With the release of latest vRealize Operations Manager 6.6 there are many new features which came along with it. Now customer can get more OOTB functionalities which were not available in its earlier releases. Some of the new features available in 6.6 version are:
- Simplified usability and faster time to value capabilities:
- New HTML5 user interface provides an easier and consistent experience.
- The Getting Started dashboard allows for quick navigation.
- Persona based dashboards provides for answers in one place. Dashboards are separated into categories such as Operations, Capacity and Utilization, Performance Troubleshooting, Workload Balance, and Configuration and Compliance.
- Out of the box integration with vSAN and vRealize Automation provides quick time to value.
- Added Native vSAN management capabilities:
- Allows for centralized management across stretched clusters.
- Ability for complete vSAN management, which includes administering performance, capacity, logs, and configuration and health.
- Fully Automated Workload Balancing:
- Ensures performance across the datacenter’s with fully automated workload balancing, across clusters and across data stores.
- Ensures DRS Configurations and provides the option to set DRS automation level for individual objects.
- Predictive DRS takes action to preempt resource contention.
- Utilizes operations analytics to optimize initial placement of workloads through vRealize Automation.
- Additional out-of-the-box compliance capabilities:
- Ability to tackle compliance problems through the new vSphere hardening dashboard
- Extends compliance through PCI & HIPAA compliance for vSphere.
- Ensures business configurations through new cluster, host, and VM configuration dashboards
Since it is really difficult to mention all the features in a single blog so here I’m picking up one of the most awaited feature – vROps Workload balance. It is now very much possible to edit vSphere DRS policies with the help of vROps and using its analytic engine intelligent placements can be made.
Before I discuss about vROps Workload Balance feature I would like to draw attention to this question – Why we need to Balance our Workload ? Why we need vROps to balance our resource workload ?
Over the years during evolution of vSphere customer have realized that while workload placement is made across the cluster the idea is to maintain optimum burden on each esxi host and if resource utilization of one host increases upto certain level the load should be balanced by another host of that cluster. We call it DRS (Distributed Resource Scheduler) and vCenter server is responsible for this intelligent placement.
But why vROps to deligate this responsibility – the answer is simple vROps has inbuilt analytic engine which runs multiple algorithms to decide which cluster is getting over utilized and what could be the best possible way to lower its burden. Also since it is tightly integrated with vCenter server hence the recommendations made are easily incorporated by vCenter server.
How to View & Edit DRS Settings: Login to vROps under Home tab Workload Balance can be found on left side. On clicking it we can see available Datacenter and its DRS status. It also gives information about vMotion occurred in last 24 hours. In this example we can see that DRS cluster state is Balanced that’s why options like Rebalance & Schedule are grayed out. However I’ll be discussing about these options as well in the course of this blog post.
DRS Settings & Current Workload gives the information about current cluster resource utilization and applicable DRS settings on that cluster.
Click on Set DRS Automation to jump into the wizard where DRS settings can be changes right from the vROps console.
Current DRS level is set to Fully Automated & Migration Threshold is Most Aggressive however it can be changed to desired setting right from this wizard
The best part is that you don’t have to login to vCenter (web) console to see the progress of this task as it can be viewed under Recent Task in vROps.
Manage DRS settings through vROps Primarily there are three DRS settings which can be set via vROps console: Consolidate Workload, Balanced Workloads & Cluster Headroom
Consolidate Workloads: It is basically for initial placement and it is an ON/Off setting. This is the only setting which is required to set while configuring vRealize Automation with vROps and doesn’t come into picture while configuring workload balance.
Balance Workload: This setting is used while balancing workload and has three stops Conservative, Moderate & Aggressive.
Conservative balancing will only be triggered if one of the underlying cluster is facing contention VS Aggressive which means DRS will try to it’s best to balance the cluster even for minimal deviation. Lastly Moderate is in between of those two extreme ends where DRS still try to balance cluster.
Cluster Headroom: This is the percentage value of resources (CPU, Memory, Disk) customer wants to keep free before it is declared Full. Setting this value 10% means that CPU, Memory or Disk resource can only be used upto 90% of their actual capacity and remaining 10% will be left un-usable & available for cluster headroom.
This setting is further derived from the Workload score Threshold (can be checked by editing active policy –> Analysis Settings) so it is recommended not to change this setting unless requested by VMware support to do so.
Please note that vROps workload balance will not move a virtual machine if it causes a cluster to either:
(i) Breach the cluster headroom threshold – which means if my cluster headroom is 10% and by moving a vm from another cluster to that cluster can reduce the headroom value to 9 or 8 which is less that 10% then vROps will not trigger that move
(ii) Cause the cluster workload to go non-Green – let’s assume there are 2 cluster one at 95% utilization and other at 70% utilization with cluster headroom set at 10% & workload threshold set at 80% now when workload balance will trigger it will try to balance both the cluster untill one of them turns GREEN. In below example as soon as Cluster B comes at 79% (which is close to 80% workload threshold) it will stop the balancing operation
Note: If both the cluster are non-GREEN in that case no move will be recommended
When is cluster Un-balance..? My cluster is considered un-balanced if any of the condition is TRUE
- Cluster utilization is such that it no longer fits into the balance workload policy – If there are 3 clusters Cluster1, Cluster2 & Cluster3 if their workload utilization is 70%, 40% & 10% respectively and DRS setting is set to Conservative (considering no contention) then the cluster is well-balanced but in the same example if DRS is set to Agrressive then it is considered as Un-balanced cluster
- Any cluster is non-GREEN – In a situation where one or more cluster or datcenter is non-GREEN
- Utilization of any cluster breaches Cluster Headroom – if cluster headroom threshold is crossed
If any of the above conditions is TRUE and cluster is considered Un-balanced and using vROps 6.6 OOTB features cluster balance operations are triggered which results into intelligent placements. With this release of vROps 8-concurrent placements can be triggered (earlier it was 1) which means in less time more number of placements can be made. Also before performing cluster re-balance operations vROps will show Before & After resource utilization so that customer can know what will be the utilization score after performing this operation even without performing actual re-balance operation.
Cluster re-balance operations can also be scheduled so that it doesn’t impact business operations and vROps will perform these placements on non-business hours.
Limitations of vROPs workload balance:
- Cross vCenter placements are snot supported untill this release
- Compute cluster needs to be on DRS mode Fully Automated
- All datastore cluster must be under SDRS (Storage DRS) in Fully Automated mode
- Source & Destination datastore must be under datastore cluster
- vSAN isn’t currently supported
- RDM is supported unless destination datastore cluster can access RDM LUN
- VMs running DRS or SDRS affinity rules are excluded
- VMs having multiple .vmdk spread across multiple SDRS cluster are excluded
As I said earlier, this blog is only concentrated on vROps – Workload Balance feature. I’ll be posting more stuff related to other OOTB features in later series… with this I end this blog… I hope it helps….. till then Happy Reading 🙂