This user hasn't shared any biographical information
Posts by Richard Garsthagen
Over the years I have been getting frequent request to the source code of the VMware scanner. So I decided to re-write it for python (version 3) and make it available. It actually handles the multi tasking better and is faster scanning.
It still shocks me how many VMware servers are directly connected to the internet, including some seriously old versions.
Here the script on github: https://github.com/AnykeyNL/vmware_scanner
Like many of you, I have an adorable home lab with a few servers, networking and storage. For my storage I have been using for many years now QNAP devices. Currently I have an TS859pro, which is an older 8 bay model with 8x2TB disks in them running raid6. In general I am happy with my QNAP, but when doing tests with virtualization, especially when you start cloning stuff, you need to be a patient person (which I am not!).
I already recently added SSD disks into my ESXi servers, so for my important VMs I get some solid disk performance (what a difference between using 1Gb iSCSI!!).
This week I also found out that QNAP has not been sitting still. With the release of their latest firmware 3.8.2 they now support VAAI (vStorage APIs for Array Integration) and while my model is End-of-Life, QNAP did also make this firmware available for my unit. Besides VAAI, Qnap also ships an plugin for vSphere, so you can directly in your vSphere client create datastores for iSCSI and NFS. While that is nice, the VAAI stuff is really the useful stuff.
I assume that most of you, like myself, just run 1Gb ethernet. This typically is by far the bottleneck for all my I/O. I do use dedicated NICs in the ESXi servers and the 2nd ethernet interface on my QNAP for iSCSI only, but still doing operation like clone/copying VMs take forever.
So I was brave and installed the new Qnap firmware today. Without doing anything else, suddenly the vSphere client was saying that Hardware Acceleration was enabled on my iSCSI datastores. So I took my simple WinXP template (3GB actual data) and tried to “create new VM from template”. In the meantime I was monitoring the traffic monitor on my Qnap. I saw almost no traffic at all and 133 seconds later the new VM was created. This is significantly faster compared to not having VAAI (234 seconds).
Network traffic with VAAI
I did some more test and strangle found that cloning from the same template that was on the NFS of the QNAP to an iSCSI datastore on the same QNAP it only took 104 seconds. I was using ethernet 1 for NFS and ethernet 2 for iSCSI, but find it strange that this copying goes faster then internally just copying the file using VAAI. Clearly QNAP can do some more optimization here.
But for a “home” nas I am really impressed with the QNAPs added functionality. They use a single software code for all their models, so even their smaller unit will have this capability (while not officially supported).
I finally got around working on a new version of vAudit. If you do not know what vAudit it, it is a monitoring / auditing tool for VMware View/Horizon environments. The standard logs from VMware View are useless as it logs actions, but not really sessions. For each session there is somewhere a log entry of connecting and later, at some point, a seperate log entry of a disconnect or logoff.
Also VMware View does not log any information about the client that users are using. This information is only available inside the desktop VMs.
So what does the new vAudit do?
Well it monitors the VMware view event log. When it sees when a new connection to a desktop is made, it will use WMI to remotely read out the view client information like client IP, protocol used and Device type (mac, win, ios, etc). As the program uses remote WMI, no extra client software is needed inside the desktop VMs. This information is then stored in a new sessionlog table (automatically created in the view event server database).
When a session disconnect/logoff is detected, vAudit finds the matching record of the login and will update this with the logoff/disconnect information (time and the way the session ended, status 2 is disconnect and status 3 is logoff).
The result… A very simple and easy readable log file, where each entry is all the information of the session; login time, logoff time, client information, pool id, desktop id, etc
Now having this information you can easily create charts and much more.
The charting is not finished yet, but I am releasing the log monitoring engine. It does not run as a windows service yet (the final version will), but if you want to test it out (and provide me with feedback) you can now download the vAudit test enigine in the download section.
Important to know: vAudit does NOT edit any of the official view database tables, it only creates an extra table (sessionlog). it only communicates to the event server (mssql) and using VMI the desktop VMs. It does not touch the actual connection servers.
I know most of you are VMware Fans out there, I do like to use vSphere from time to time as well But working now for a bit more then a year at Oracle I have to say, some very exciting things can happen when you marry great management software, virtualization, database and middleware solutions together.
I am a big Cloud Computing fan (understatement) and for me cloud is NOT about “The Cloud” but about changing the way we deliver IT services to our users. Not complicated and slow, but easy to consume and quickly available. Virtualization of course introduced a new way how we can deliver people with their hardware, much faster and fully automated. But how often do users really ask for hardware to get hardware!? Most people want hardware because they want some functionality on top of that, like an application or database.
Last week oracle released an updated version of their Cloud Management solution (Enterprise Manager 12c) which is like a vSphere / vCloud Director / vFabric Data Director / vFabric TC Server IN ONE! (and more). Yes one product, meaning one self-service portal with ONE menu. On this menu you can offer users:
- Assemblies (like vApps, but with deeper software integration in the apps and config running on a group of VMs)
- Pre-installed VMs (like a VM with preinstalled Database or middleware software)
- Database as a Service (No virtualization, but a full database instance running on an existing database hardware platform)
- Database Schema as a Service (No virtualization, just offer zillions of users a schema out of one existing database system, super low overhead)
- Middleware / java as a Service (No virtualization, instant provision java applications to an existing middleware platform, with full automation to scale up and down)
- Testing as a Service, Using virtualization you can easily get any kind of test environment, PLUS!! full automatic execute workload tests against these test environments! Just pick the a workload test, which you can design once, from the menu and all will be taken care off, including full automatic reporting!
For the Database/Schema as a Service, this is not just based on getting an empty database but also includes the capabilities to get snapshots from an existing database, great for testing purposes. And we support both Thick and Thin Snapshots. So if you want to run a test against a gigantic database, you can get a thin snapshot made in seconds, use little extra diskspace and you are all set, and all this using no virtualization (no multiple OSes and software stacks) but build-in capabilities of our database and management platform.
So if you want to run an IT restaurant, where your users can not just ask for pots and pans to cook up their own food, but where you can serve an actual meal, check it out
Here on run-virtual we always like to promote free things And one of the leaders in giving free things is Veeam, with their previous awesome home-lab give-a-ways and free passes to IT shows.
And again veaam is stepping up to the plate, you can win a free pass for TechEd 2013!
Well I never before really got involved with the top virtualization blogs, mainly because I do not consider my blog to be one of them But I do think we all, us the people reading the blogs, should really take a moment and participate. Most of the 200+ blogs are run purely on a voluntary basis and most bloggers, like myself, spend a lot of private time into sharing knowledge with us all.
So have a look, pick your 10 best blogs. Select the best blog in a few categories. Unfortunately there is no categorie for best Hardware Gadgets for virtualized environments blog, else I could still maybe have won a price
Well I am still making progress with my analog alarm/monitor box, but really still need to come up with a better name for it! If anyone has any suggestions, please shoot!
I just finished the board design (I hope it is correct) and ordered a few boards, I should have them just after valentines. All the other components I already have in stock, so I should hopefully be able to offer a few kits for any of you that would like to have one of these yourself.
The software development is also going steady. I finished last weekend the configuration software (webbased), fully based on ajax. You can setup for each meter what object you want to meter (VM, Host, ResourcePool, Cluster, Datacenter) and then select any of the available counters for that object. A few are percentage based, so the meter can display 0% to 100%, but you can also have absolute numbers displayed, as you can define in the config software the min and max value.
I guess before I receive the circuit boards I really need to finish the casing, as I have not done much work on that. Hopefully can make some progress on that coming weekend. My current prototype just have a faceplate. In the end I will design 2 versions; A 4 meter version and a 6 meter version.
To be continued….
While I am playing on making cheap, cool hardware gadgets for VMware, others are focused on finding the best possible hardware to run your ESXi (or other hypervisor) on. Alex from the UK has made some real awesome progress on this in the past few days. Using a newly released Intel NUC, which is a new barebone system for 260 EUROS (inc VAT) ex memory, but capable of max 16GB he found a way to run ESXi v5 on it. Not only is this a good enough system for home lab; 16gb RAM, 2x Intel I3 Cores, 1x 1GB LAN, and the option to add extra via mini-PCI-express cards (~43 euros) it is also Quiet!!! (like in silence), which probably for most home labs will be awesome. My servers have to be in my garage (out side of the house) due to their noise. O and from the energy conscious geeks, it also just uses a few Watts
The total system with 1nic and 16GB ram comes down to 348 euro! (inc VAT). This put having 2 physical lab nodes into the possibilities of many virtual geeks out there… and without pissing of their girl friends /wifes (at least noise wise).
Read here the full article by Alex how he got everything up and running.
Is this a new trend? I hope so! I have been working on my analog hardware monitoring box for vSphere environments the last couple of weeks. This weekend I hope to finish writing my software so I can finally release a full working version. But I am not the only one experimenting with Arduino’s and Raspberry PIs to engage them with your vSphere servers and other datacenter components. Bouke, a training consultant (also Dutch), has also played with the Raspberry PI and interfaced it with vSphere. In his case he is using pyton, where I am using PHP. He just released a video showcasing his 16 LED micro monitoring box. One of his customers has a big 16 VMware vCenter large environment. So each LED represents the status of a vCenter server. It is not just pinging the servers, but really using the API to check that vCenter is running.
Check out Bouke’s blog post: http://www.jume.nl/entry/umu-my-raspberry-pi-vsphere-monitor
Let the REVOLUTION of Micro Hardware Gadgets start
Today the VRC Team released a new report in a long list of reports regarding Desktop Virtualization. This report goes into detail about how Anti Virus solutions impact you performance for VDI environments. It does not only investigate the traditional methods of doing anti-virus, but also looks at the various offloading (via network and via vSphere vShield API) methods.
Interesting to see is that the vShield offloading method does improve well on Disk i/o, but CPU is not being reduced.
The most important thing is that the document does not just describe the perfomance result with the default behaviour of the most common anti virus solutions, but discusses what kind of optimisations can be made and what their impact are. So a very useful document for anyone running or building a VDI environment.
The 80 Page report can be downloaded here: http://www.projectvrc.com/white-papers