Eco4Cloud Workload Consolidation vs. Red Hat OpenShift Virtualization

Eco4Cloud is well suited in VMware farms that have standard VM deployments, especially with small VMs, with nothing blocking vMotion and hardware fully compatible with IPMI 2.0 (minimum). It doesn't work well if many VMs are nodes in a Microsoft cluster because they would be unmovable and therefore their host would never be put in stand-by mode. If every host in the cluster has at least one guest VM that is a node of a Microsoft cluster, no host in the cluster will ever be put in stand-by mode.

I believe that Red Hat OpenShift Virtualization is a very promising option for hypervisor at the moment and it seems to be improving quickly. Red Hat is doing a great job working with customers on feature set requests and improving UX and stability

Incentivized

• It consolidates workloads in VMware clusters using an algorithm that seems to work better than VMware DRS. It keeps headroom for workload peaks and to withstand the number of host failures set for the cluster in vCenter.
• If the workload is very variable, Eco4Cloud will not continuously turn a host on and off but it will settle to keep the host running. It sort of "rightsizes" the cluster keeping only the right number of hosts running. This is another main difference when comparing Eco4Cloud with VMware DRS/DPM which strictly follows the workload and sometimes isn't fast enough in turning on the hosts when the workload is ramping-up.
• To do a good job Eco4Cloud needs to analyze the hosts and the VMs. Therefore, the troubleshooting component of Eco4Cloud can figure out if there are problems in the configuration of the hosts (e.g. vMotion settings). It can also determine if a VM is stuck and can't vMotion (e.g. using the physical CD-ROM) or if it's oversized or undersized.
• Something Eco4Cloud has and others don't is a technology named Smart Ballooning. With Smart Ballooning Eco4Cloud recovers unused RAM from the VMs and gives it back to the host way before the host goes over the 94% RAM threshold. This allows us to reach higher levels of overbooking.

• Easiness of deployment, we can deploy the full stack completely and hardened easier than other container products
• Used to host different workloads (CNF and VNF).
• Easy to scale on both the worker and the infra layer
• Infra nodes do not require subscriptions, which means we are only paying for our workload only, and no hidden costs.

Incentivized

• It would be useful to have a version that works with OpenStack.
• It should be made compatible with VMware VSAN.
• It would be useful to have a version that works with hybrid clouds using the new "vMotion anywhere" or share nothing vMotion feature.

• Because of how different virtualization is done, there is a learning curve to overcome when switching to it from another hypervisor.
• Documentation needs to be improved. Just finding the details to set it up is a burden that will keep some people using it.

Incentivized

OpenShift is really easy of use through its management console. OpenShift gives a very large flexibility through many inbuilt functionalities, all gathered in the same place (it's a very convenient tool to learn DevOps technics hands on) OpenShift is an ideal integrated development / deployment platform for containers

Incentivized

The product is maturing at an amazing rate, but there is still work to do to reach existing VMware customers. Most notably are the HA/DR features, and backup/recovery. We need to have faster backup capabilities, and I know this is in the works with change block tracking - but it is a challenge today. Some users are also scared by the number of volumes created by Red Hat OpenShift Virtualization, as opposed to the number of VMFS volumes they are used to.

Incentivized

We have already hab cases in which our customers have a better application performance even in legacy architectures running on Jboss, and even more on cloud native architectures. Customers who hosted their applications on Jboss and wildfly, when running on top of Red Hat OpenShift could benefit from more stability and availability, and in some cases, gains of performance.

Incentivized

Speedy support

Incentivized

The learning curve is quite high but worth it.

Incentivized

We compared the consolidation feature and we obtained exactly the same amount of servers in stand-by. The main difference was that Eco4Cloud turned off the unneeded hosts in 2 days while VMTurbo Operations Manager required one day per host. This behavior is caused by the different algorithms implemented by these two products. Compared to Eco4Cloud, VMware DRS/DPM was very aggressive (we tried it in vSphere 5.0). We haven't tested it with vSphere 6 yet.

Some of the various other Kubernetes out there, including Rancher, is probably one of the biggest one. Rancher has a reputation for being very easy to implement. I've found that Red Hat OpenShift is just as easy to implement as Rancher. People just don't really know about it. Nutanix is another hypervisor out there competing with Red Hat OpenShift. Again, I think OpenShift virtualization but maybe isn't as old as Nutanix. I do feel it's more mature.

Incentivized

• Eco4Cloud causes a reduction in power consumption both on old VMware clusters and especially on new ones.
• It also causes a delay in CapEx investments because you buy more hosts only when really needed.

• It has allowed us to see where we need to be in the container world. I'm going to call it a net neutral impact, not negative or positive. It has given us a sense of what we are ready for and what we're not ready for. You know where you stand.
• You don't know what you don't know, so it helps us know what we want to know.

Incentivized