The SDDC, as theorized by VMware is going, from a technical side, towards automation and industrialization. One of the major tool to achieve this goal for customers is vROps.
vROps 6.0 significantly improved the automation level one can easily achieve with improved dashboards to understand better its environment and application, giving the ability to control deeper and easier the way everything is going and finally allowing automated action (even limited, the feature is paving the way for the future for sure).
The 6.1 release gave the ability to have a single place to monitor both vSphere and VMs, not only outside the VMs but also withing the VM, by merging Hyperic and its agents within vROps.
To go further with automation and giving the infrastructure manager, architects and, in fine, the CIO the ability to optimize its datacenter and its associated costs, vROps 6.2 improves the Workload placement by monitoring resources across clusters and provide recommandations of Workload placement AND workload migration.
So, what’s the goal ? What could I achieve ? What’s in it for me ?
Currently, on the first hand, vROps monitors the platform to understand how it behaves and identify, over sized VMs, undersized VMs, provide guidance on how clusters are utilized, …
On the other hand, DRS is closely monitoring the resources used on the cluster and on each host to ensure every VM has the exact resources it needs.
vROps 6.2 is here to integrate the way DRS works, but across clusters. Let’s imagine you have a VM that is constrained for Memory on its current hosts and DRS cannot migrate it on any other hosts within the cluster efficiently. vROps can now calculate a better placement for this VM and allow the vCenter to migrate from its cluster to another one. Before doing so, not only does it calculates a potential better placement but it also compute more than 100 checks.
To make a recommandation, vROps also includes vSphere constraints : Resources pools, DPM, Endpoints agents, Admissions controls, Reservations, …
After the recommendation is made and validated by vROps, it can launch automatically this migration to ensure the workloads are allowed to run at the maximum potential and log a reason for this recommendation (CPU or RAM balance).
And what’s the relations between vROps 6.2 new placement feature and DRS ?
Well, vROps 6.2 determines the appropriates cluster to place a VM, should it be a new one or an existing one that is currently suffering from constraints resources.
After it generates and launch a migration order, vROps hands over to DRS, through the DRS API, to place the VM within a new cluster. DRS determines the appropriate host to place the VM. Thus, it needs DRS to be configured on the “Fully Automated” level. The VM is migrated and the world (at least, your vWorld) is better now !
Some important facts from the dark side
It is important to note that vROps will never recommend and/or automatically process a migration across vSphere Datacenters or vCenters. It won’t also process a Datastore migration nor it will affect a VM configured in an active DRS affinity rule and
After the “don’t’s”, it is also important to note that vROps 6.2 Workload placement algorithm takes in account the concept of Customer Datacenter. Thus, you could consider a CDC composed of specific clusters (database or any other criteria) and vROps 6.2 will consider this entity to move the workloads across the members of this CDC. Thus, you could stay compliant against any policy you might have set or match performances expectations over a specific subset of clusters and VMs. (Remember that no migration across DCs or vCenters will be recommended, even if CDC spans across DC or vCenters).
I trust this feature is really valuable for vROps users. It delivers an improved way to balance workloads across the datacenter and optimizes the resources available. It also allows the CIO that every workload has the resource it deserves to provide the end user with the highest quality each user could expect.
And to ensure that these resources are efficiently used, not only does it automatically balances workloads across clusters, but it provides new dashboards to help monitoring that every resources is used efficiently.
As a final note, vROps upgrades process is really easy to upgrade : download the .pak, upload it towards the master node of your infrastructure and enjoy !
Any advises ??
Well, vROps is a very rare tool that could be used by many different kind of users from my experience :
- System Engineer for the virtualization platform, Infrastructure Manager : troubleshoot, deep dive on performance considerations, …
- System Engineer tasked with managing the OS : control the behaviour of their VMs, troubleshoot their VMs (with Endpoint agents or not)
- Production director : overview of the available capacity, optimize the delivery of service, SLAs
- Many other stakeholders
Obviously, each kind of users deserves its own set of dashboard and will use the product differently. I would recommend having some workshop to set each users expectations. And, release after release, after integrating more and more management packs, users are able to get more and more information on their infrastructure (physical or virtual BTW).