Dec
    14

    AOS 5.0 – Adapt Not React – Performance

    In AOS 5.0 is Adaptive replica selection is intelligent data placement for the extent store. Rather than use a random selection placement decisions are based on this capacity and queue length, these metrics are used to create a weighted random selection. The current algorithm was great for spreading all of the work load around for fast rebuilds but could cause issues with heterogeneous clusters. With mixed clusters with different tiers size, CPU strength, and running various workloads could have some nodes could be taxed more than others. It also didn’t take in to account the need for rebuilding data if the affected nodes had heavy running workloads.

    This new algorithm can prevent weaker nodes from getting overburden and their hot tier from filling up and reduce the risk of having busy disks. It can also allow for lower utilized nodes to send their replicas to each other and allow busier nodes to have less replica traffic being delivered to them. If we take the example of our storage only nodes we can ensure that replicas will go to the storage only nodes while we’re not sending replicas to other computer-based nodes. This new algorithm also reduces the need to run auto balancing from a capacity perspective. By reducing the need to react we also reserve CPU cycles for workloads and save on wear and tear of the drives.
    In a rudimentary static placement systems this ability to have adaptive replicas would also solve the problem of moving data that then blows up your cache.

    The two less used nodes send their replication traffic to each other. The high-performing node is not impacted by incoming replica traffic.

    The two less used nodes send their replication traffic to each other. The high-performing node is not impacted by incoming replica traffic.

    Since we have a high performing NoSQL database collecting disk usage and performance stats for each disc we can use those stats to create a fitness value. If we can collect stats for a disc we assume the worst case and place a low number for the probability. If we can’t grab stats there is likely chance that something bad is happening to that disc. The disks once assigned a fitness value can be selected by a weighted random lottery to prevent some nodes taking all of the traffic.

    As the product continues to mature were trying to avoid problems from even happening. Whether VDI, Splunk, SAP, SharePoint, SQL your workloads can get very consistent high performance on top of data locality.

    The doctor says prevention is always the best medicine.

    Dec
    10

    Get Ready for AOS 5.0 – Nutanix

    This authentication behavior is changed in AOS 5.0. If you are using Active Directory, you must also assign roles to entities or users, especially before upgrading from a previous AOS version. If you’re not using AD, pass Go and collect $200!

    For customers upgrading their clusters to AOS 5.0:

    * Customers upgrading their clusters to AOS 5.0 will see a pre-upgrade check warning if user authentication is enabled for the Active Directory (AD) service and role permissions are not assigned to any user. The upgrade process will fail in this case.

    Warning - no role mappings

    Warning – no role mappings


    * The AOS 5.0 Prism service (part of the Prism web console) will not authenticate AD users if role permissions are not configured for those users. This situation effectively locks out existing AD users that previously were allowed to access the Prism 4.x web console and other components such as the Nutanix command line (nCLI).
    Add a Role mapping for your AG Groups or Users

    Add a Role mapping for your AG Groups or Users


    To upgrade successfully in this case and to maintain existing access, assign roles (role permissions) to entities that are allowed access to Prism before attempting to upgrade your cluster.

    Dec
    01

    Integrated Single Node Backup with Nutanix

    Integrated backup for remote branch offices and small to medium sized business. Single Node backup is using the NX-1155 which is quotable today . Single Node Backup is apart of AOS 5.0

    Nov
    14

    Docker Datacenter 2.0 for Virtual Admins

    Just a short video walking thru how easy it is to get an environment up and running with Docker Datacenter 2.0 on top of AHV.

    High level points:

    * If you can deploy an VM you can setup Docker Datacenter
    * Management of new docker hosts is easliy done with pre-generated code to paste into new hosts
    * Docker Datacenter has the ability to run both services and compose apps side by side in the same Docker Datacenter environment

    Later this week I hope to have a post talking about the integration with Docker Datacenter and the Docker trusted registry.

      Oct
      31

      Eliminate Standalone NAS & What’s new with Horizon 7

      Thought I would post the links to 2 new on-demand webinars. The Horizon 7 webinar has some Nutanix but mostly focused on Instant Clones, App Volumes and user impact.

      Horizon 7: New Features and How it Impacts User Experience

      The AFS webinar has some great questions and there is a demo at the end as well.

      Eliminate Standalone NAS for your file server needs with Nutanix Acropolis File Services

      Sep
      16

      Serve Files with Enterprise Cloud Agility, Security, and Availability with Acropolis File Services

      afs

      Nutanix continues on its Enterprise Cloud journey at the .NEXT On-Tour event in Bangkok, Thailand. Today, we are proud to announce that we are planning to support Acropolis File Services (AFS) on our storage only nodes, the NX-6035C-G5. Acropolis File Services provides a simple and scalable solution for hosting user and shared department files across a centralized location with a single namespace. With Acropolis File Services, administrators no longer waste time with manual configuration or need Active Directory and load balancing expertise. If and when released, this will make 6035C-G5 nodes even more versatile, adding to the current capabilities of serving as a backup or replication target and running Acropolis Block Services.

      [read more]

      Sep
      16

      Build Large File Services Repositories on Nutanix’s Largest Capacity Nodes, the NX-6035C-G5

      Nutanix continues on its Enterprise Cloud journey at the .NEXT On-Tour event in Bangkok, Thailand. Today, we are proud to announce that we are planning to support Acropolis File Services (AFS) on our storage only nodes, the NX-6035C-G5. Acropolis File Services provides a simple and scalable solution for hosting user and shared department files across a centralized location with a single namespace. With Acropolis File Services, administrators no longer waste time with manual configuration or need Active Directory and load balancing expertise. If and when released, this will make 6035C-G5 nodes even more versatile, adding to the current capabilities of serving as a backup or replication target and running Acropolis Block Services.

      [read more here]

      Aug
      07

      Battle Royale: View Composer VS Instant-Clones – Deploy

      Horizon 7 added Instant-Clones with the ability to clone a full desktop in 4-5 secs. What is the catch? Not really a catch, but no explanation that it takes a bit of time to prep the desktops. For testing purposes, I decided to clone 100 desktops with View Composer and 100 desktops with Instant Clones.

      For these tests I used NX 3460-G4, Win 10, 2 vCPU, 2 GB of RAM

      Impact of cloning 100 desktops with View Composer

      100View5

      You can see hypervisor IOPS and disk IOPS. The impact is really shown on what is happening on the backend and CPU used to create the desktops. So roughly 16,000 IOPS to create the desktops with Composer.

      Impact of cloning 100 desktops with Instant-Clones

      instant-clone1009
      You can see an initial bump in IOPS due to the replica that has to be copied without VAAI. The replica also has to get fingerprinted with does take some time. In my testing it took about eight minutes. The reduction in IOPS is amazing. While you still need performance for running the desktops, you don’t have to worry about provisioning destroying your performance. Disk IOPS was ~ only 1200 IOPS at its peak.

      Summary VC vs Instant Clone

      Deploy 100 Desktops
      View Composer: 5 min
      Instant Clone: 14 min —– virtual disk digest – 8.22 min
      —– Clone 100 desktops 1.4 min

      While the overall process took longer the impact is a lot better with Instant-Clones. With hundreds of desktops Instant-Clones is powerful tool to have in your back pocket. Once Instant-Clones gets GPU support I think they will really take off as the default choice. If you have related questions to performance I encourage you to talk to your Nutanix SE and they can get put you in touch with the Solution and Performance Team at Nutanix.

      Related Articles

      Tale of Two Lines

      Aug
      01

      The Tale Of Two Lines: Instant-Clones on Nutanix

      There was a part of me that wanted to hate on Instant Clones that are new in Horizon 7 but the fact is they’re worth the price of admission. Instant-clones has very low overhead to provide true on-demand desktops or as VMware is tagging it, Just-In-Time desktops.

      On-demand desktops with View Composer..... not happening

      On-demand desktops with View Composer….. not happening

      In my health care days the non-president desktops and shift change always resulted it some blunt force trauma around 7 am and 7 pm when staff would start their day. They only real way to counter balance the added load of login storms was to make sure the desktops were pre-built. This of course means you need so have some desktops sitting around doing nothing waiting for the these two time periods in the day, or use generic logins and then the user never disconnects which was another bag of problems.

      Instant-clones ability to clone a live running VM by simply quiescing the VM is really amazing. Have you ever changed the name of the a desktop and then windows tells you to reboot? If your like me your try to do 5 or 6 other things before you have to reboot which usually ends up in a mess. Instant-clones uses a feature called clone prep to add the VM to AD and change it’s name, all while not having to reboot the VM. When you see a power on operation inside of vCenter it’s actually just quiescing the desktop so there is very low overhead.

      The steps during Clone Prep. MS does not support Clone Prep but they didn't for View Composer so I don't see it being any different.

      The steps during Clone Prep. MS does not support Clone Prep but they didn’t for View Composer so I don’t see it being any different.

      When I went to test instant-clones I wanted to see if on-demand desktops was actually possible without destroying node densities. I had two test runs with Login VSI, 1 run with 400 knowledge users with all the desktops pre-deployed and 1 run with 400 knowledge users but I only started with 50 desktops. I had set the desktop pool to always have at least 30 free desktops until the pool got to 400 desktops.

      Instant-clones delivers on-demand desktops with very low overhead.

      Instant-clones delivers on-demand desktops with very low overhead.

      The darker blue line represents the on-demand test and you can see that the impact over 400 hundred users is pretty small. This is pretty remarkable from a CPU and memory consumption on boot that is being almost eliminated.

      It’s not all unicorns and rainbows however, instant clones does have some limitations in the first release:

      No dedicated Desktop Pools
      No RDS Desktop or Application Pools
      Limited SVGA Support – Fixed max resolution & number of monitors
      No 3D Rendering / GPU Support
      No Sysprep support – Single SID across pool
      No VVOL or VAAI NFS Hardware Clones support (Smaller desktops pools may take longer to provision)
      No Powershell
      No Multi-VLAN Support in a single Pool
      No Reusable Computer Accounts
      No Persistent Disks – Use Writable Volumes \ Flex App \ Unidesk \ RES …….

      vMotion Is supported

      Like anything use case will dictate when this gets used but its a powerful tool inside of Horizon. I plan to show some of the differences between View Composer and Instant Clones in my next posts. Also keep in mind that you still need high IO to service your desktops. Size for the peaks or face the wrath of your end users.

      Jul
      28

      The Impact On App Layering On Your VDI Environment

      I was testing instant clones in Horizon 7 and it was pretty much a requirement to use some form of application virtualization and get your user data stored off the desktops. My decision on what to select for for testing was based on that I had already had ProfileUnity from Liquidware Labs and App Volumes is bundled in View at the higher layers. I wanted to see the impact of layering on CPU and login times. I has also used UberAgent to collect some of the results. While testing I would run one test run with UberAgent to collect login times and then one with UberAgent agent turned off to collect CPU metrics.

      I used three separate applications, each in their own layer.

      * Gimp 2.8
      * iTunes 10
      * VLC

      I used AppVolumes 2.11 since 3.0 is kind of dead in the water and not recommend for existing customers so I can’t see a lot of people using it till the next release. ProUnity was version 6.5

      I first did a base run with no App Stacks or Flex Apps but with a roaming profile being stored on Acropolis File Services. The desktops were running horizon 7 agent and office 2013 and were instant clones. The desktops were Windows 10 with 2 vCPU and 2 GB of RAM. When you see the % listed is a factor of both CPUs.

      Base Run
      baserun

      So not to bad 14 secs login, probably some clean up I could do to make it faster but also not that realistic if your thinking about enterprise desktop so I was happy with this.

      I did test with 1 layer at a time until I used all of the 3 applications. There was a gradual increase in CPU and login time for each layer. The CPU cost comes from the agent and attaching the vmdk to the desktop.

      App Volumes with 3 AppStacks

      3appstacks

      So with 3 layers the CPU jumped by ~20% and the login time went up ~9 secs with App Volumes.

      3 Flex Apps

      3appstacks

      flexapp

      With 3 Flex Apps CPU jumped a bit and login times went up ~4 sec.


      Overall Review

      layeringreview

      What does this all mean?

      Well if you have users that only disconnect and reconnect and rarely log out then this means absolutely nothing for the most part. If you have a user base that gets fresh new desktops all of the time and things like large shift changes then it means your densities will go down. I like to say “Looking is for free, and touching is going to cost you”. Overall I still feel this is a small price to pay to have a successful VDI deployment and layering will help out the process.