Shared Nothing Migration within VMTurbo 5.3

August 25th, 2015 by

VMTurbo’ s Journey

VMTurbo’s ability to control virtualized environments in a constant state of automated service assurance has been at the center of the platform’s value proposition since its inception. By creating an economic abstraction of the datacenter that automatically brokers workload demands to the underlying supplies of resources, VMTurbo enables organizations to assure application performance while simultaneously increasing efficiency across multiple layers of the IT Stack.

In past releases, VMTurbo expanded its core control into other facets of the IT Stack including capabilities around Hybrid Cloud, Application QoS adherence, and SDN to name a few. While each release has become a proof point in VMTurbo’s journey to run any workload anywhere, anytime; a key themes behind each release remains the idea of “workload mobility”.

In VMTurbo’s newly announced v5.3 release, VMTurbo Operations Manager now has the ability to leverage shared nothing migrations within vSphere environments to make intelligent placement decisions (svMotion and vMotion) within a single action execution.

What Is Shared Nothing Migration and Why Is It Important?

Introduced in vSphere 5.1, shared nothing migrations allow the user to vMotion and svMotion a virtual machine across environments that don’t have shared storage. This ability can be extremely useful in infrastructures that remain on local storage and still want to leverage the flexibility of migrating workloads.   It can also be useful in migrating workloads to logically separated environments during HW refresh, VC upgrades, and datacenter migrations. Additionally, with vSphere 6.0’s introduction of long distance vMotion capabilities and enhanced network support, VMTurbo is directly positioned to take advantage of what this has to offer.

vmotion-shared-nothing-migrationImage Source: http://blogs.vmware.com/vsphere/files/2015/07/vMotion-1.png

The Challenge

The challenge with leveraging shared nothing migrations is that the user must now determine which actions are the right actions to maintain performance. More importantly, this must be done continuously in real time as the state of the environment changes to prevent performance issues from taking place. Which VMs do we move and why? If we move the virtual machine from Local to Local storage, how do we understand the trade-off we are making on resources like CPU and Memory utilization on our hosts? What are the implications of VM movements and eventually allowing a software system to automate placement recommendations across disparate environments in a single action across hosts and data stores? The more boundaries (clustering, rules, and policies) that we remove in our environment, the greater the trade-offs become.

Let’s examine a few use cases where VMTurbo enables customers to leverage shared nothing migrations with confidence:

Local Storage to Local Storage

vmturbo-migrations

Above illustrates and example of VMTurbo recommending a shared-nothing migration between ESX5 and ESX6, both of which have local storage attached. VMTurbo is able to analyze all interdependencies; both compute and storage, to intelligently recommend which VMs will be the best to control the hosts in a desired state. For customers opting for local storage vs investing in a full-blown SAN, this allows for flexible workload automation in environments that don’t have shared storage.   At the same time, it is able to drive the environment into a state where local storage is being used to its maximum level of utilization and performance demands are being met for the application workload.

Local Storage to Shared Storage

vmturbo-move-vm Another potential use case would be a scenario where workloads are migrating from local to shared storage. This use case can be two-fold. For customers currently leveraging a mixed environment of local and shared storage, VMTurbo offers the ability to offload risk from local environments to available shared storage in the event of performance risk and can then make these suggestions in real time. In doing so, VMTurbo is able to understand the specific workload demands and which shared storage would be best suited for its application QoS requirements while respecting organization business constraints. Secondarily, VMTurbo can now allow organizations to plan and execute mass migrations from local to shared environment with the implementation of new SAN environments and drastically cut down on the time it takes to execute these projects without risk. Did we mention that these actions can be fully automated?

Converged Fabric – Nutanix

With converged infrastructures such as Nutanix increasing in adoption, shared nothing migration and intelligence in placement can be critical to assuring performance while utilization all supplies of resources.  Together, a group of Nutanix nodes forms a distributed platform called the Acropolis Distributed Storage Fabric (DSF).  DSF appears to the hypervisor like any centralized storage array, however all of the I/O operations are handled locally to provide the highest performance.  While this provides high levels of QOS assurance within node-clusters, VMTurbo’s direct integration into Nutanix architecture along with shared-nothing capabilities now expands VMTurbo QoS assurance capabilities allowing the platform to make VM placement decisions across Nutanix clusters in real time given changes in demand and resource usage. This can allow customers to be flexible across distributed Nutanix architectures.

VM Migrations-VC Upgrades, Cross-Cluster and Datacenter Migrations

vmturbo merge clustersSo you may be asking yourself: how do I control where shared-nothing migrations take place in my environment under VMTurbo’s control? Well you’re in luck. Using VMTurbo’s Policy view, users are able to designate “Merge” policies that tell the platform where shared-nothing migrations can take place. It’s as simple as selecting the clusters that you want to merge into a migration zone applying the policy, and letting VMTurbo’s decision engine determine the rest.

In the case of specific migration goals, users are also able to specify which workloads that they would like to migrate to destination environments during scenarios such as VC Upgrades, cluster-migrations, and Datacenter migrations. This can be done through “Workload placement” policies in the Policy view of VMTurbo’s interface. And because there is no longer a requirement for the environments to be connected via the same shared storage, VMTurbo can execute these recommendations during designated change controls so long as the network is in place to support it.

Shared nothing capabilities add yet another proof point on VMTurbo’s journey to run any workload, on any infrastructure anytime. Feel free to contact your VMTurbo account manager for more information on how to leverage this new feature.

Leave a Reply

Your email address will not be published. Required fields are marked *