 10K msg/sec Execution Venue Market Data Service 160K msg/sec
  • 35. TIMESTAMP,BID_PRICE_1,BID_QTY_1,ASK_PRICE_1,ASK_QTY_1 1405890300827,1.35263,50,1.35296,50 1405890300908,1.35263,50,1.35296,49.9 1405890300909,1.35263,50,1.35296,47.9 1405890301090,1.35263,48.2,1.35296,47.9 1405890301193,1.35263,48.2,1.35296,46.1 1405890301194,1.35263,48.2,1.35296,44.6 1405890301280,1.35263,48.2,1.35296,42.6 1405890301571,1.35263,48.1,1.35296,42.6 1405890303037,1.35263,47.1,1.35296,42.6
  • 36. List<String[]> lines = ! parseAsList("data.csv");! for (String[] line : lines)! {! double bid = parseDouble(line[1]);! double ask = parseDouble(line[3]);! ! // Stuff...! }!
  • 37. Iterable<String[]> lines = ! parseAsIterable("data.csv");! for (String[] line : lines)! {! double bid = parseDouble(line[1]);! double ask = parseDouble(line[3]);! ! // Stuff...! }!
  • 38. public interface CsvObservable! {! default void onError(Exception e)! {! }! ! default void onComplete()! {! }! ! void onEvent(! int row, int column, ! CharSequence value, boolean endOfLine);! }
  • 39. parse("data.csv", ! (row, column, value, endOfLine) ->! {! switch (column)! {! case 1:! bid = parseDouble(value);! case 3:! ask = parseDouble(value);! }! ! // Stuff...! });
  • 40. Ops/sec GC Count GC Time (ms) Iterable 8.20 104 154 Observable 16.75 0 0
  • 41. public class CsvObservableDelegate ! implements CsvObservable<CharSequence>! {! private CsvObservable<String> delegate;! ! public CsvObservableDelegate(CsvObservable<String> delegate)! {! this.delegate = delegate;! }! ! public void onEvent(! int row, int column, ! CharSequence value, boolean endOfLine)! {! delegate.onEvent(! row, column, value.toString(), endOfLine);! }! ! // Other delegate methods! }!
  • 42. Garbage Collection Stallin’ your app since 1959
  • 45. mean 95% 99% 99.99% JDK6 1.386 1.82 5.75 15.72 JDK7 1.44 1.79 11.75 20.66
  • 46. Solutions •Don’t Generate Any • Use Zing (Azul Systems)
  • 47. Why Use Java? (why not C/C++/..)
  • 49. public class Doubles ! { ! public static void sum(! double[] sum, ! double[] addendA, ! double[] addendB)! {! for (int i = 0; i < 0 + sum.length; i++)! {! sum[i] = addendA[i] + addendB[i];! }! }! }!
  • 50. #include <immintrin.h>! ! void sum(double* c, double* a, double* b, int len)! {! __m256d rA_AVX, rB_AVX, rC_AVX;! ! for (int i = 0; i < len; i += 4)! {! rA_AVX = _mm256_load_pd(&a[i]);! rB_AVX = _mm256_load_pd(&b[i]);! rC_AVX = _mm256_add_pd(rA_AVX, rB_AVX);! _mm256_store_pd(&c[i], rC_AVX);! }! }!
  • 53. More Java vs. C++ https://github.com/real-logic/simple-binary-encoding
  • 54. What did I learn? • Details matter (performance measurement) • Let the problem define the solution • Design without bias • Understand the costs/benefits of trade-offs • Design to let others make trade-offs • Some trade-offs are not technical
  • 55. Q&A
  • 56. Links • Measuring Latency - http://goo.gl/6bllUP • Event Sourcing - http://goo.gl/VcZI7l • LMAX Architecture - http://goo.gl/UI5z0T • Source Code • http://github.com/mikeb01/qconsf2014