Application Acceleration by Fusion-io CTO, Pankaj Mehra - Fusion-io Analyst Day 2014

Application Acceleration Pankaj Mehra February 6, 2014

Safe Harbor Disclosure

5

Note on Forward-looking Statements Certain statements in this presentation may constitute "forward-looking statements" within the meaning of Section 21E of the Securities Exchange Act of 1934 and Section 27A of the Securities Act of 1933, including, but are not limited to, statements concerning our future plans, expectations, beliefs, intentions and prospects, financial guidance for our third fiscal quarter of 2014, our progress on, and effect of, our key technology and strategic initiatives, our sales execution efforts, our expectations regarding the market opportunity and our ability to execute our go-to-market strategy, expectations concerning our product portfolio and our strategic partnerships, and benefits and value of our products and solutions to our customers and end users. These statements are based on current expectations and assumptions regarding future events and business performance and involve certain risks and uncertainties that could cause actual results to differ materially from those contained, anticipated, or implied in any forward-looking statement, including, but not limited to, risks associated with changes in the demand for our products, our expectation that large and concentrated purchases by a limited number of customers will continue to represent a substantial majority of our revenue and our ability to sustain or increase our revenue from our large customers or offset the discontinuation of concentrated purchases by our larger customers with purchases by new or existing customers, the continued adoption by customers of our ioMemory platform products, growing our sales through OEMs, resellers and channel partners and maintaining our relationships with OEMs, resellers and channel partners, including the timely qualification of our products for promotion and sale by our OEMs, long and unpredictable sales cycles, changes in the competitive dynamics of our markets, including the potential for increased pressure on the pricing of our products, reduced gross margins, increased sales and marketing expenses, the potential that we or our customers may not realize the benefits we currently expect from our acquisitions of ID7 and NexGen Storage, our ability to develop or acquire new products to meet customer needs and expectations, including additional software solutions to be integrated with our storage memory products, our acquisition and strategic partner strategy and disruptions in our business, operations and financial results as a result of acquisitions and strategic partner relationships, as well as the risks inherent in the integration and combination of complex products and technologies from acquisitions, undetected errors, defects or security vulnerabilities in our products, worldwide economic conditions and the impact these conditions have on levels of spending on datacenter technology like ours, our ability to recruit and successfully hire new executive officers, and such other risks set forth in the registration statements and reports that Fusion-io files with the U.S. Securities and Exchange Commission, which are available on the Investor Relations section of our website at www.fusionio.com. You should not rely upon forward-looking statements as predictions of future events. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee that the future results, levels of activity, performance or events and circumstances reflected in the forward-looking statements will be achieved or will occur. Fusion-io undertakes no obligation to update publicly any forward-looking statement for any reason after the date of this presentation.

Use of Non-GAAP Financial Information This presentation includes a discussion of “non-GAAP financial measures” as that term is defined in Regulation G promulgated by the U.S. Securities and Exchange Commission. The most directly comparable GAAP financial measures and information reconciling these non-GAAP financial measures to the Company’s financial results prepared in accordance with GAAP are provided on the company’s investor relations website at http://investor.fusionio.com under “Financial Information.”

Application Acceleration

6

▸ What is application acceleration?

▸ Why does it matter? ▸ What is the solution, and

differentiated value proposition?

▸  A confluence of businesses and their technologies striving to deliver user experiences: faster, simpler, smarter •  CPU & Graphics for raw horsepower •  Networking for connectivity •  Cloud, SaaS & Mobility for untethering

•  Flash technology accelerating data supply to

the CPU for analytics and offering fast persistence for inexpensive transactions

•  Context brokers exploiting favorable data:technology trends to gather and analyze big user data with scale-out iron

•  Ultimately, wearable computing and activity recognition, toward total context-awareness


7










8










9










10




Business value of

information

Time since ingest

Raw, new information

“right time”

Contextualized information

Net gain from contextualization

0

Across industries, businesses must process more new information deeper and faster than their competition.

▸  App SLAs >103-5 Infrastructure SLAs

From Fast Storage to Application Acceleration

11

SSDs

Mar

ket

Rel

evan

ce


Workload Optimized

Plug & Play Future Proof Investment

Application Acceleration and the Flash Memory Tier

12

Flash memory is solid state memory technology whose performance, cost, and endurance place it between DRAM and HDD. ▸ Directly benefits almost

every subarea of application acceleration

DRAM Host CPU

Flash Memory

HDD, SAN, NAS

Application SLAs Application Performance Management

Server Virtualization

Memory Management

Data Management

Storage Performance Management


13




In Server Application Acceleration

Application Acceleration Appliances

PC

Ie

DRAM

Host CPU

Flash As Disk Flash As Memory

ioMemory: Superior Architecture

PC

Ie

SA

S

DRAM

Data path Controller

NAND

Host CPU

RAID Controller

SC

14

NAND

App OS App

OS

14

15

1.  Performance • Primary apps (DB) • Log – Index – Checkpoint

2. SAN Relief • Virtualized servers/desktops • Random read IOPS • Write vectoring

First Steps Beyond Vanilla SSDs

16

Plug & Play

▶  Enterprise-grade PCIe SSDs

▶  5% of Data Tier

▶  5-10x performance! ▶  Strategic deployment ▶  Early adopters ▶  Databases


17

Workload Optimized

▶  Virtualization ▶  In-memory

Computing

▶  5-10x VM density! ▶  SAN relief ▶  Early majority

Plug & Play




1.  Performance • Primary apps (DB) • Log – Index – Checkpoint

2. SAN Relief • Virtualized servers/desktops • Random read IOPS • Write vectoring

18

4. Comprehensive Application Acceleration with a Flash Memory tier

3. Memory Relief • Fast swap & sparse map

at page grain

Persistent Memory

Modules Features

Standards-based access

Persistent Memory Foundation

19

Workloads

PM

F Stack

Transparent and API Interfaces

NVM File System

Logical volume management

Virtual Storage Services

Persistent Memory Manager

RDMA Layer

Flash Translation Layer

Device-independent virtual storage services • Snapshots (memory & block) • Clones (memory & block) • Atomic writes • TRIM-compression • Memory logging (ACM) • Memory checkpointing • Memory metadata

Accessed as Memory and Managed as Storage

Uncommon value under

common standards


20

Workload Optimized

Plug & Play



▶  Virtualization ▶  In-memory

Computing


▶  5-10x VM density! ▶  SAN relief ▶  Early majority

Future Proof Investment


▶  5-10x memory relief!

▶  Native flash interface ▶  Application extensions

Workload Optimized Application Acceleration

Types of Flash-Based Application Acceleration

▸  Out-of-the-box benefits of flash for today’s applications

▸  Representative methods •  App configuration, data placement •  Workload-aware persistent memory

based system software for standard application programming interfaces

▸  Benefits •  Existing database applications

▸  Applications directly program the persistence •  Avoid double-buffering using atomics •  Gain compression using TRIM •  Offload indexes from memory •  Speed up logs using auto commit

memory

▸  Benefits •  Re-architected or new applications

21

Flash-Aware Application Acceleration

Persistent Memory Benefits for MySQL

22

Latency, Throughput and Endurance

MariaDB: 70% Transactions/sec Improvement Sysbench 99% Latency OLTP Workload

Percona: 2-4x Latency Improvement

XFS Double Write Atomic Writes

200

Late

ncy

0 0 Seconds 3600

Upcoming: Better Compression for Databases

23

Trim-Enabled Compression: Making MySQL Compression Viable

Compression Performance Penalty

Transaction rate

Application Acceleration Reduces Server Sprawl

24

Use

ful A

pplic

atio

n W

ork

per S

erve

r

SSD (block) NVM File System

Application-Transparent Acceleration

NVM File System

Atomics API

Compression API

Index Optimization APIs (future)

Log Optimization API

Thank You

f u s i o n i o . c o m | D e l i v e r i n g t h e Wo r l d ’ s D a t a . F a s t e r.

Technology

Application Acceleration by Fusion-io CTO, Pankaj Mehra - Fusion-io Analyst Day 2014