Venture capitalist Matt Ocko’s 20-year track record of success in the startup world has given him unique insight into how AWS has changed the venture financing process. In this session, you’ll learn about industries susceptible to disruption by AWS-based startups, and where VCs are willing to take new risks on those startups, including the heavily-regulated medical, government, financial, and industrial sectors. Matt will talk about how new, supercomputing startups are now possible because of AWS technologies. Hear about how using AWS technologies can actually reduce risk – and reduce time to customer penetration – from a VC perspective, and how to go from ‘AWS to Series A’ in 5 easy pieces.
3. It’s Not That Bad, and AWS Can Help
• Reduce your startup cost
• Accelerate your time to
market
• Create a differentiated
startup, get VCs’ attention
• Prove to VCs there’s an
enterprise-grade product
• Help get customers more
rapidly
All of which
help to
• Preserve founder
equity
• Increase chances of
funding
• Decrease time to get
funding
• Increase valuation in
Series A
• Increase valuation in
subsequent rounds
4. The Five Easy Pieces
• Be brave
– Even in heavily regulated industries; go where others fear
• Be simple but smart
– Don’t use features you don’t need, keep failure modes minimal
• Design for resiliency and uptime
– Assume you will be hacked, crashed, fat-fingered
• Plan for scale
– Design and test for success
• Build for the zero-touch sale
– Use AWS to your advantage to make customer engagement frictionless
5. The VC Perspective
• Be brave
• Be simple but smart
• Design for resiliency
• Plan for scale
• Build for zero-touch sale
Don’t want to fund the 11th SaaS
whatever
Simple=fast adaptation=lower risk
Enterprises don’t tolerate broken
consumer s---
See above, plus tiny businesses
uninteresting
First to customer w/ best experience
wins
6. Be Brave
• Now possible to attack regulated industries
– VPC to meet regulatory hurdles
– CloudHSM for customer crypto friendliness
– Good soft crypto for overlay on top of S3
– Glacier for long-term records retention
– CloudTrail and Loggly for auditability
• Good enough for CIA, good enough for F500
• HIPAA, SARBOX, Dodd-Frank are good barriers
7. Be simple but smart
• Don’t overengineer
– If you’re not doing crazy TPS, ELB works fine; Beanstalk!
– Redshift and results vs. months of engineering
• Consider costs before implementing
– Is it cheaper to run your own stuff on bare nodes, or…?
– Do not underestimate ops costs when doing this calculation
• Be reductionist about core processes
– Use AWS to simplify, where possible – huge time value
8. Design for resiliency
• Think in multiple AZ from day one
– AWS features to manage this complexity (RDS) where possible
• Stateless where possible, backup where not
– Either via AWS features or via S3
• Real time monitoring
– Whether CloudWatch, NewRelic, etc.
• Redundancy of data, state, compute
– Either your own code, or novel solutions like CloudVelocity
9. Plan for scale
• Simplicity pays off: clone and conquer
– Ability to add capacity by cloning your modules, AWS for rest
• Consider in advance how demand will burst
– And where you need to code vs rely on AWS
• Understand your costs; don’t recklessly burn $
– Don’t use nodes like cannon-fodder; solutions like Cloudability
• Culture is part of successful scaling
– Focus on results vs. engineering fetish, NIH
10. Build for the zero-touch sale
• There’s no such thing, but key aspiration
– First to the customer, with lowest friction, wins
• Be creative with AWS to drive towards this
– More than customer sandboxes: complete systems
– Start with sample data, then allow upload
– Heck, sell the customer the final state sandbox
• Consider how AWS elasticity solves post-sale
– Ability to clone customer config for debug, hot standbys, rollback
11. EXAMPLES
• Companies that have executed to the “Five
Easy Pieces”
• Covering who they are, and some (not all) of
how they use AWS
12. BlueTalon
• Emerging leader in secure data collaboration and
unification across enterprise boundaries
• Cloud-based BlueTalon Virtual Database
integrates data from multiple databases on the fly
• Demanding Fortune 500 customers
13. BlueTalon on AWS for Developers
• Bring data from on-prem
databases into cloud apps
• Integrate data from disparate
databases
• Implement data policies for
privacy and selective access
14. Automate Setup with AWS VPC
AWS Feature
• VPC with Private and
Public Subnet and
Hardware VPN Access
– Spin out pre-configured
combinations of nodes with
security rules
– Pre-configured VPN
connectivity to datacenter
BlueTalon Advantage
• Quick customer setups
– hours not days
• More secure setups
– no accidental security
holes
15. Integrate AWS IAM in BlueTalon UI
AWS Feature
• APIs to use AWS IAM
within custom web apps
– Reuse IAM identities in
BlueTalon through APIs
– Use right level of IAM
access for operations
within BlueTalon
BlueTalon Advantage
• More secure operations,
– no need to see customer
data to setup BlueTalon
• Seamless administration
– Operations transfer
continously from AWS to
BlueTalon
23. How Prism uses AWS
• S3
– Easily and securely store at-rest data
– Seamless on-disk encryption
• ELB (Elastic Load Balancers)
– Handle all of our API calls/traffic over SSL
– Provide SSL for all API calls and data transmission without the need for
additional computational overhead
• EC2 (Elastic Compute Cloud)
– Global infrastructure enables near-infinite scalability and worldwide
deployment
24. How AWS has helped Prism
• Deploy Instantly
– Add on high-volume customers without worrying about network or server load.
• Connect cameras to the cloud
– Intelligently extract images and meta data at less than 50kb per second
• Store data infinitely
– Any number of users can access images or analytics over any time period
• Analysis on demand
– Immediate understanding of activity and movement in any physical space
• Present dynamic visual summaries to customers
– Transform mountains of data into intuitive imagery
26. AWS
customer
Downstream
application
sequencer
• 70GB of input data per sample
• >2000 Core*hours
• 10s-100s samples / day
• AWS was a natural solution
• But, EC2 accounted for 90% of the cost
AWS natural solution for Moleculo
28. How many core*hours can $100 buy?
-
2,000
4,000
6,000
8,000
10,000
12,000
Rackspace Google cloud Amazon on
demand
Amazon
reserved
Hardware Spot
Instances