4. Spark: Innovating in the Cloud 1st Sept
Supporting & growing 6 startups
The Ventures could act like startups, we
& Act with
“Build now, scale for later”
5. Spark: Innovating in the Cloud 1st Sept
Show of hands: “Does any one know much about Spark Ventures?”
Created 6 ventures from scratch - creation of lean Canvas to POC and business case and delivery
Bigpipe – ISP
Qrious – Big data platform
Semble – mobile wallet
Morepork – Home automation
Lightbox – Video on Demand
Invest in 2 others
Putti and Vigil
Started Spark Ventures just over 3 years ago…why…to create new revenue streams available in the digital space as demonstrated by this recent study conducted by EY with global Telcos (TV and Video being at the top…so Lightbox fits the bill there). Traditional Telco revenues were declining or static at best….
“the reason Telecom is now Spark”
Part of shared services group providing Ops, Customer Care, IT and Software Delivery to the Ventures
Mindset shift – non traditional view of IT, automation, non silo’d way of working, multi skilled with cloud and SaaS experience,
Keep costs low - they are start ups and as such want to keep their cost as low as possible
People - Hiring a team based on attitude, fit, and their multi skills/experiences, customer first approach, happy to work in a fluid environment. Had to bring in experienced AWS resources (such as eIT) before we built up our skills internally. Collective recruitment, where team interviews potential candidate.
Scale with Stability – Automate, grow as the Ventures grow, working with a hybrid environment
Guiding principles - to base architectural decision on – such as cloud first, build now, scale for later, 80% fit is ok, automate and reuse, helped empower our people
High level over view of AWS:
MultiAZ highly available dev/test/prod environment includes EC2 instances, Elastic Load balancer, RDS multi AZ instances, security groups, monitoring/logging (cloudwatch), SSL implementation.
Scale – can scale up easily as the business grows such as auto scaling or Reserved instances or shut down without affecting the other Ventures
Cost – can attribute costs to a business (as we have created individual accounts per Venture), monthly fee, no need to buy expensive infrastructure especially as they were start ups.
Hybrid model – work loads both in public and private clouds
DevOps – Ops and Dev working closer together to deliver to sprint time frames (by attending Sprint Planning sessions), created a sandpit environment so that devs can start working on a POC while AWS team build proper environments.
Closer relationships with Product Manager to ensure clear understanding of requirements and user stories
Cloud first – supports our cloud first guiding principle
DR – inherently in place due to using multi AZs
Billing – understand costs, per month, look to optimise, we also use a 3rd party SaaS solution (Cloudyn) to assist us.
Delivery – allows us to keep up with the rate of change required by the Ventures – standing up environments in a few hours for devs to start development work
Automation – All builds are templated, we used Puppet but moved to Packer and Cloud Formation
Deployments – can complete deployment to production in 5-10 mins
Re-use as much as we can – designs to build templates
A quick view of some Venture stats
spinning up the environments from 1 day to 15-30 mins. Manual> Puppet> Packer
Deployments – 5-10 mins using bamboo and Elastic bean stalk (EBS)
A high level view of the under lying technology we use to create and support the Ventures