From StorPool, Migration From Traditional Hypervisors to KVM

From traditional software ‘Stack to New-Age’ IT with a migration to KVM.

For the last few years, companies have witnessed the rising popularity of open-source (OSS) technologies driven by large- and hyper-scale deployments, the cloud-native movement and the pursuit to make business operations more efficient. These are some of the tools and techniques used to implement the ‘digital transformation’ everyone is talking about. And it is all driven by the changing nature of business, where value delivery needs to be faster, more efficient, 24/7, easily accessible online and thus re-invented by and with technology.

The current reality requires any company to innovate and adopt modern IT, which is resilient, automated, scalable, cost-efficient and vendor-independent. Migration to a New-Age IT stack will help your company be more agile and prepared for the changes of today and tomorrow. This is best achieved by the adoption of intelligent software, thus we witness the rise of software-defined technologies and datacenters, like SDDC, SDN, SDS.

This process inevitably means that the old, traditional ways of designing and building IT systems are being replaced with latest-gen designs, tools, and systems, which deliver what the businesses of tomorrow will need. And since most workloads today are virtualized, the starting piece is the Hypervisor. I.e. switching from a traditional hypervisor to KVM is one of the first steps for successfully adopting New-Age IT.

Some of the main challenges in the migration process is: how to do it and do it properly, so as to reduce risk and reduce downtime.

StorPool Ltd. has developed a step-by-step guide on easily migrating VMs from traditional hypervisors like VMware, Hyper-V and XenServer to KVM.

The document is targeted to help senior technical staff, who are building and managing infrastructure and the applications running on top of this infrastructure.

The proposed method can reduce the downtime in many cases to less than a minute, even when large disk images have to be copied over low-bandwidth links. This can be achieved by altering the workflow as follows:

Convert VM metadata and define a new VM at the target hypervisor
Take a snapshot of the VM disks on the source hypervisor
Copy the snapshot to the target hypervisor
Convert the VM disk image format
Take a second snapshot, if needed, and copy the latest changes to the target hypervisor
Apply the changes to the target image.
Stop the VM at the source hypervisor
Copy the data changed since the last snapshot to the target hypervisor
Apply the changes to the target image.
OS morphing
Start the VM at the target hypervisor

While the overall number of steps has been increased, the downtime is reduced by limiting the time when the VM is non-operational to only three short steps, while the most time-consuming steps – copying large amounts of data are executed outside of this critical time frame.

The presented techniques are successfully used to migrate CentOS and Windows Server VMs from VMWare ESXi, Hyper-V and Xenserver to KVM, but it can be used with other guest OSs and other types of hypervisors.

Full paper on migrating from VMware/Hyper-V /XenServer to KVM (requires an email).

Read also:
StorPool V19 Resolves Challenges of Large-Scale Data Center Storage
Features:multi-stack deployments support, lower latency for NVMe storage, and improved bare-metal Kubernetes support
December 26, 2019 | Press Release
StorPool With Record of 13.8 Million IO/s
Performance test report with NVMe-class storage performance in VMs
May 7, 2019 | Press Release