What is the end user experience of your VDI platform? Losing sight of the end user experience is a fundamental mistake in VDI. Yet it is surprising how few products monitor the user experience of accessing a desktop through VDI. Today I spent an hour learning about Login PI, from Login VSI. You may be familiar with Login VSI from the product that bears this name and from vendor VDI benchmarks. Login VSI measures the responsiveness of a VDI platform with a controlled desktop count and a consistent real-world workload inside each desktop. They drive real applications inside the VDI desktop and measure the response time of those applications. For benchmarking, Login VSI drives a large number of desktops with a consistent workload. It helps to identify the maximum number of desktops that can be supported at an acceptable response time. This is great to identify the capacity of a platform or the impact of a change, before it is committed to thousands of live users.
What I saw today was Login Pi, which is a monitoring product rather than a benchmarking product. It launches a connection to a single desktop, inside a production desktop pool alongside production users. The single desktop runs through a sequence of application functions and reports the response times. These test desktop connections can be launched from multiple remote sites. Usually from each site where your users reside and to multiple desktop pools. Each session reports back the connection success, network bandwidth, and application response times. The reports go to a central monitoring console where a rolled-up view of the user experience is shown. Right now, the dashboard is fairly simple, it will tell you overall health and let you identify problem clusters, sites or datacenters. I really like the simple and clean dashboard, not overloaded with data. Login Pi will help VDI administrators identify the scope and location of VDI user performance issues. When a user or group of users complain of slow performance, Login PI will help identify whether it is a network issue or a datacenter issue. There are displays to help identify whether it is the WAN to one site or the network inside one data center. The application response time information will highlight data center issues distinctly from network issues. This is not something that end users understand when they report a slow desktop. You may realize that I really like Login Pi. I loved concept when the product was launched a little over a year ago and I like what they are doing to develop it further.
I sat down for a few minutes with Josh De Jong while we were both at Tech Field Day 13. We only managed to grab a few minutes before we had to rejoin the TFD train. That is the way TFD events run, they are very packed and it is hard to snatch a few minutes away. There is also the fear that you can miss out on the funniest and most interesting conversation if you find a quiet place to record, so we were in the awesome SolarWinds training room for the recording.
I did manage to find a few minutes to record with another delegate. Hopefully, I’ll get that posted soon.
SolarWinds has hit my radar a couple of times, both as an exhibitor at Tech Field Day events and as a management tool that I see deployed at customer sites. After a briefing and Tech Field Day 13 I decided that it was high time that I gave the tires a kick. One of the first questions is what products do you want to test? SolarWinds have about twenty products just in their on-premises IT and network management category. They also have a service provider category and a SaaS category, which I won’t be looking at today.
As you probably know, vSphere 6.5 has removed the vSphere Client Windows application. Now all vSphere management is through the web client or the new HTML client, confusingly also called vSphere Client. My lab has been vSphere 6.5 for a few months and running the vCenter appliance since there is now an upgrade process from the Windows vCnter 6.0 to VCSA 6.5. Naturally, my lab is a messy place, I try things that end up breaking and leaving a mess around. I ended up with a few orphaned VMs. Orphaned VMs are when vCenter has a VM registered but the ESXi server does not have the same VM. Happily, my orphaned VMs were disposable and I did not loose anything when they went away dues to a VSAN experiment. But I was left with three orphaned VMs in my inventory, making it look messy. In the Windows vSphere client, you can simply right click on the VMs and remove them from the inventory.
In the Windows vSphere client, you can simply right click on the VMs and remove them from the inventory. In the web client, the right click menu is much shorter and does not include Remove from Inventory. After a little poking around through all the sub-menus, I did locate the right option and was able to remove the orphans. It is quite deep in the submenus, rightly so as you will seldom need to remove VMs without deleting the files (Delete from Disk action). The web client is not a direct replacement for the old vSphere client, expect to take some time to learn new ways to do old tasks.
January is summer here in New Zealand, our holiday season starts with Christmas and school is out until early February. Traditionally almost nothing happens in New Zealand during January. Certainly nothing business related. Despite this, I wrote a few articles here in January, a sure sign that I wasn’t traveling. In addition, I wrote on TechTarget and TVP, as usual.
I wrote about preparing to deploy HCI in Streamline your implementation of hyper-converged technologies. This is about things that happen after you commit to HCI.
I also wrote about some of the changes in positioning that VMware is doing with its cloud architectures in Is Hybrid DMZ Reference Designs for vCloud Air what it claims to be?
I am a little cynical about IT as a science in enterprises, I thin it is more of a craft. No Matter How Many Tools You Have, IT Is Practised by People.
I also had some thoughts on HPE buying SimpliVity. HPE’s HCI Just Got Real: It Bought SimpliVity. I really hope that my friends are treated well by HPE, they worked hard to make SimpliVity what it was before it was acquired.
Today I wrapped up my first training course fo O’Reilly media. It was two half days of live online training and was all about operating in a vSphere environment. I will be teaching the course again every couple of months for a while. The next one is in April and is starting to fill up. These courses are free to attend for subscribers to Safari Books. I am also planning to develop some more courses. Maybe a performance tuning course or some deep dives into VMware technologies.
I wrote before about why I decided to put an application into a Docker container. Today I’m going to cover a bit more of the how. I originally wrote the application fast and dirty. It was my first serious use of Python and my first distributed system application. Like many coders, I used what I knew rather than losing time learning lots of new things at the same time.
I am back at the front of the classroom, although in this case, it is an online classroom. Starting in February I will be teaching a course on “Operations with VMware vSphere” for Safari Books Online. If you are a Safari subscriber then you can register here on the site The course is included in the Safari subscription, no additional fee to attend.
Of course, if you read this blog you are unlikely to be the audience for this course. I will be covering using VMs for applications rather than building and operating the vSphere infrastructure. The course is scheduled to run every two months, provided there is audience demand. So please promote the course to your application teams, it will help them to be better citizens of your virtual infrastructure.
Next week is another great Tech Field Day event for me. This time it is Tech Field Day 13 in Austin. It seems that I only attend odd numbered TFD events, TFD9, VFD3, VFD5, TFD11, SFD11, and now TFD13. I don’t think that here is any conspiracy here, just seems to be how it works out.
It has been a few months since I last got an update for SimpliVity, so the launch of their OmniStack V3.61 was a good reason for an update.
• There are some nice manageability updates in this version. You can now take an existing SimpliVity cluster and convert it into a stretch cluster. The SimpliVity Data Virtualization Platform will redistribute the VM data copies across the nodes. Then advise you when it is safe to relocate half the nodes to another datacenter.
• The deployment and upgrade processes have been simplified and made more robust. They now use what I call a “Trust and Validate” approach to verify all the configuration settings provided by the customer.
• I wasn’t surprised to hear that VDI has been a huge growth area for SimpliVity. Their first VDI Reference Architecture had some amazing numbers for user density, which lead to great cost per user desktop. In this release, they have a VDI license model for OmniStack. A discounted license cost that disables the backup and replication features. These features are seldom used in VDI deployments. SimpliVity has also published (or are about to publish) Reference Architectures for a few VDI platforms: Epic healthcare, Workspot, and Citrix XenDesktop 7.11.
• One aspect of SimpliVity that I like is the use of compute nodes. There are hypervisor hosts that consume the SimpliVity datastores. Essentially delivering more compute capacity to the cluster. SimpliVity support compute nodes as a normal part of a deployment, rather than just as a migration process.
• The other news from SimpliVity is that their partnership with Huawei has been very successful. This Chinese company’s name is not well accepted in the US. The rest of the world has a lot of Huawei equipment, particularly in telco businesses. Expect to see more models of Huawei servers supported as OmniStack nodes in the future.
Disclosure: in 2015 & 2016 SimpliVity was a customer of mine. I have written training materials and white papers for them, I also have many friends at SimpliVity. SimpliVity did not solicit or have any control over this blog post (or any other on demitasse.co.nz)
Each month I write articles that are published on other sites, which is a large reason why I don’t write so much on this blog. It looks like I only got one blog post here in December, but had at least 7 more articles published. I write most weeks on TVP strategy, this writing is fun as there is minimal editorial control of the content, I write about what interests me.
Over on TechTarget the content is more controlled. I work with TechTarget editors to agree topics. Most of my editors are happy if I go somewhere a little different with topics, so I still get some control.