SLIDE 1
Unevenly
Adrian Colyer
@adriancolyer
blog.acolyer.org
350
Foundations Frontiers
Unevenly Distributed Adrian Colyer @adriancolyer blog.acolyer.org - - PowerPoint PPT Presentation
Unevenly Distributed Adrian Colyer @adriancolyer blog.acolyer.org - - PowerPoint PPT Presentation
Unevenly Distributed Adrian Colyer @adriancolyer blog.acolyer.org 350 Foundations Frontiers 5 Reasons to <3 Papers 03 Applied rainstorm 04 Lessons The Great 02 Conversation Raise Expectations 05 01 Uneven Thinking Brain
SLIDE 2
SLIDE 3
Brain storm
01 02 05 04
rainstorm
03
5 Reasons to <3 Papers
Thinking tools Raise Expectations Applied Lessons The Great Conversation Uneven Distribution
3
SLIDE 4
Frank McSherry
Scalability - but at what COST?
4
SLIDE 5
5
SLIDE 6
But you have BIG Data!
6
Zipf Distribution “Working sets are Zipf-
- distributed. We can
therefore store in memory all but the very largest datasets.”
SLIDE 7
Musketeer
7
One for all?
SLIDE 8
Approx Hadoop
8
32x!
SLIDE 9
Improve your API Design
The Scalable Commutativity Rule
9
SLIDE 10
Raising Your Expectations
10
SLIDE 11
TLS
11
54
CVEs
Jan ‘14 - Jan ‘15
! Error prone languages ! Lack of Separation ! Ambiguous and Untestable Spec
Surely we can do better?
SLIDE 12
Do Less Testing!
12
Relative Improvement Cost Improvement Test Executions 40.58% Test Time 40.31% $1,567,608 Test Result Inspection 33.04% $61,533 Escaped Defects 0.20% ($11,971) Total Cost Balance $1,617,170 Microsoft Windows 8.1
SLIDE 13
13
SLIDE 14
Lessons from the Field
14
SLIDE 15
at Facebook
A Masterclass in Config Mgt
15
SLIDE 16
lessons from Google
Machine Learning Systems
16
Feature Management Visualisation Relative Metrics Systematic Bias Correction Alerts on action Thresholds 01 02 03 04 05
SLIDE 17
And the Syntopicon
The Great Conversation
17
SLIDE 18
Robotics Security Distributed Systems Databases Machine Learning Programming Languages
Broad Exposure to Problems and their Solutions
Cross-Fertilization
And Many More
Operating Systems, Algorithms, Networking,Optimisation, SW Engineering,... 18
SLIDE 19
TPC-C - 1992
19
SLIDE 20
TPC-C Published Record Holder
20
Mar 26th 2013
Date
Oracle 11g r2 Enterprise Edition w. Partitioning
Database Manager
8,552,523 (8.5M)
Performance (tpmC)
142,542 (143K)
Performance (tps)
$4,663,073
System Cost
8
#Processors
128
#Cores
1024
#Threads
SLIDE 21
and I-Confluence Analysis
Coordination Avoidance
21
TPC-C
SLIDE 22
Multi-Partition Transactions at Scale
22
SLIDE 23
Turning your world Upside Down
Unevenly Distributed
SLIDE 24
Human computers
at Dryden by NACA (NASA) -
Dryden Flight Research Center Photo Collection
http://www.dfrc.nasa. gov/Gallery/Photo/Places/HTML/E49-54.html. Licensed under Public Domain via Commons - https://commons.wikimedia.org/wiki/File: Human_computers_-_Dryden.jpg#/media/File: Human_computers_-_Dryden.jpg
SLIDE 25
Computing on a Human Scale
25
10ns 70ns 10ms 10s 1:10s 116d
Registers & L1-L3 File on desk Main memory Office filing cabinet HDD Trip to the warehouse
SLIDE 26
Compute
HTM Persistent Memory NI FPGA GPUs
Memory
NVDIMMs Persistent Memory
Networking
100GbE RDMA
Storage
NVMe Next-gen NVM
Next Generation Hardware
All Change Please
26
SLIDE 27
2-10m
Computing on a Human Scale
27
10s 1:10s 116d
File on desk Office filing cabinet Trip to the warehouse
4x capacity fireproof local filing cabinets
23-40m
Phone another office (RDMA)
3h20m
Next-gen warehouse
SLIDE 28
The New ~Numbers Everyone Should Know
28
Latency Bandwidth Capacity/IOPS Register 0.25ns L1 cache 1ns L2 cache 3ns 8MB L3 cache 11ns 45MB DRAM 62ns 120GBs 6TB - 4 socket NVRAM’ DIMM 620ns 60GBs 24TB - 4 socket 1-sided RDMA in Data Center 1.4us 100GbE ~700K IOPS RPC in Data Center 2.4us 100GbE ~400K IOPS NVRAM’ NVMe 12us 6GBs 16TB/disk,~2M/600K NVRAM’ NVMf 90us 5GBs 16TB/disk, ~700/600K
SLIDE 29
Low Latency - RAMCloud
29
Reads
5μs
Writes
13.5μs
Transactions
20μs
5-object Txns
27μs
TPC-C (10 nodes)
35K tps
SLIDE 30
No Compromises - FaRM
30
TPC-C (90 nodes)
4.5M tps
99%ile
1.9ms
KV (per node)
6.3M qps
at peak throughput
41μs
SLIDE 31
No Compromises
31
“This paper demonstrates that new software in modern data centers can eliminate the need to compromise. It describes the transaction, replication, and recovery protocols in FaRM, a main memory distributed computing
- platform. FaRM provides distributed ACID transactions
with strict serializability, high availability, high throughput and low latency. These protocols were designed from first principles to leverage two hardware trends appearing in data centers: fast commodity networks with RDMA and an inexpensive approach to providing non-volatile DRAM.”
SLIDE 32
DrTM
The Doctor will see you now
32
5.5M tps on TPC-C 6-node cluster.
SLIDE 33
Some things Change, Some stay the Same
33
SLIDE 34
A Brave New World
34
Fast RDMA networks + Ample Persistent Memory + Hardware Transactions + Enhanced HW Cache Management + Super-fast Storage + On-board FPGAs + GPUs + … = ???
SLIDE 35
Brain storm
01 02 05 04
rainstorm
03
5 Reasons to <3 Papers
Thinking tools Raise Expectations Applied Lessons The Great Conversation Uneven Distribution
35
SLIDE 36
A new paper every weekday
Published at http://blog.acolyer.org.
01
Delivered Straight to your inbox
If you prefer email-based subscription to read at your leisure.
02
Announced on Twitter
I’m @adriancolyer.
03
Go to a Papers We Love Meetup
A repository of academic computer science papers and a community who loves reading them.
04
Share what you learn
Anyone can take part in the great conversation.
05
SLIDE 37
THANK YOU !
@adriancolyer