Scaling Investigations Next-Generation Visual Graph Analytics - - PowerPoint PPT Presentation

scaling investigations
SMART_READER_LITE
LIVE PREVIEW

Scaling Investigations Next-Generation Visual Graph Analytics - - PowerPoint PPT Presentation

Scaling Investigations Next-Generation Visual Graph Analytics through GPU Cloud Streaming G R A P H I S T R Y Leo Meyerovich, CEO @LMeyerov Graphistry Inc. 2017 Graphistry Inc. 2017 info@graphistry.com info@graphistry.com ABOUT Graphistry


slide-1
SLIDE 1

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

G R A P H I S T R Y

Scaling Investigations

Next-Generation Visual Graph Analytics through GPU Cloud Streaming

Leo Meyerovich, CEO @LMeyerov

slide-2
SLIDE 2

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

ABOUT Graphistry is a place investigators can record scalable investigations, run them, and visually analyze the results

slide-3
SLIDE 3

Graphistry Inc. 2017 info@graphistry.com

Investigating through event data is important… but HARD

  • Customer 360
  • System outages
  • Genomics & health records

...

  • Anti-human trafficking
  • Cybersecurity
  • Retail & billing fraud

Goal: Quickly answer progression, root cause, correlation, scope…

slide-4
SLIDE 4

Graphistry Inc. 2017 info@graphistry.com

Today: Search & Dashboards over Data Marsh

slide-5
SLIDE 5

Graphistry Inc. 2017 info@graphistry.com

100+ Increasing details & data sources ~infinity Increasing resolution & duration (+ Increasing iterative workflow complexity)

slide-6
SLIDE 6

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

Handling 100X More Data in the Analyst Loop: Three 100X Technologies

Make AI visual with Graph Automate with Visual Playbooks

Graphistry Inc. 2017

See 100X+ more data with GPU Visual Analytics

info@graphistry.com

slide-7
SLIDE 7

Graphistry Inc. 2017 info@graphistry.com

Tod Today

  • Rethinking event data as graph analytics
  • GPUs for interactive analytics
  • GPUs for data-driven design
  • Graphistry’s client↔cloud end-to-end GPU architecture
slide-8
SLIDE 8

Graphistry Inc. 2017 info@graphistry.com

Problem: How to Understand 20+ Columns? ??

select * from ALERTS where PRIORITY > 8

srcIP dstIP alert score time … … … … …

slide-9
SLIDE 9

Graphistry Inc. 2017 info@graphistry.com

Physical Topology: Behavior Between Entities

srcIP dstIP alert score sensor … … … … …

slide-10
SLIDE 10

Graphistry Inc. 2017 info@graphistry.com

N-Partite: Relationships – IP <-> Alert

srcIP dstIP alert score sensor … … … … …

slide-11
SLIDE 11

Graphistry Inc. 2017 info@graphistry.com

Hypergraph: Items <-> Dimensions

srcIP dstIP alert score sensor … … … … …

slide-12
SLIDE 12

Graphistry Inc. 2017 info@graphistry.com

Hypergraphs vs. Strongly Connected Components

e1 e2

#Edges = O(#events * #attribs) #Edges = O(#events * #attribs^2)

slide-13
SLIDE 13

Graphistry Inc. 2017 info@graphistry.com

Why GPUs for Graph Analytics?

  • Goal: < 100ms interactivity
  • ML for many columns: clustering, community detection, …
  • Irregular data parallel algorithms increasingly well-understood!
  • 60X+ over other visual graph analytics tools
slide-14
SLIDE 14

Graphistry Inc. 2017 info@graphistry.com

Tabular Analytics Too! For Node Table and Edge Table

Hi Histogr grams Fi Filter ter

slide-15
SLIDE 15

Graphistry Inc. 2017 info@graphistry.com

100X Speedups: We Make Your Device Span GPU Client + Cloud

15

Optimized networking

  • 1. Scalable rendering engine

leverages WebGL (GPUs) in browsers

GPU analytics & viz GPU rendering

(No JavaScript!)

GovCloud

  • 2. Scalable analytics engine

leverages GPUs in the cloud

20ms 50ms

slide-16
SLIDE 16

Graphistry Inc. 2017 info@graphistry.com

Data-Driven Design for Scale

16

slide-17
SLIDE 17

Graphistry Inc. 2017 info@graphistry.com

Designing for Scale: Data-Driven Visualization

In Graphistry, algorithms & automation assist defining:

  • Layout – clustering (ForceAtlas2) & sorting
  • Size – data attribute or degree; semantic zoom
  • Color – data attribute or community
  • Transparency – if edge start/end nodes are on screen
  • Camera – bounding box
  • Labeling - most prominent data
  • Curvature – edge direction

17

slide-18
SLIDE 18

Graphistry Inc. 2017 info@graphistry.com

Result: Can finally see & explore big systems for killchains, patient cohorts, correlated incidents, …

slide-19
SLIDE 19

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

Handling 100X More Data in the Analyst Loop: Three 100X Technologies

Make AI visual with Graph Automate with Visual Playbooks

Graphistry Inc. 2017

See 100X+ more data with GPU Visual Analytics

info@graphistry.com

slide-20
SLIDE 20

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

Handling 100X More Data in the Analyst Loop: Three 100X Technologies

Make AI visual with Graph Automate with Visual Playbooks

Graphistry Inc. 2017

See 100X+ more data with GPU Visual Analytics

info@graphistry.com

slide-21
SLIDE 21

Graphistry Inc. 2017 info@graphistry.com Graphistry Inc. 2017 info@graphistry.com

Curious about more reliable & scalable visual investigation workflows through 100X technologies? Now piloting with teams doing security, development, and more! info@graphistry.com

G R A P H I S T R Y