Hey all,
I am trying to do a Hypervisor Cluster comparison between AHV and ESXi.
To kick this off, does AHV have a way to load balance the VM’s and move the load to other hosts / nodes with less usage as DRS does for ESXi?
Thanks
Hey all,
I am trying to do a Hypervisor Cluster comparison between AHV and ESXi.
To kick this off, does AHV have a way to load balance the VM’s and move the load to other hosts / nodes with less usage as DRS does for ESXi?
Thanks
yes AOS Dtynamic Scheduler takes care of this but it works slight different (better)t to DRS.
From the Nutanix Bilble
A different approach towards resource optimization
When you look at existing scheduling / optimization platforms (VMware DRS, Microsoft PRO) they are all focused on balancing workloads / VMs evenly across cluster resources. NOTE: how aggressively it tries to eliminate skew is determined by the balancing configuration (e.g. manual -> none, conservative -> some, aggressive -> more).
For example, say we had 3 hosts in a cluster, each of which is utilized 50%, 5%, 5% respectively. Typical solutions would try to re-balance workloads to get each hosts utilization ~20%. But why?
What we're really trying to do is eliminate / negate any contention for resources, not eliminate skew. Unless there is contention for resources there is no positive gain from "balancing" workloads. In fact by forcing unnecessary movement we cause additional requisite work (e.g. memory transfer, cache re-localization, etc.), all of which consumes resources.
The AOS Dynamic Scheduler does just this, it will only invoke workload movement if there is expected contention for resources, not because of skew. NOTE: DSF works in a different way and works to ensure uniform distribution of data throughout the cluster to eliminate hot spots and speed up rebuilds. To learn more of DSF, check out the 'disk balancing' section.
At power-on ADS will balance VM initial placement throughout the cluster.
yes AOS Dtynamic Scheduler takes care of this but it works slight different (better)t to DRS.
From the Nutanix Bilble
A different approach towards resource optimization
When you look at existing scheduling / optimization platforms (VMware DRS, Microsoft PRO) they are all focused on balancing workloads / VMs evenly across cluster resources. NOTE: how aggressively it tries to eliminate skew is determined by the balancing configuration (e.g. manual -> none, conservative -> some, aggressive -> more).
For example, say we had 3 hosts in a cluster, each of which is utilized 50%, 5%, 5% respectively. Typical solutions would try to re-balance workloads to get each hosts utilization ~20%. But why?
What we're really trying to do is eliminate / negate any contention for resources, not eliminate skew. Unless there is contention for resources there is no positive gain from "balancing" workloads. In fact by forcing unnecessary movement we cause additional requisite work (e.g. memory transfer, cache re-localization, etc.), all of which consumes resources.
The AOS Dynamic Scheduler does just this, it will only invoke workload movement if there is expected contention for resources, not because of skew. NOTE: DSF works in a different way and works to ensure uniform distribution of data throughout the cluster to eliminate hot spots and speed up rebuilds. To learn more of DSF, check out the 'disk balancing' section.
At power-on ADS will balance VM initial placement throughout the cluster.
Thanks this is great! I found it in the bible and will dive in for more information!
Right the key for Dynamic LB of the workloads is only to kick in when there is contention that's the goal to avoid any impact or stun on the workloads for sure.
Thanks!
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.