Simplifying VM Hardware Migration (and why it just got much easier to adopt the latest data center infrastructure)

It’s becoming hard to ignore the business case for phasing out legacy data center infrastructure in favor of the latest generation of processors. 

The disadvantages of the status quo are now clear, as new, data-intensive workloads ramp up the pressure on existing infrastructure. To be successful with AI and machine learning, for example, most enterprises will need increased data center capacity. 

Without infrastructure that is more efficient, this can only lead to more servers to run an enterprise’s VMs. This is not a realistic option for organizations faced with maximized data center footprints. Even less so, given the additional power and cooling that would demand – and the additional cost of that energy and its associated carbon output. 

Even the most risk-averse IT manager accepts these are strong arguments. Nevertheless, for all the potential performance and efficiency improvements to gain by moving to servers with AMD 3rd and 4th Gen AMD EPYC™ processors, they would tend to urge caution when changing processor architectures because of the challenges of migration. VMs need to be migrated cold, which means rebooting servers, which means application downtime. 

Or so the thinking goes.

The New Tool From AMD and VMware That Changes Everything 

Reboots are an unavoidable feature of migrating VMs between hardware architectures, that’s a fact. 

But what’s also a fact is that reboots are standard practice when applying operating system, application and security patches. IT staff know these processes well, and because most organizations have highly available application environments with load balancers that are built for redundancy, a cold migration can be viewed as similar to routine patching work. 

The basic steps are: power down systems, perform operations and updates, power up systems, and verify completion and proper operation. The good news is this established workflow that IT professionals have used for years can be used to migrate virtual machines between hardware architectures. Applications are moved (failed over) to other instances during the migration process in the same way that applications are failed over during patching and upgrade processes. 

And the better news is that this process just got easier thanks to a new open-source tool created by engineers at AMD and VMware. VMware Architecture Migration Tool (VAMT) encapsulates those known best practices for patching and automates key stages in the plan and migrate workflow. It does all the heavy lifting to get VMs up and communicating, right up through the validate stage. 

Simply put: with VAMT your IT teams control when reboots occur using standard automation practices. Applications are then moved back at your convenience, after the migration process completes.

Explained: VAMT’s Key Features 

The migration process does not have to take long. In-depth tests conducted by Prowess Consulting show that engineers successfully cold-migrated 40 virtual machines from systems with Intel Xeon Scalable Processors to servers with AMD EPYC™ Processors in less than 30 minutes.¹

Key VAMT features that make this possible include: 

• Change window support, which provides the ability to stage which machines go first and which machines go second. This is a key feature that enables customers to perform an entire migration of their multi-tier architecture without experiencing any downtime. 
• The speed of VM migration to new CPU architectures always depends on the speed of available storage, network bandwidth, and how many VMs are being moved. VAMT’s throttling process, included as a configurable tab, allows customers to dial up or down on the VM migration velocity to fit their resources. This feature makes VAMT ideal for enterprise organizations and small-to-midsize businesses regardless of total resources.  
• Supports logging and reporting of processes. These logs and migration status reports can then be sent to an email address for easy access and review.  
• VM success validation ensures that VAMT verifies that VMware tools load successfully after migration is completed. This acts as an indicator of migration success.  
• Built in rollback scenarios allow VAMT to mitigate any risk of booting failure. This feature uses VMware Tools, installed on every VMware environment, to perform a validation that the OS boots, with automation hooks to allow end-users to insert their application specific validations to approve a successful migration. If it fails, it automatically executes a rollback to the snapshot and notifies the operator. 

One Less Obstacle to More Efficient Data Centers  

Live or cold, VM migrations often require similar tasks, depending on the features to be enabled within the VMs themselves. Enterprise IT teams have the talent to do this. What they often don’t have, however, is the time or resources. 

VAMT has taken all the hard work of cold VM migration and automated it, right up to the validation stage where your IT team’s deep application knowledge and experience adds the most value. It also saves enterprises the hassle and expense of hiring a professional services team or consultant, building their own equivalent tools, or standing up a “second stream of infrastructure”. 

Ultimately, VAMT means there’s now one less obstacle between businesses and the adoption of the data center transformations they need to innovate faster, drive efficiency, and lower operational costs. That’s timely when increased server performance is needed to keep up with enterprise compute demands, for AI and elsewhere. 

For more on VAMT, read the white paper.