VAST Data - IT Press Tour #44 June 2022

Transcript

1. IT PRESS TOUR JUNE 2022
2. MOMENTUM REPORT AS OF FEB 1, 2022 1: Source: https://www.gartner.com/reviews/market/distributed-file-systems-and-object-storage/ 2: Source: https://www.ark-bigideas.com/2022/en/pages/download POSITIVE Cash Flow $1.2M Q4-FY22 Average Selling Price 303% 100% Customers Surveyed Recommend1 Q4-FY22 Net Revenue Retention Demonstrating Confidence in Universal Storage from Existing Customers INFINITE RUNWAY $300M End of Year 3: Bookings Run Rate 3.8X Year over year growth October 31, 2021 January 31, 2022 OVER $200M ON OUR BALANCE SHEET AN UNPRECEDENTED SOFTWARE SUCCESS STORY
3. 800 EMPLOYEES 330 EMPLOYEES 600 EMPLOYEES MAKING HISTORY
4. 4 OUTPERFORMERS VASTRONAUTS HAVE THE STRENGTH OF 2.5 MEN
5. REALLY MAKING HISTORY INFLATION-ADJUSTED CUMULATIVE GROSS PROFIT * * estimated
6. Full-Year Gross Profit & Estimated RPO (Profit) Net Loss FY2022 ACTUAL Y/Y Gross Profit Growth Rate
7. MACHINE LEARNING ERA Multi-Core CPUs Petabytes of Data Single Cloud ANALYTICS ERA Single-Core CPUs Terabytes of Data Single Data Center DEEP LEARNING ERA 1000 Core GPUs Exabytes Of Data Multi-cloud To Many-edge 45TB AVERAGE $SNOW CUSTOMER CAPACITY 12PB AVERAGE VAST CUSTOMER CAPACITY (266x) DATA IS EVOLVING
8. COMPUTE DATA STORAGE PLATFORMS MUST EVOLVE WITH DATA ANALYTICS ERA MACHINE LEARNING ERA DEEP LEARNING ERA
9. WEB-SCALE COMMODITY DATA FABRIC 1,000s OF CONTAINERS • FULLY PARALLEL DATA SERVICES • COMPOSABLE EXABYTES OF HYPERSCALE LOW-COST FLASH THE DASE ARCHITECTURE CONFIDENTIAL
10. SW ENCRYPTION REPLICATION (<1M RPO) FIPS 140-2 EASY INSTALLATION VMS HIGH AVAILABILITY NFS VERSION 4.1 DATA FLOW MONITORING CAPACITY ESTIMATIONS UPLINK (CLOUD VMS) S3 VERSIONING SMB LINUX SUPPORT NFS KRB AUTHENTICATION USER QUOTAS DIRECTORY-LEVEL SNAPS GEMINI LICENSE MGMT S3 MANAGEMENT IN VMS NON-CONTIGUOUS POOLS SIMILARITY NFS 4.1 NCONNECT S3 MULTIPROTOCOL 30TB QLC SUPPORT 130% MORE FILES / DBOX S3 DIRECTORIES S3 RENAME BACKUP TO S3 AT-REST ENCRYPTION BUNDLE OBFUSCATION S3 HTTPS REMOTE SUPPORT TUNNEL VIEW POLICIES OUR LAST 15 MONTHS
11. THE VAST ADVANTAGE ULTIMATE SIMPLICITY "It has quickly become the most reliable storage product in our environment” – Gov Customer LINEAR SCALABILITY & RESILIENCE “The only system that gets better as it scales” – Leading Hyperscale Company RADICAL COST-EFFICIENCY “All-flash performance and resilience for the TCO of HDD… you're future proofing” – Ivy League U CONSOLIDATE & ENABLE INSIGHTS “Our pipeline went from 24 hours to 15 minutes” - Invitae PAST Complexity, Compromise & Constraints Unified Data, No Limits
12. EARLY VERTICAL FOCUS Light on regulation, strong on size, speed, and scale EXPANDING HORIZONTAL FOCUS More regulation that creates more need for data mgmt & security features FINANCIAL SERVICES LIFE SCIENCE HPC SITES & UNIVERSITIES WEB & SAAS COMPANIES MEDIA COMPANIES BIG DATA BACKUP DEEP LEARNING ELK Stack
13. $3M: FIRST USE CASE EXPANSION $12M NEW USE CASES $1M: LAND INITIAL USE CASE UNIVERSAL STORAGE AI AI BIG DATA BACKUP SPLUNK BIG DATA BACKUP SPLUNK $1M OF BUSINESS CAN EVOLVE INTO $16M IN 3 YEARS
14. GLOBAL COVERAGE 150 COUNTRIES SERVED 24 x 7 x 365 – Follow the sun 10 LANGUAGES Multilingual Support & Sales 145 PARTNERS Global Network Present With Open Locations Not Currently Present With Open Locations
15. FBI EXABYTES OF VAST DATA
16. “Choosing VAST was viewed as a risk at our organization. It is now considered a big win!” “Maintaining and managing the system is a breeze…. effortless.” “Storage performance on our compute cluster with parallel jobs is incredible.” “A magic combo of cost and performance that only improves with each new version.” “You Want Happy Users, Buy VAST Data.” 100% RECOMMENDED
17. “ …not only is it a vendor to watch in enterprise storage, it is the vendor to watch. GROWING DATA DEMANDS ARE DRIVING VAST DATA’S CONTINUED HYPERGROWTH, APRIL 2022
18. USE CASES
19. 1,000s OF APPLICATIONS ~20 APPLICATIONS FINANCIAL SERVICES LIFE SCIENCE HPC SITES & UNIVERSITIES WEB & SAAS COMPANIES MEDIA COMPANIES BIG DATA BACKUP DEEP LEARNING ELK Stack
20. THE GAME IS CHANGING DATA PROTECTION RAPID DATA RESTORATION HADOOP CLUSTERS FAST S3 DATA LAKES & ABSTRACTION MACHINE LEARNING DEEP LEARNING & INFERENCE @ SCALE
21. DATA PROTECTION HADOOP CLUSTERS MACHINE LEARNING SCALABLE ALL-FLASH BACKUP & RESTORE WITHOUT LEGACY FLASH TAX SCALABLE & RESILIENT FAST S3 WITH HDD ECONOMICS THE SIMPLEST PATH TO SCALABLE DL. FULL STOP. VAST CHANGES THE NARRATIVE RAPID DATA RESTORATION FAST S3 DATA LAKES & ABSTRACTION DEEP LEARNING & INFERENCE @ SCALE
22. HARD DRIVE ACCESSES CAN RESULT IN 50X SLOWER I/O FOR AI APPLICATIONS. STORAGE TIERING DOESN’T WORK FOR A.I.
23. EVERYTHING ELSE /scratch
24. BIG ARE WHAT WE DO CLUSTERS
25. THE ZERO-SUM SHARED-NOTHING FAILURE GAME SIZE OF NODE (TBs) # OF NODES Probability Of Node Failure Increases Time-To-Rebuild Increases
26. SIZING FOR HA
27. CONTINUOUS PLATFORM INNOVATION REDUCING COSTS • INCREASING DENSITY • REDUCING LEAD TIMES New! 30TB drives - Reduced $/GB By 20%+ New! Ceres Platform – Reduces HW Cost ~12%* New! SCM From Kioxia – Reduces Cost & Improves Supply Chain Dual Sourcing Reduces Costs and Risk *COMPARING LIKE FOR LIKE DRIVES
28. A NOTE: NOT A HW COMPANY MANUFACTURER APPLIANCE UNIVERSAL STORAGE SOFTWARE COMMODITY STORAGE HARDWARE
29. 2 x BLUEFIELD-1 SMART NICs PER TRAY 4 x 100Gb ETH/IB PER TRAY 2 x HOTSWAP DNODE TRAYS REDUNDANT PCI SWITCHES ENABLES SINGLE-PORTED DRIVES 8 x HOT-SWAP SCM DRIVES U.2 DRIVES ON SLEDS 22 x HIGH-CAPACITY QLC RULERS 15TB or 30TB NVMe DEVICES CERES DBOX 1U NVMe-oF JBOF
30. POWER 91% LOWER 83% LOWER PARITY VS. HDDs Universal Storage FlashBlade PowerScale Archive (A300) Universal Storage FlashBlade PowerScale Archive (A300) Universal Storage FlashBlade PowerScale Archive (A300) UP TO UP TO SPACE COST
31. 600PB: IN JUST 14 RACKS CBOX + DBOX MAX ARCHIV CLUSTER SIZE W/ TOR SPINES 304PB OF USABLE SPACE | 600PB @ 2:1 5.1TB/S OF READ SPEED 640GB/S OF WRITE SPEED 2X 64 Port 400GE Spines 16 Racks 8 uplinks per rack 4 Uplinks to each Spine
32. CONFIDENTIAL SUPERPOD. SUPER SIMPLE. Q3’2022 NVIDIA’S END-TO-END DATA CENTER VISION
33. NEW FOR ITPT’22 DATA REDUCTION
34. BLOCK SIZE PERSPECTIVE 128KB 32KB 8KB Commvault HNAS: 1/4th Less Susceptible to Noise DataDomain: 1/8th Less Susceptible to Noise
35. DEVICE PERFORMANCE: MB/s (LOG SCALE) HDDs STINK. AMIRIGHT?
36. HDD DEDUPLICATION SYSTEMS WERE NOT DESIGNED FOR FAST RESTORES A CRITICAL FAILING OF THE RANSOMWARE AGE
37. FLASH PROVIDES NO-COMPROMISE RESTORE PERF. RANDOM ACCESS ENABLES NEW DATA REDUCTION OPPORTUNITIES; VAST’S DATA STRUCTURE IS VARIABLE LENGTH @ ~32KB VAST IS THE INSURANCE POLICY RESEARCH PERFORMED BY
38. 8KB 1 byte 8K OPPORTUNITIES FOR DELTA
39. SIMILARITY VAST’S NEW GLOBAL COMPRESSION METHOD INCOMING DATA SIMILARITY CLUSTERING DELTA COMPRESSION Compressed Together Hashed to data cluster by similarity Stored to QLC https://vastdata.com/whitepaper/#similarity-reduction-to-the-rescue
40. After we have our chunks, send them through the write pipeline to be deduped, matched for similarity, and compress the rest NEW ADAPTIVE CHUNKING: 4.3 (April ‘22) HOW IT WORKS
41. Commvault backup files of some SQL Servers Some Unstructured Data Some Lab Virtual Machines 7.88:1 (+70%) 1.96:1 (+50%) Similarity w Adaptive Chunking 6.83:1(+30%) Data Domain 4.7:1 1.3:1 5.3:1 SURPRISE… WE GOT A DELL ☺
42. LEGACY STORAGE STINKS. AMIRIGHT? mirroring limits scaling limits
43. CONFIDENTIAL MIRRORED GLOBAL COMPRESSION DICTIONARY IN SHARED-EVERYTHING SCM AVAILABLE TO 10,000 STATELESS CONTROLLERS A SHARED-EVERYTHING DICTIONARY
44. ONE MORE SURPRISE… DATA-AWARENESS IN 4.4 (July) FLOATING POINT NUMBERS <?php $a = 5.205; $b = 1.3e3; $c = 6E-10; ?> INTEGERS 4356 is : int(4356) -39 is : int(-39) 47 is : int(47) 75 is : int(75) AN ADDITIONAL 25% REDUCTION SEEN IN TESTING; MARKET DATA LIFE SCIENCE DATA DATA WAREHOUSES
45. THE DATA REDUCTION LAYER CAKE WE’RE BAKING SINGLE COPY DICTIONARY: 10,000 CONTROLLERS SAVE $1000s PER CONTROLLER SIMILARITY WITH ADAPTIVE CHUNKING 2X MORE REDUCTION THAN THE BEST… WILDLY BETTER THAN OTHER FILE STORAGE NEW DATA AWARE COMPRESSION AN ADDITIONAL 25% COST REDUDCTION ($100K/PB), MORE STILL TO INVENT!
46. PAST ARCHIVE UNIVERSAL STORAGE BACKUP NEARLINE FLASH HISTORY’S FASTEST GROWING COMPANY… IS JUST WINDING UP.
47. VAST DATA THANK YOU