Weitere ähnliche Inhalte Ähnlich wie Bitmovin LIVE Tech Talks: Analytics for Workflow Automation (ft. Touchstream Media) (20) Mehr von Bitmovin Inc (17) Kürzlich hochgeladen (20) Bitmovin LIVE Tech Talks: Analytics for Workflow Automation (ft. Touchstream Media)1. All rights reserved. © Bitmovin Inc 2019
April 17TH
, 2020
Analytics for
Workflow Automation
BITMOVIN | TOUCHSTREAM
All rights reserved. © Bitmovin Inc 2020
2. Moderator & Panelists
All rights reserved. © Bitmovin Inc 2020
Sean
McCarthy
Moderator Ninja
Christoph
Prager
Product Manager
Analytics
Brenton
Ough
CEO & Co-Founder,
Touchstream
Pedro
Tavanez
OTT Monitoring &
Development
3. ● Intro
○ The Bitmovin Story
● Benefits of Automation
● Bitmovin Analytics
○ What it is
○ How it works
○ Architecture
● Multi-Codec
○ What it is & Why It Matters
○ Context Aware Encoding
● Operational Monitoring - Touchstream
● Q/A
Agenda
All rights reserved. © Bitmovin Inc 2020
4. Bitmovin’s Backstory
Chris Müller
CTO & Co-Founder
Stefan Lederer
CEO & Co-Founder
Summary
● Company founded in 2013
● Co-created the MPEG DASH standard
● YCombinator 2015
● 30+ US Patents
● Athena Research Project
Solutions
ENCODER PLAYER ANALYTICS
All rights reserved. © Bitmovin Inc 2020
5. All rights reserved. © Bitmovin Inc 2020
Exceptional
Viewer
Experience
Increase viewer
satisfaction by
exceeding expectations
on video quality
Faster Time
to Market
Get assets, products
and, services to
market faster and
more efficiently
Operational
Optimization
Control cost,
mitigate risk and
better manage
complexity
6. PLAYER
Reach viewers on largest number
of devices using a multi-device
player. It’s flexible and easy to
deploy, utilizing built-in SVOD and
AVOD features, giving complete
control of the viewing experience.
ENCODER
Deliver content in the highest
quality at the lowest streaming
costs. Prepare content at high speed
and deploy in any environment.
ANALYTICS
Fix playback issues with real-time
data before your viewers see it.
Monitor infrastructure performance
insights, anytime, all the time.
Visualize data, your way.
All rights reserved. © Bitmovin Inc 2020
7. Why Automation Matters
It’s a pillar of optimizing operations
Cost Reduction
Free up Engineering Resources (Eliminate Repetitive Tasks)
Accuracy and Consistency
All rights reserved. © Bitmovin Inc 2020
8. Cost adds up across the delivery
chain: transcode, storage, delivery
$$$/(HR)
PREPARATION COST
$/GB
Storage Costs
$/GB
CDN Costs
$/Impression+
All rights reserved. © Bitmovin Inc 2020
Benefits of Automation
+ +
CDN / Distribution
Storage
Consumption
Encoder
Touchstream
Real-time player analytics capture valuable
data which can inform other systems
9. PRODUCT HIGHLIGHTS
Real-time data enables optimized video and ad workflows based on viewer experience
1. All data in real-time: go from macro-view to
micro-view while your audience is still watching
2. Industry benchmarks, alerts, and suggestions
to focus technical resources on the most
important issues
3. Data analysis tools - Integrated with widely
adopted tools (e.g. Grafana, Google Studio,
Tableau) making data accessible across
different business and technical users and
bulk export of the dataset
All rights reserved. © Bitmovin Inc 2020
Analytics
10. ● Customers can query data
in many ways
● Central piece is the API
● Real-time: data is available via
the API and the dashboard
within 5-8 seconds
● Access data in maximum
granularity
● SQL-style query language
Real-time data
All rights reserved. © Bitmovin Inc 2020
User
Bitmovin
Analytics
Dashboard
Data
Export
Dashboard
Visualization
Tools
API
12. Codec Efficiencies Compared to H.264
HEVC
40%
More efficient
than H.264
VP9
40%
More efficient
than H.264
E.g. a HEVC encoded file will use 40% less bitrates than a H.264 encoded file
AV1
30%
More efficient than
HEVC/VP9
All rights reserved. © Bitmovin Inc 2020
13. All rights reserved. © Bitmovin Inc 2020
Adjust, optimize and deliver
This dramatic increase in streaming video consumption
will not break the internet. However, there is public
pressure directed at content distributors to lower
bandwidth consumption while still delivering a
high-quality experience for their viewers.
40-50% Higher Efficiency
15. Suitable use-cases
● Predictable distribution of
viewers amongst assets
● Predictable device/browser
fragmentation
● Long asset life cycles: Cost
amortized over a longer life cycle
● Long-tail: Also non-top titles on
average achieve break-even
viewing figures
The 40% less distribution cost for 76%
of your plays, have to outweigh the
additional costs of encoding 100% of
your assets in VP9 in addition to H.264
All rights reserved. © Bitmovin Inc 2020
Produce Multiple
Codecs Initially
VIDEO CODEC SUPPORT
16. All rights reserved. © Bitmovin Inc. 2020
Bitrate ladder efficiencies The graph shows the different measure taken to
reduce bitrates and the relative bitrate reductions for every step
Hulu Japan - Multi Codec Case
All rights reserved. © Bitmovin Inc 2020
Bitrate (kbps) Resolution
200 256 x 144
400 512 x 288
750 640 x 360
1200 853 x 480
1600 1024 x 576
2500 1280 x 720
3500 1280 x 720
4500 1920 x 1080
6500 1920 x 1080
Total Size 1566.2 MB
Bitrate (kbps) Resolution
240 512 x 288
456 854 x 480
693.44 1024 x 576
1317.54 1280 x 720
2104.61 1920 x 1080
3361.86 1920 x 1080
Total Size 587.3 MB
Bitrate (kbps) Resolution
240 854 x 480
416.87 1024 x 576
792.05 1920 x 1080
1378.81 1920 x 1080
2400.26 1920 x 1080
Total Size 383.6 MB
Standard
H.264 Ladder
Per-Title
H.264 Ladder
Per-Title
H.265/VP9 Ladder
-34.68%-56.92%
Average Bitrate
2350 Kbps > 1045.6 Kbps (-56%)
Storage Consumed
1363 MB > 383.6 Kbps (-72%)
QOE (Full HD Available)
4500 Kbps > 792.05 Kbps (-82%)
Hulu Japan
achieved 30%
CDN savings
18. ● Un-predictable viewing patterns
● Ever changing viewer
characteristics: browser and
device distribution change regularly
● Short asset life cycle: Encoding costs are
amortized over a short period of time
● Potential use cases: Publishers, News
broadcast, Sports platforms
All rights reserved. © Bitmovin Inc 2020
Re-Encoding top
performing assets
?
19. eg: Cost based or Performance Based
● A solid asset performance prediction model
○ The better this model, the earlier you can re-encode
○ The faster you can save $ or improve performance
○ Based on historical or real-time data per asset
● Know your costs
○ Encode costs
○ Storage costs
○ CDN Costs
● Understand Codec Support Distribution
Build a Model
All rights reserved. © Bitmovin Inc 2020
20. ● Query the Bitmovin analytics API
Count the number of impressions/viewing time + codec support for an asset:
querybuilder.COUNT('IMPRESSION_ID')
.filter('VIDEO_STARTUPTIME', 'GT', 0)
.filter('VIDEO_ID', 'EQ', '<VIDEO_ID>')
filter('SUPPORTED_VIDEO_CODECS', 'EQ', ‘VP9’)
.query()
● Result triggers an AWS Lambda function
● AWS Lambda function process the result informed by
CDN/encoding/storage costs
● If threshold is met, AWS Lambda function starts encoding
job in o Bitmovin Cloud Service
● Depending on customer setup, this can trigger an update to CMS
All rights reserved. © Bitmovin Inc 2020
Connect the APIs - Sequence of Events
21. All rights reserved. © Bitmovin Inc. 2019
● Re-encoding top performing assets can be an attractive entry
point into experimenting with multiple codecs
● When combined with automation organizations
reap benefits at scale
○ Save time manually analyzing the data and
viewing patterns of assets
○ Combining with a trained ML model can
maximize cost savings
○ Optimize bitrates and execute a 3-pass or per-title
encode if QoS metrics dictate
○ Leverage other elements of the workflow to achieve the
same for live (cap bitrate at player or manipulate
manifest at CDN edge)
All rights reserved. © Bitmovin Inc 2020
Benefits of Automating
22. The Only Open OTT
VirtualNOC
Built for operations by operations experts
23. Why Touchstream?
Entirely cloud based
23
Simple, focused & operations friendly
dashboards
Stream monitoring takes a unique
approach to End-to-End monitoring
Direct integration with all major CDNs to allow data sharing for faster
problem resolution
Some of our customers:
24. Here’s How We Tackle The
Problem
Alert you about problems
before customers complain
We monitor the availability of
every endpoint, every minute,
every day
We monitor every channel,
every format and every bitrate
24
Quickly identify root cause
We verify the delivery network
is meeting their SLA
We use our PoPs (cloud pollers) to
retrieve data from your CDN and
Origin end point
28. Major Sports
Championship
Hours before the game, Touchstream detected
stream problems that would have impacted
100,000’s viewers.
Operations was alerted and the problem resolved
before the game started.
28
30. Touchstream’s library of
hardware/cloud encoding
monitoring scripts
Touchstream data retrieved via our
PoPs (cloud pollers)
3rd party
integrations
Touchstream VirtualNOC
End-to-End Monitoring
32. Automation - API Driven
Discover assets and events
32
Discover and configure analytics
integration
Automatically configure stream
monitoring
Feedback loop to dynamically update monitoring based on real time
analytics
36. All rights reserved. © Bitmovin Inc. 2019
When Automation is Necessary
All-In - encoding all codecs initially (No automation necessary)
When you understand your viewing patterns this makes the most sense
Scale
Predictable distribution of viewers amongst assets
Predictable device/browser fragmentation
Most relevant in SVOD workflows
VOD Automation - For assets with unpredictable viewing patterns
If you can’t gauge the audience size, device
Live Automation -
All rights reserved. © Bitmovin Inc 2019
37. All rights reserved. © Bitmovin Inc. 2019
Why the Bitmovin API is well suited to support
this
Real time data
Granularity of the data
All rights reserved. © Bitmovin Inc 2019