The last couple of weeks I’ve been busy with a couple vR Ops designs and implementation in very different environments, and the question I get a lot is what the differences are between vCOPS and vR Ops. First of all I must point at the naming difference where vR stands for v Realize and Operation manager has become a part of much larger suite. A suite that will give you the opportunity to leverage, monitor, automate and build hybrid cloud environments.
Back to the question:
The vR Ops architecture consists of 1 Virtual Machines (VM) that works on a scale out basis, which differs from ealier version that consisted of a vApp with 2 VM’s and which was based on a Scale-up architecture. You’ll get a better picture looking at figure 1 and reading the information below.
As shown in the figure above, the deployment of vR Ops starts with a single VM (which will become the Master Node) and can easily be scaled out with additional nodes (which can be data nodes or remote collectors). To provide HA ,a master node can have a replica node (holding the same data as the MasterNode) which will take over if the master node fails. see the figure below for more information.
The Master node as well as a replica node holds the Global xDB and is responsible for collecting data from the vCenter Server, other vR Ops suite product and 3rd party data sources (metrics, topology and change events) and storing that raw data in its scalable File System Database (FSDB).
I’ll dive into other differences and more in depth posts in a later stage, but for now I just wanted to get this information out 😉
During the beta fase of vSphere 6 I was looking for a couple of workarounds for problems during installation process in my homelab. One of those problems is that (as with vSphere 5.5) certain not supported hardware are the onboard realtek NICs on the cheaper “homelab” motherboards. During this search I came across this workaround (login needed) by Andreas Peetz explaining a method to install the drivers onto the vSphere host(s) in your environment. Thanks Andreas, it worked well for me!!
here is a workaround for you … I have created a package that includes the original VMware net-r8168, net-r8169, net-sky2 and net-s2io drivers and uses the name (net51-drivers), and published it on my V-Front Online Depot.
If your host is already installed and has a direct Internet connection then you can install it from an ESXi shell by running the following commands:
esxcli software acceptance set –level=CommunitySupported
esxcli network firewall ruleset set -e true -r httpClient
esxcli software vib install -n net51-drivers -d http://vibsdepot.v-front.de
As you can see I had to add the –no-sig-check to install the vib. It might be this is not needed in your situation, though.
If you need the VIB file for injecting into an installation ISO or offline installation then you can download it from
(right-click the above link and save as …)
I have not yet tested this myself, and of course this is completely unsupported by VMware! Use at your own risk, but – honestly – I expect it to just work …
As Andreas points out this is totally unsupported by VMware! But hey, it’s your homelab so probably most of it is unsupported ;-P
The other thing to notice is the awesomeness of the VMware community! Always helping one another and making sure people are able to get hands on experience with the little resources they have at home.
After an upgrade to our HP BL460c Gen8 blades we had multiple blades showing a Purple Screen of Death (See screenshot)
In our environment we have multiple DC’s as well as multiple versions of the HP BL series blades. Multiple adminstrators across multiple DC’s experienced the same problem. The problem only seems to affect the latest HP BL460c Gen8 v2 series (as far as we can tell), but it might be other vendors with Intel Haswell Xeon v2 CPU’s might be affected as well (please let me know if you’re having the same problems on other hardware).
We decided to roll the affected servers back to 5.5 U1 and wait for a solution but Frank Büchsel (@fbuechsel) of fbuechsel.eu provided me with following workaround. In the BIOS when you turn of the Intel VT-D option, your server will start without trouble:
VMware is working hard to process the logfiles and making sure this problem is solved ones and for all! Keep you updated on the status. If you have a solution or questions/answers please give us a tweet or something!
After a long and extensive investigation we’re able to say the problem in our environment has been solved. After checking all Firmware we reached the point where the only difference between a purple screen and a stable running ESXi 5.5 U2 host was the firmware of the HP FlexFabrics (Emulex) in the server. When we used the 4.6 (or lower) firmware we had Purple screens, but after updating to the latest 4.9 firmware the Purple screens were gone, and the servers ran stable.
VMware is still examaning the logfiles, but htis seems to solve our problem for now! If you experience any problems with this, please let me know…
There’s always a lot of gossip around VMware Partner EXchange (PEX), and it’s not any different during this years edition. It all started about 2 weeks ago when Joe Kovar at CRN posted: VMware Limits Some Storage Vendors Presence At VMware PEX. In the post Joe told us that VMware asked Nutanix and Veeam were asked to stay away from PEX, which VMware hosts this week in San Francisco. My good friend Hans de Leenheer did another post on this: VMware uses its right to bear arms! where the replies are of most interest to me… I’m not choosing sites, but Nutanix and Veeam not being present on this Partner Event is very strange, but as always there are most likely two sites of the story…
During the event PernixData their Flash Virtualization Platform (FVP) 1.5 software, which includes support for vSphere 5.5, the vSphere webclient and as a result they now support the whole VMware vSphere 5.x family. The way PernixData describes the 1.5 release is:
Robust vSphere Support: With FVP supporting VMware vSphere 5.0 to 5.5 now, a customer has the oppertunity to choose the VMware release of their choice.
Seamless management through vSphere Web Client: With VMware moving more and towards the Web Client and away from the vCenter Client this move by PernixData is great. Being able to manage you’re complete vSphere environment including the PernixData through the Web Client is a great achievement.
Seamless deployment: Installing FVP within ESXi can be done quickly (within 20 minutes) without changes and reboots needed to the VM’s, the hosts or primary storage.
Clustered platform compatible with all VMware Operations: It’s important if you’re implementing new software to make sure your normal operation will not be compromised. PernixData FVP 1.5 uses it’s patent pending Flash Cluster Technology to support all VMware operations and products like vMotion, HA, DRS, VDP, SRM, Horizon View and vCloud Director. Making sure you can keep on using VMware vSphere, but in a way you never experienced before.
For a great post on the PernixData FVP 1.5 release please read this post by Chris Wahl: Man Alive! PernixData Unleashes FVP 1.5
Atlantis Computing announcement
Another big announcement was done by Atlantis Computing during PEX. Atlantis announced their ILIO USX software. Atlantis released the ILIO software a couple of years ago. With the ILIO software Atlantis uses the server RAM to build an in memory storage solution providing Real-Time Compression, Inline Deduplication, Thin provisioning and Fast Clone solutions. Let’s dive a little deeper into the ILIO USX solution
As a VMware consultant I see a lot of customers struggle with the rapid growth of their virtual environments. While more and more application being virtualized, a lot of the customers are struggling with the right buck for the money. Not all environments are greenfield (duh…) and investments made in the near past need to be taken in account when making sure that an enviroment performs the way it should. As already said not everything is greenfield, but in some cases building a new environment along side the “old” one provides the oppertunity to make use of new players on the market like Nutanix and Simplivity. But wait a second, if they can build this, wouldn’t it be possible to this myself also? Yes, with Atlantis ILIO USX you can.
Using their unique In-Memory Storage Technology Atlantis is able to create an In-Memory very high performance solution, but still leverage the underlying storage solution a customer already uses. With the need of an NFS share for the front end, one might think it might need a very large NFS share and let the tiering be done by the storage solution, with USX you have the possibilty to choose the need for storage is seperated, or as Atlantis calls it: “Application Defined Storage Volumes”:
The release is already available, but one must remember that this is a 1.0 release. There are multiple improvements coming in the near future, including :
QOS – Quality Of Service (A very cool feature if you ask me)
There still are two things that are really cool with ILIO USX:
Working alongside VMware Virtual SAN. Where most “caching solution” vendors are looking at ways to talk bad about VMware’s Virtual SAN whereever possible, Atlantis ILIO USX proves it can work with VMware Virtual SAN just fine. More information can be found here: VMware Virtual SAN with Atlantis ILIO USX
Working with IBM (the new IBM X6 series) that supports the Diablo UltraDIIMM technology shows the great possiblilties Atlantis ILIO USX offers. With Flash technology provided to leverage the memory channel interface lateny can be in nano seconds in stead of microseconds. And not have to worry of dataloss in case of a system outage…
Awesome annoucements if you ask me, and more then worthy of reading more on this. So here are a couple of posts by my fellow bloggers:
Marcel van den Berg: Atlantis releases Unified Software Defined Storage ILIO USX
Marco Broeken: Atlantis ILIO USX, In-Memory Performance for Servers
Andrea Mauro: Atlantis ILIO USX – Unified Software-defined Storage
Atlantis Computing: ILIO USX page
After re-installation of my homelab, I decided it was time to use the VMware vCSA instead of the Windows based vCenter Server. Doing some testing on Infinio and , I wanted to install some windows XP machines. After creating a XP template I wanted to use the wizard to customize the XP installations but got the following message:
In the old days this was rather easy to solve, but with the VMware vCSA appliance I wasn’t so sure… During my google search, the first I came across was this article this article by Hugo Phan, and this solved the problem.
But then when I logged into the mgmt interface of the vCSA https://FQDN-vCSA:5480 I noticed there is another way also. Let me explain:
First log on to your vCSA. When logged in, on the first window (vCenter server – summary) you’ll find sysprep files and the upload button:
In the sysprep files upload window first choose the windows OS of choice, and then choose files:
Make sure all files from the deploy.cab are uploaded to the sysprep folder on the vCSA, and click close:
In the post from hugo you can find the links for the OS’es that need the sysprep files (from windows Vista and newer OS’es sysprep is build in), so make sure you’ll read that post as well.
During VMworld, and in the months prior for the event the buzzword seems to be Software Defined (Everything). And although I think highly of innovative and perfectly developed software, hardware is still the driver behind the software force. Without hardware, there would be no software, and because of the innovations at the hardware level, software is able to do it’s awesomeness these days. So naming it software defined is a bit stupid IMHO!
As said, I think highly of great software. But when you think of it lots of the software houses that developed software for many years just used the innovation of hardware in a boring way. Using the extra resources the hardware would (or could) offer is what a lot of these software houses did instead of doing the same innovation in their software. as the hardware vendors did. It made developers lazy. So a revolution in Software development is needed, but giving all credits to software isn’t.
With virtualization of the X86 hardware (with VMware as the main driver behind this force) the hardware vendors (Like Intel and AMD) developed more and more cool features on the hardware side that can be leveraged by software. After server virtualization we started virtualizing the storage and now we start virtualizing the network. This are all awesome achievements and I hope to see much more on these great technologies.
Did a server (before Software Defined) do anything without software installed on it? Or a switch or router? Have you ever seen a NAS or a SAN perform without software? So why does it now all of a sudden has to be named software defined? Hardware needs Software, as Software needs Hardware, so let’s rename it to something both of these awesome technologies are equally represented 😉
Lots of questions. Do I have the answers? not really, but I guess the most of the renaming and rebranding has to do with Marketing. Renaming and creating buzzwords sells. So doing renamed technologies that already existed is all about making decision makers drool, and buy the new software (and hardware) products. It’s all about selling, and the developers just keep on creating the awesome software they are creating as well as the hardware vendors will keep on creating incredible cool hardware.