17 Nov
Google Cloud Application Migration
Google Cloud Platform (GCP) is Google’s public cloud service, which competes with Amazon Web Services and Azure (see our articles about Google Cloud vs AWS and Azure vs Google Cloud). GCP supports virtual machines like its competitors, but has a special focus on modern, containerized workloads and the ability to efficiently run machines learning models.
While Google is a smaller player in the cloud services market, it is rapidly catching up, and is perceived to have several important advantages:
Pricing
Google prices services by the second, with discounts for long-running instances with no upfront commitment. Also, instance prices can be up to 40-50% lower than other options.
Private global network
GCP leverages Google’s private global fiber optic network, which offers very fast direct interconnectivity between their data centers.
Live migration
GCP is able to migrate workloads seamlessly from one virtual machine to another with no downtime.
Security
GCP offers the same security measures used for Google’s own global-scale services like Gmail, Google Search and Google Docs.
Google Cloud Migration: A 5-Step Strategy
When planning your migration strategy to GCP, Google advises you follow these five steps.
1. Assess
Evaluate applications and workloads’ suitability to the Google Cloud. Primary considerations include:
- Hardware and performance
- Number of users and licensing issues
- Compliance
- Dependencies between applications
Divide your apps into three categories:
- Easy to move
- Hard to move
- Can’t move
2. Pilot
Take one or two applications, preferably from the “easy to move” bucket, and migrate them. Run the apps in production for a while and measure performance. Understand the licensing requirements if you scale up your Google Cloud workloads or add more applications, and always plan for a rollback to on-premise or another cloud in case migration fails.
3. Move Data
Google advises moving all your data to the cloud first, then moving the rest of your applications. Consider the storage tiers offered by Google Cloud Storage (Standard, NearLine Storage and ColdLine Storage), SSDs vs. regular hard disks, and database services like Google Cloud SQL, Datastore and Bigtable. Plan how you’ll physically move the data—data transfer, sending an offline disk to the Google data center, streaming to persistent disks, etc.
4. Move Applications
If you can, perform a direct “lift and shift” of your applications to Google Cloud. For example, by creating a local virtual machine representing your workload and importing it as a Google VM, or backing up your app to GCP and thus automatically creating a cloud copy. If a simple option is not possible, consider rebuilding applications in the cloud with a combination of custom VMs and GCP infrastructure services.
5. Optimize
Now that applications are running in the cloud, consider ways to make them better:
- Make your apps redundant across GCP availability zones
- Plan for disaster recovery using Google Cloud Backup
- Set up elasticity with autoscaling groups
- Set up monitoring for your workloads with Google Stack driver
- Move static assets to cold storage
- Use Google’s Deployment Manager to launch and scale new instances
Once your strategy is laid out, you should get familiar with the Google Cloud tools and services available which can facilitate the migration process and help overcome any storage challenges that may lie ahead.
