Chris Wolf’s Post

Broadcom, NVIDIA, and Dell power the magic of virtualized ML/AI and deliver near or better than bare metal performance, as shown with MLPerf Inference 4.0

One of the features of vCenter is the ability to display various performance metrics of the virtual machines and hosts, such as CPU, memory, disk, and network usage. The memory graph parameters that vCenter shows are:   ⚜ Memory Active: This is the amount of host memory that is actively used by the virtual machine for its guest operating system and applications. It does not include memory that is cached or idle. This parameter can help users estimate the actual memory demand of the virtual machine and adjust its memory size accordingly.   ⚜ Memory Consumed: This is the amount of host memory that is allocated to the virtual machine, including the memory overhead. It does not reflect the memory savings from techniques such as transparent page sharing or memory compression. This parameter can help users monitor the memory pressure on the host and the potential for memory reclamation.   ⚜ Memory Usage: This is the percentage of host memory that is consumed by the virtual machine. It is calculated by dividing the memory consumed by the total host memory. This parameter can help users compare the relative memory consumption of different virtual machines and hosts.   ⚜ Memory Balloon: This is the amount of host memory that is reclaimed from the virtual machine by using the balloon driver, which is part of VMware Tools. The balloon driver inflates the memory usage of the guest operating system, forcing it to swap out some memory pages to the disk. This frees up host memory for other virtual machines or host services. This parameter can help users identify the virtual machines that are experiencing memory contention and the effectiveness of the balloon driver.   ⚜ Memory Swap In/Out: These are the rates of host memory pages that are swapped in or out from the disk for the virtual machine. Swapping is a last resort technique to reclaim memory when the host is running low on memory and the balloon driver is not sufficient. Swapping can degrade the performance of the virtual machine and the host. These parameters can help users diagnose the root cause of memory swapping and take corrective actions.     ✍ To summarize, vCenter provides various memory graph parameters that can help users monitor and manage the memory performance of their virtual machines and hosts. By understanding what these parameters mean and how they relate to each other, users can optimize their virtualization environment and ensure the best possible performance and efficiency. 😊   [ #vmware #vcenter #graph #networkengineer ]

This is a hot topic especially with AI technlogy getting integrated into the data center space. #CPI #datacenters

Today marks a significant milestone as we officially launch our Virtualization Ready Modules with integrated hypervisor! This is the first step towards Solutions and Application Ready Modules, solidifying our commitment to innovation in meeting customer needs. Key benefits include cost savings, flexibility, security with hardware isolated partitions, and faster time to market. The hypervisor is seamlessly integrated into the firmware of all new x86 platforms, making virtualization more accessible. Our website and press releases are ready for the official launch. Stay tuned for updates and press coverage. #hypervisor #congatec #hypervisoronmodules #virtualization

Amazing how time flies - it is now 100 days for me @congatec.   While I am far from knowing everything our company is about I can draw a first resume: an outstanding team and brand working with a leading edge customer base. In short - an amazing workplace in the "Bavarian Silicon Valley". Being deeply routed in Germany's technical heritage and embracing global (business) culture is forming a strong base allowing us to be a leader for computing in the quickly expanding Embedded market place. We are about to re-invent computing on embedded modules and the launch of our #hypervisor on COM (see below) is just the starting point of much more to come - stay tuned. And this is what I like about being part of #teamcongatec: staying focused on simplifying the use of Embedded Technology while embracing new technology as it becomes applicable - to enable better devices for our customers who drive innovation globally in so many ways. And these ways are the reason why I like to work as an embedded technology provider: so many insights (& contributions) in various application areas supporting our society - be it leading edge medical devices, advanced automation to optimize assembly or traffic, enhancing efficiency of power generation or distribution, control of agriculture vehicles, charging electrical cars, automating (plastic) bottle re-collection or "just" parking fee retrieval and so much more. For all of that we do provide computing and in my past 100days I got new stunning insights in what technology can do. Thank you #embedded market for having me, thank you - our customers - for working with us and thank you #teamcongatec for getting me on board!

1,000W per GPU, and these things are normally clustered 8 to a server, 4X rack ++ so that is in the area of 40,000W per rack, and then there are backup power supplies (hot swappable), then UPS and all the other overhead. I wonder how many data centers can accommodate this load?

This could be a potential gamechanger for simplifying work load consolidation in the embedded space. #congatec #hypervisor #virtualization #embedded

Is this a technical post? You bet it is! Why would I need to know about memory tiering? "...memory capacity (and performance) has not grown at the same rate as CPU capacity..." OK, how does memory tiering help? Similar to caching in storage, you can have a small amount of high performant memory ready for high leverage situations? Like what? vMotions on VMs with large memory footprints, it can make them faster and less disruptive. Any closing thoughts? This MIGHT be the first positive sign coming out of the VMware by Broadcom acquisition. This seems to be Broadcom tech that can be leveraged by VMware. https://lnkd.in/gN8mY9-a

The pressure is on for companies to consolidate infrastructure, free up data center space, and enhance their competitive advantage. Read my take on how AMD EPYC server processors can help data centers deliver the results they need with the efficiency they require:  https://lnkd.in/gawrhKyS

General availability of Azure HBv4 and HX-series VMs, powered by these new AMD 3D V-Cache processors. Organizations can now solve some of the most compute intensive challenges faster than ever. For example, top of stack 4th Gen AMD EPYC processors with 3D V-Cache Technology can deliver up to 2.6x design jobs per day across top CFD workloads versus top of stack Intel Xeon CPUs. #amd #epyc #togetherweadvance