@ WEKA | June 6, 2022 Session Agenda Pt. 1 | On the Record (55 minutes) • Welcome and Introductions • WEKA Business & GTM Update – Liran Zvibel & Jonathan Martin • Fireside Chat with Hitachi – Jonathan Martin & Jason Hardy, Global CTO at Hitachi Vantara Break – Take 5 Pt. 2 | Under Embargo (55 minutes) • WEKA 4 Launch Preview – Colin Gallagher & Nilesh Patel Break – Move to Dinner (30 Minutes) Dinner: Orchard City Kitchen 5:30 – 8 PM
Facts • Founded in Tel Aviv • Headquartered in Silicon Valley • $140M funding to-date in Series C2 • Data Platform for AI and next-gen workloads • 40 patents granted, 95 pending • Over one exabyte of data under management • Nearly 200 enterprise customers – incl. 8 of the Fortune 50 • Deep GTM partnerships with leading cloud & server vendors • Named a Visionary in the 2021 Gartner MQ for DFS & OBS • 4.9/5 stars rating in customers reviews on Gartner Peer Insights – 5/5 stars for Customer Support & Service Strong Backing & Board Strategic Investors Key Board Members • Dan Warmenhoven (Ex-CEO NetApp) • Dror Nahumi (Norwest) , Menashe Ezra (Gemini) & Roni Hefetz (Celesta/WRVi)
100 Gbit Networking Containers Next Gen Workloads Computer Vision AI Natural Language Processing High Speed Data Analytics DevOps GPU/xPU/ARM WEKA Innovations Fine grained distribution O(1) data structures Load balanced data & metadata No read-write-modify Low latency networking Zero copy data path STORAGE MARKET DISRUPTION
WEKA’s Long-Term Vision A single, scalable, highly performant software Data Platform for hybrid cloud and edge AI Data Pipelines Tier 1 Application (ERP & CRM) Data Lakes & Data Warehouses WEKA Data Platform Ubiquitous Data Services across Hybrid Cloud & Edge Scientific Computing Datacenter Core Multi-Cloud Near-Edge
next gen workloads different? Distributed GPU processing, high velocity & volumes of data built on data pipelines with highly variable IO patterns make next gen workloads very challenging for traditional storage solutions Hybrid Cloud & Near Edge Containers & Microservices By 2023, 90% of enterprises that implement AI pipelines will use containers By 2025, AI will be the top category driving infrastructure decisions, resulting in a tenfold growth in infrastructure requirements In 2025, Cloud spend will eclipse On-Prem spend for the first time
10-100x faster business outcomes for next gen workloads DATA COPY DATA COPY DATA COPY DATA COPY DATA COPY Zero Copy Architecture | Zero Tuning IO Algorithms S3 NFS SMB GPU DIRECT POSIX HDFS
10-100x performance & scale Legacy Storage Protocols optimized for writing data inside a server measured in tens microseconds (µs) Hardware. Doesn’t play well with cloud Fragmented into BLOCK | FILE | OBJECT for Storage, and Data Lake / Data Warehouse WEKA Data Platform New parallel data-plane and control-plane protocols load-balancing IO and metadata over fast (100gbit+) networking delivers 10-100x faster performance than Legacy Storage, with linear scalability up to 14EB Software. Cloud Native. Deployable in hybrid and edge environments A Data Platform that delivers all the benefits, none of the compromise New parallel data-plane and control-plane protocols
application servers processing WEKA data 4+ access protocols support complex data pipeline AWS Lift Lift & shift on prem to AWS 6PB of genomics data for 12M customers in AWS 12K GPUs World’s largest GPU Supercomputer 1 Week on WEKA was equivalent to one year on the competition! Autopilot Allowed customer to be amongst the first movers on autopilot, years before the competition 70% Save Tiering between flash and HDD tier allowed 70% of cost saving compared to competition S3 + GDS Ingest data from instruments using fastest S3, then process on GPU DB using GPU Direct Storage AI Yield Manufacturing 4.0 project to improve yield on phone & watch displays for Apple & Samsung Auto Scale Easily add compute when they need so, 5PB of capacity 5x perf 500% more performance compared to the alternative on-premises solution, 20% cost reduction compared to FSX for Lustre Edge2Cloud Treating the wet instruments as Edge, pushing all processing to the AWS cloud CLOUD DATACENTER CORE DATACENTER CORE EDGE / CLOUD
Platform 4 CONFIDENTIAL SECTION: UNDER EMBARGO UNTIL JUNE 15 at 6 AM PT Colin Gallagher Vice President, Product Marketing Nilesh Patel Chief Product Officer
workloads to cloud: devops, operations, QA, backup • Haven't moved high data gravity, large scale, or high-performance workloads § Security posture § Availability § Performance characteristics Most Organizations Have Made the Leap to the Cloud Have One of More Cloud Initiatives "We are 100% cloud… …except for the 3 workloads we run our business on."
Can’t Compromise Cloud can’t be fundamentally different Same across Core, Cloud, Edge Speed, scale and simplicity Double Double Identical Code and Capabilities
Gen Workloads in AWS + 230PB of data moved to AWS; Available in AWS Marketplace Scalable architecture from on-premise to the cloud Cloud native with robust enterprise feature set Highest performance with zero tuning or customization
Data Platform for AI & Next-Gen Workloads NEW: Advanced Data Reduction Capabilities for Class Leading Economics NEW: Simplicity at Hyperscale Introducing WEKA 4 The first high performance scalable data platform for on-prem, edge – and multicloud
for even better economics Leverage the same data across multiple applications Simplify the migration to and between clouds Move any workload to any cloud, even the “impossible” ones The Industry’s First Multicloud Data Platform for AI
Multicloud Data Platform for AI ON WITHIN TO WITH BETWEEN Move or Backup to Clouds Migrate or DR Between Clouds Tier and Reduce Data Within a Cloud Use the Cloud for Data Tiering Run Natively on the Cloud
for All Cloud Use Cases Move or Backup to Clouds Migrate or DR Between Clouds Tier and Reduce Data Within a Cloud Tier Data to Cloud Run Natively in the Cloud
Cloud Simpler Auto-scale Performance Automatically Tier & Restore Data Easily Manage at Hyperscale All Protocols Access the Same Data Easily Move Data to Cloud Eliminate Data Copies
in the Cloud with WEKA Extreme I/O performance at low latency Sequential & random I/O, small & large files Massive scalability of capacity & performance
Cloud Move data to the cloud to leverage compute for bursting or hybrid workflows Incrementally and continuously promote filesystem changes non-disruptively
Data Reduction File System-Wide for Greater Effective Capacity Media Choice Cost, Availability, Scale TLC or QLC Data Tiering Local & Cloud Object Storage Global Namespace + +
Expand Usage NFS v4 Core NFSv4.1 features natively integrated into the WEKA stack Instant Object Retrieval Milliseconds retrieval from lowest cost storage (e.g. AWS Glacier) for rarely accessed data SMB-W Natively integrated WEKA stack for improved latency and small file performance Incremental Snapshots Incrementally and continuously promote filesystem changes without remounting
itpresstour
line_developers_tw
shinyasaita
ynt0485
kinunori
jacopen
cscengineer
cmusudakeisuke
kworkdev
hiroramos4
iotengineer22
findy_eventslides
inesmontani
oripsolob
elsirapls
akihiro_kokubo
jeffersonlam
ryanjones
dougneiner
aleyda
trishagee
mojombo
mkilby
