Public Cloud providers (AWS, Microsoft Azure and GCP) have been acquiring migration-tools to provide fast ways to migrate existing workloads to the public cloud. As we have not found any comprehensive study of tools and hands-on-experience reports so we decided to do it with our skilled cloud engineers and consultants.
We selected tools listed in chapter 2 for the following reasons:
- they are preferred tools by cloud vendors
- CloudEndure – AWS
- Azure Migrate: Server Migrate – Microsoft Azure
- Migrate for Compute Engine (formerly Velostrata) – GCP
- customer demand for tools has been high during 2019
- we need to understand fundamental differences/restrictions of the tools
Our teams included skills from all public cloud providers and plenty of migration experience from various methods (https://aws.amazon.com/blogs/enterprise-strategy/6-strategies-for-migrating-applications-to-the-cloud/). The trick was that we formed our teams so that AWS experts used GCP tools and GCP experts Azure and so on.
Selected tools and experiences
CloudEndure (Acquired by Amazon Web Services)
CloudEndure offers highly automated disaster recovery and migration solutions into AWS.
With support for any source infrastructure and all applications running on supported operating systems, CloudEndure ensures that your entire IT landscape will remain robust and reliable as it continues to grow.
CloudEndure requires installation of agent to the operating system. The Installation is easy and fast for Linux and Windows operating systems. For small linux installation it took less than one minute:
After the agent is installed the machine appears to CloudEndure website under the migration project. The first initial data transfer takes about 1 minutes per 1 GB of compressed and encrypted data in moderate environment. The data is transferred to CloudEndure replication instance via AWS elastic (public) ip and 1500/tcp port. The replication instance is automatically created by CloudEndure to the customer target AWS account when first agent connects to CloudEndure. The replication instance saves information to EBS volume (one volume per each disk from each machine) and takes snapshot after the initial sync phase is completed.
When initial phase is completed, the agent starts transfer only changes to replication instance, and user can start actual migrated instances to the AWS account. In CloudEndure you define Blueprint (template to describe new migrated instance), which ip address it should use, which subnet, which security groups etc. The new instance is created in a few minutes:
- Migration tool without any big issues
- Straightforward approach
- We managed to migrate all source systems within given time frame
- Automatic migration target EC2-type choosing algorithm isn’t spot on. You can choose the correct target instance type when defining Blueprint.
- CloudEndure migrates licenses to Bring-Your-Own-License model in AWS. Eg. if source machine has Windows Server with SQL Server the migrated instance will be Windows Server, not Windows Server with SQL Server instance.
- We are sure that the CloudEndure console will be integrated to AWS Console (look and feel, user logins, etc.).
- Agentful approach with block level copy of disk (Download and run agent on source server. No installation needed.)
- Tool is very scalable
- All source machines supported if you can install the agent
- Free of charge (migration target to AWS)
- Continuous data replication
- Setup the tool (1 hour) – this was really straightforward task
- Time to implement first migration (30 minutes)
- Migration of the source machines (1 hour)
Google Cloud with Migrate for Compute Engine (formerly Velostrata)
Cloud migration creates a lot of questions. Migrate for Compute Engine (formerly Velostrata) by Google Cloud has the answers. Whether you’re looking to migrate one application from on-premises or one thousand applications across multiple data centers and clouds, Migrate for Compute Engine gives IT teams the power to migrate these systems to Google Cloud.
- There are major issues in the documentation with Migrate for Compute Engine
- Some challenging phases in the migration (some limits, naming etc issues)
- Integration to Google Cloud Platform has been partly implemented
- Learning curve quite steep (major usability challenges with UI), you just have to know how to use the product
- Seems that REST API is the best way to use this, which is, of course, the right way in any large scale operation.
- Migration sources supported: VMware, AWS, Azure
- Needs worker in the source environment (for example AWS AMI is available from Market Place)
- Run in cloud mode for transparent/seamless migration
- Agentless (Block mode based migration approach)
- Setup the tool (4 hours) (This time will reduce dramatically when we have more experience)
- Time to implement first migration (1 hour)
- Migration of the source machines (90 minutes)
Streamline your migration journey
Discover, assess, and migrate all your on-premises applications, infrastructure and data. Centrally plan and track the migration across multiple Microsoft and independent software vendor (ISV) tools.
Azure Database Migration Service (DMS) – Cloud and agent-based migration solution. Migrate data from other Azure PaaS SQL Databases or SQL Server on Virtual Machines to Azure database services.
- Heavy capacity requirements (amount of memory, number of vCPUs)
- Customizable assessment (infrastructure can be customized and not transferred as is)
- Automation possibility with push installation, automated deployment
- Limited OS support (AWS Linux kernel not supported but this is usually not a problem in VMware environments)
- Relatively well documented but UI / documentation / terminology sometimes inconsistent
- Community support lacking. Does anybody really use these tools?
- Discovery functionality limited
- Agentful approach (Agent needs to be installed all servers (AWS / physical servers)
- Agentless features available for VMware vSphere
- Configuration / Process server pushes data to Azure Migration Server Migration tool.
- Doesn’t include features for assessment / discovery of cloud workloads
- No need to allow inbound traffic to source environment from external sources
- Setup the tool (3 hours)
- Time to implement first migration (1 hours)
- Migration of the source machines (1 hours)
Result / Verdict
The migration tools market is booming at the moment. There are plenty of tools available to assist migration, but there are still major challenges in all tested tools. Migration tools can assist you in your migration challenge but there is definitely a need for skilled experts who can analyze your current environment and propose a suitable approach for migration. In most cases, your preferred approach is re-factor / re-platform (especially if you have CI/CD pipeline already built) but the tooling is there to help you. Azure Database Migration Service (DMS) is a really powerful tool for migrating MS SQL servers to PaaS SQL.
Can we migrate your workloads and data also?
Solita offers comprehensive end-to-end migration services for all major public cloud platforms. We are able to assist you with all migration needs from refactoring to mass-migration.
Read more from our reference case:
Interested to learn more – please contact:
Head of Cloud Consulting