Introduction
Hello all, we need a preliminary design for Zero Trust Architecture in cloud services. I have a draft of the boundaries but the goal of this meeting is to identify the “must haves” for each service type. “Must haves” include things like boundary isolation, segmentation, no public ips, or public ips acceptable, etc.
Service Types Compute Data Storage Cognitive Utility On-prem to Cloud Cloud to On-prem Cloud to Cloud Cloud to 3rd party SaaS
The goal here is to say if we are working in a particular service in the cloud, we need at a minimum “x requirements met”. I will show you an example. Define the automation we need DevOps to build to accommodate (there is a backlog item today) Draft a policy around ensuring we are protecting our build out
Comments from meeting: What is the level of “best practice” we should use? Can we use Microsoft to help scope the PaaS and ecosystem? ZTA and the environment is massive ZTA for on-prem is being implemented with Guardicore Has applications that tie components together Can we tie it into the CMDB
What does it mean to Seagen? What are we doing today?
What problem will it solve?
Why do we want to do this?
What’s in it for me?
What will it give me?
Fundamentals Missing Source of truth - what defines our tags and applications 802.1x - what constitutes a Seagen owned device (InTune, Lab, cloud service, etc.) Unified tagging scheme for applications How do we classify applications Membership
Apply security to what architecture or systems?
Inventory
Monitoring and Remediation
Hub and Spoke Model for cloud
Next Steps Invite Gartner and Microsoft to describe what ZTA is Use the buzz word to help us How does Seagen define ZTA? What problem are we trying to solve? What is allowed to talk to what, systems and people
Problems
Applications and ecosystems
Segmentation and permissioning applications that talk to each other
Automation of onboarding new environments into the network
Developing and accessing cloud resources
Cloud native applications or ecosystems
Managed containers
ZTA between cloud providers (Azure to GCP)
Talk back to production systems on premises from cloud
What is the workflow and how does it apply
Standardize how we get things up and running (model, RA)
Technology covered via ZTA strategy
Where does threat modeling occur?
Train developers on risks, interdependencies
App Dev - configuration, what should or should not be stored locally
How we handle problems in Azure, critical, high, etc?
Who does it?
What are we trying to protect?
Understanding our data, how it travels, who sees it, who should not see it
Does everything have a baseline level of protection?
Model based on the basis that everything is dangerous, like if all of our systems were on the internet
Control plane or data plane in the design
Where they are logging on
Device
Data, is it normal or abnormal
Real time monitoring and catching issues
RA for how to connect and where
Scenarios
TOPS data science in the cloud needs to get to a system that is currently locked down on-premises. How would this work?
Approval process
Personas
Application Developer
Data Engineer
Machine Learning/Data Scientist
Cloud Engineer
Ownership
Delivery
Scenarios mapped to security profile
Secure Zones
Subscription
VNET
Subnet IPs
Firewalls
VPNs
Deliverables by EOY
What is the structure in cloud when we use and configure it? RA
Publish standards
Definition of an application
Process for identification
Make sure we know what we have
Understand the data we want to grant access to
Verify an identity or things in play, you are who you say you are. Any perspective.
Tie an instance to a security group
We need a source of truth on grouping
Need unified tagging strategy across environments (By end of November)
Devices
Resources
Members
Groups
What does the app do
Where does the data live
What happens on the wire
Holistic view of applications