Identifying The Scope Of Cloud Transition For DevOps Processes

There are various aspects while choosing one cloud platform over others. I am here trying to put some basic thoughts for DevOps processes which can help you decide what path we should take. Here you go.

Build and deployment

• What all tools are used to create builds (Ex. Jenkins)
• What tools are used to store builds and configurations (Ex. Nexus server)
• What scripting language is used to support the build process (Python, Ruby, Shell, etc)
• What is the build promotion strategy (Ex. Dev->Stage->Prod)

Infrastructure creation

• What is the deployment infrastructure (on-prem or cloud)
• Which cloud is being used
• What kind of VM’s are being used in one cloud infrastructure, are they available in other cloud infrastructure
• Region availability of one cloud infrastructure in another cloud service provider (for West US2, West EU, East EU, Singapore, etc) for solving latency related problems
• What tools are used to create infrastructure (ARM template is recommended in Azure but Terraform is a common template language for both Azure and AWS)

Monitoring

• What tools are being used in one cloud environment vs its availability in other cloud infrastructure (direct API integration, plug-in availability).
• If the same tool is available, what are the approaches of integrating it in the newly chosen cloud infrastructure (Ex. Datadog, NewRelic).

Security agent

• What security agents are being used in one cloud and if that has integration available in other cloud infrastructure nodes (ex. Qualys agent)

OS support in desired VMs

• A product can have a combination of modules running on different OS configurations (Linux (with different flavors), Windows, etc). Are they available directly in VM’s provided in the infrastructure list?
• The cost of certain VMs is higher in one region that another.

Log analysis

• Agents like Sumo needs to be checked if they have support on desired cloud infrastructure VM configuration

The memory footprint of used agents

• It is quite possible that the memory footprint of one agent in one environment is controlled while in other machine configuration it is shooting up. Different Linux flavors may depict this behavior so it is recommended while transitioning to use the same VM configuration, OS build versions, etc.

Configuration management

• Need to understand closely what tools are used for configuration management (Chef, Puppet, etc)
• What additional features these tools are providing (process or VM monitoring, resiliency, etc)

Load balancers and geolocation-based traffic management

• Checking closely on load balancers rule, the configuration is important and how this is available in different cloud setup for comparison and availability.
• What traffic management tools/services are being used and their setup rules (for example Azure traffic manager provides 6 different configurations to choose from and one of them is performance-based routing (landing the traffic to the nearest region based on lowest latency)

Region failover

• How region failover is supported in one region (traffic manager covering all required regions and monitoring health checks to divert traffic when needed).

Security rules (NSG)

• What ports are being used for specific modules and what is denied?
• Are these available to be automated (through some scripts)