DevOps at Target & .Next

    I was trying to find some examples of companies that weren’t just web companies using DevOps to change their business. I stumbled across the below video of Heather (@hmmickman) and Ross (@rossclanton) that talk about their teams and company’s story about getting aboard the DevOps train. The thing that I heard above people and process was API’s. Even with the use of a few API’s, automation can be used to change the game in your environment.

    DevOps - Target

    Target’s use of API’s

    In the context of test and dev environments(production too) it’s automation that drives cost down. Hypervisors like Nutanix’s Linux KVM version offers something unique in that department being hyper-converged. Imagine having something that has predictable performance, the ability to quickly deploy your applications with VAAI, run docker containers, deploy and move nodes around the environment, attach storage automatically, use Packer to build consistent environments for KVM, ESXi and your local desktop. They’re so many choices to get the job done today, it’s really exciting to see all the options today. Puppet Chef, Salt and so on. I would encourage people to register for .Next and see what the future has in store with you and Nutanix.

    DevOpsDays Minneapolis 2014 — Heather Mickman & Ross Clanton, DevOps at TGT from devopsdays on Vimeo.


    Nutanix is DevOps Kungfu, are you?

    While this video is about DevOps, I think it can be applied to most if not all businesses if you want to be successful. Adam Jacob mentions in the below video, “Do not be timid about the scope or the influence – we are re-inventing the way we run businesses”. Nutanix does have lofty goals but that’s apart of the lure of Nutanix. Would you want to be apart of something that is mediocre? Web-scale gets poked at but that’s cool, it’s all about the end state plus life is too short to take everything seriously.

    Some of the other points really resonate with me when interacting with sales, engineering and product management at Nutanix:

    • Design for the safety, contentment, knowledge and freedom of both your peers and your customers.
    • Each choice you make needs to make life better for the humans involved
    • The ability for individuals to act without fear of unintended consequences
    • Access to knowledge is a leading indicator of social progress.
    • Pull over Push – The “Pull” of demand aligns process and resources, rather than the “Push” of central planning and prediction
    • Empowerment
    • Small batches, near term hypothesis – validation comes from customers

    DevOps doesn’t have to be technical, It’s about you figuring out what works in your environment. Are you DevOps Kungfu?


    Top 20 KB Arictles for VMware Horizon View

    More of a bookmark for myself. Why repeat the past?

    1. Manually deleting linked clones or stale virtual desktop entries from the View Composer database in VMware View Manager and Horizon View (2015112)
    2. Pool settings are not saved, new pools cannot be created, and vCenter Server tasks are not processed in a Horizon View environment (2082413)
    3. Generating and importing a signed SSL certificate into VMware Horizon View 5.1/5.2/5.3/6.0 using Microsoft Certreq (2032400)
    4. The time stamp on a snapshot image does not update automatically in VMware Horizon View (2074301)
    5. Network connectivity requirements for VMware View Manager 4.5 and later (1027217)
    6. Manually deleting replica virtual machines in VMware Horizon View 5.x (1008704)
    7. Connecting to the View ADAM Database (2012377)
    8. VMware Horizon View Best Practices (1020305)
    9. Forcing replication between ADAM databases (1021805)
    10. Restart order of the View environment to clear ADLDS (ADAM) synchronization in View 4.5, 4.6, 5.0, and 5.1 (2068381)
    11. Provisioning View desktops fails due to customization timeout errors (2007319)
    12. Generating a Horizon View SSL certificate request using the Microsoft Management Console (MMC) Certificates snap-in (2068666)
    13. Legacy applications fail to start with the VMware View 6.0.x agent installed (2091845)
    14. Troubleshooting a black screen when logging into a Horizon View virtual desktop using PCoIP (1028332)
    15. Resolving licensing errors when deploying virtual Office to a system with Office installed natively (2107369)
    16. Troubleshooting SSL certificate issues in VMware Horizon View 5.1 and later (2082408)
    17. The Event database performance in VMware View 6.0.x is extremely slow (2094580)
    18. Configuring USB device filters and redirecting Human Interface Devices (HIDs) in View Client (1011600)
    19. Finding and removing unused replica virtual machines in the VMware Horizon View (2009844)
    20. PCoIP Disconnection Codes (2012101)

    First Posted at: http://blogs.vmware.com/kb/2015/04/refresh-top-20-horizon-view-kb-articles.html


    Docker and Virtualization – Both are needed

    Today we released the Nutanix Next Community Podcast on Docker. Our guest was Nigel Poulton who I was lucky enough to first meet at a Tech Field Day event back in 2012. Nigel has a couple of courses available on Docker at Pluralsight and has great insight knowing Docker and spending a ton of time working with infrastructure.

    You can check out the full podcast on iTunes or here:

    A question that gets raised will Docker cut into a significant portion of the virtualization game? Will Hyper-V and VMware lose potential revenue to people running docker? It’s a hard question to answer and I think we tried to address it in the podcast in a round about way. Unless your a Joyent and doing something very custom like with their SmartOS it’s doubtful to me. Developers don’t tend to manage infrastructure and DevOps is only a buzz word for a lot Enterprises. Enterprises tend to be slow moving so a common management platform is important and that tends to be the virtualization layer. If the delvoplers didn’t already move out and consume public cloud resources, virtualiaztion will still be needed. There were lots of reason why people moved away from bare metal installs and lots of those same reasons still apply. Yes Docker can provide isolation but host management, security and protecting the workloads are still very important. You still need to backup and manage your images and any persistent data that may be stored. Does it make sense to make for a Enterprise Plus license from VMware to run Docker? Probably not but maybe there is the right use case. I am still need to get my head around Docker Swarm \ lattice and how it will all tie together. Like AWS, I see Xen and KVM based hypervisors flourishing here. Get the features that you lose out on from going bare-metal and but lower the cost. This is were I can see people still running Docker on a virtualized host because of the familiarity with the management layer.

    From a Nutanix perspective whichever hypervisor you want to run, ESXi, Hyper-V or KVM you get:

    • Per VM Management, per disk level metrics
    • Hardware Management thru one-click upgrades
    • Auto-scale your infrastructure thru the Prism UI or API’s
    • Snapshots/backup – think protecting your private repository
    • Network/Security Management

    Docker Resources
    Want to Learn Docker – Look no Furthur!
    Docker on Windows – Some Insight
    Nigel Poulton Blog
    Pluralsight Video – Docker Deep Dive
    First Look: Native Docker Clustering

    Have a comment, please add it!


    VMware Horizon DaaS 6.1 – 100 Full Clone Desktops deployed on Nutanix [4:36]

    100 Full clone desktops in 4:36 minutes using VAAI on Nutanix. Low impact and high data reduction for full clone desktops.


    VMware App Volumes on Nutanix Best Practices with #UberAgent

    Well I am happy to say there really isn’t a lot of knowledge needed to wrap up the best practices for running App Volumes on Nutanix.

    In general create 1 container for all of your App Stacks and turn on inline dedupe(Performance Tier). Could you put the App Stacks on the same volume as your desktops? Sure but then you can’t inline dedupe with no performance penalty. By turning inline dedupe on any time you go to create or update your App Stacks your applications will get fingerprinted. When the App Stack gets attached to your desktop it will be a read heavy workload. Any reads that have fingerprints associated with them will go into the content cache which is deduped on read. Your applications have a great chance to be served out of RAM instead of SSD or HDD! The RAM happens to be sitting right beside the CPU so you’ll save some CPU cycles to boot from the Nutanix Controller Virtual Machine(CVM). With the use of Nutanix Shadow Clones all of the caching can be done locally regardless of were the AppStack vmdk is being hosted.

    Because you don’t have to keep inline dedupe turned on you can turn on inline compression for desktops. VAAI and inline compression will save a ton space if your plan on doing full clone desktops. The space saved will allow you not to buy storage heavy nodes and save on power and cooling too. VAAI can save around 20X of space and inline compression can save over 2X plus performance improvements if you’re moving big files around on the desktops.

    Inline Compression Savings

    Inline Compression Savings

    I was running Uber Agent and Splunk to grab some results when I was working on Horizon DaaS thanks to Helge Klein . There is a ton of information that the Uber Agent can grab for you.

    Here are the log on times for 300 users in a 48 minute window generated with LoginVSI and inline compression turned on.

    UberAgent - Logon Times

    UberAgent – Logon Times

    I found attaching an App Stack to a desktop added just under 2s to the logon time compared to applications that were nativity installed. Pretty small penalty if you consider the consequence of updating your apps with View Composer or traditional methods.

    One graph that I thought was super interesting that was generated UberAgent was the Total Start IO. The total number of IOs generated from the applications starting up.

    UberAgent - showing how IOs are used starting applications

    UberAgent – showing how IOs are used starting applications

    Imagine if you could get all of those IOs severed from a dedupe RAM cache? :-)

    The below graph is taken from the Prism UI. It shows in blue the hit rate in the content cache which is over 97% and the green line shows the hits.

    App Stacks being delivered with RAM

    App Stacks being delivered with RAM

    The below picture shows the physical savings with inline dedupe turned on from a performance perspective. You can also manually fingerprint your golden image since it won’t be changing much and also enjoy the benefits of inline dedupe without the overhead.

    Inline dedupe saving  from the applications and fingerprinting the golden image.

    Inline dedupe saving from the applications and fingerprinting the golden image.

    The next graph is just the total number of IOPS. Why? Because everyone loves IOPS! Just over 4,500 IOPS. I turned off video during the test so more time could be spent launching applications.


    To recap, 1 container for all of your applications and turn on inline deupe => done.


    The Nutanix Next Community Podcast – 14, 30min shows

    Here is a listing of the latest guests that have been on the podcast so far. Next week we have Douglas Brown from dabcc.com. Load them on your phone for the drive home.

    The full list is at https://soundcloud.com/nutanixnext


    Nutanix News – New NuSchool Video’s, Yahoo Japan and getting your DC fit.

    Over the last week a new video on Protection domains was added online.

    NuSchool – Protection Domains

    The Nutanix Bible was also updated to highlight data path redundancy:

    Yahoo Japan Corporation announced it’s deployment using the Nutanix Virtual Computing Platform for desktop virtualization. The article talks about why Nutanix was selected and the VDI Assurance program that Nutanix offers. < Read more here >

    Marketing is starting a new campaign to get your data center fit. Looks to be some sweet swag but I would also like to point out the new TCO work that has been done as well. Check both out, old boring topics made fun.



    Nutanix Cloud Connect – EBS & S3

    Using a cloud destination like AWS to provide on-demand backup is quick and easy to provide protection for your important workloads. Nutanix customers can maintain and manage their infrastructure though Prism like they have always done before. Nutanix hides the complexity by using Prism to have all Async Remote site appear the same way one once the site is setup. Additional physical gear can be avoided and existing datacenter floor space can be saved for running workloads. The inclusion of Cloud Connect can help ensure recovery from large outages as AWS provides added value of having worldwide availability zones.


    Data that is sent across the WAN can be compressed and the granularity of what is sent is at the byte level. If 32K of data is changed Nutanix will send 32K of data. If only 4K of data is changed then we will only send 4K of data.

    One physical Nutanix cluster can replicate too many remote AWS clusters(no limits); however, today one remote AWS instance can only replicate to one physical cluster.

    The NOS instance uses both Amazon EBS (Elastic Block Storage) and Amazon S3( Simple Storage Service) for data. EBS is used to store the metadata for the cluster and is stored on SSD. For added resiliency the EBS metadata is snapshotted every time a replication occurs to prevent against corruption that might be caused by AWS. Only 1 snapshot is kept at a time. When the next replication occurs the old snapshot will get deleted after a successful replication.

    The replicated user data from the local site is stored in S3 buckets. The NOS instance will get attached a thin provisioned 100 TB thin provisioned disk called the Cloud Disk. Janus helps to identify the disk either local, S3 or Azure based for support of future releases.

    Nutanix provides the automation to maintain your AWS sites with a script that is stored on the AWS Controller Virtual Machine to delete all EBS and S3 storage. This prevents an accidental data deletion on the remote cluster if someone accidental deletes the AWS remote site in the Prism UI.

    All information based on NOS 4.1.1

    More Information on Cloud Connect: 

    Janus – Nutanix Cloud Management Service


    Nutanix Cloud Management Service – Janus

    Providing backup services into Amazon is made simple by the deployment of the Nutanix Controller Virtual Machine (CVM) into Amazon Elastic Compute Cloud (EC2). Deployment is handled by using the Prism UI and supplying the customers Amazon Web Services (AWS) access key and secret key to give a fully automated solution. Under the covers the automation is provided by a new cluster service in NOS 4.1 named Janus. Janus is the Nutanix Cloud Management Service, it will provide the framework for transport between private and public clouds.


    The Janus services are exposed via cerebro which allows customer to use the same Prism interface to manage their environments.

    The Janus service is responsible for:
    • Managing AWS credentials
    • Listing AWS regions
    • Listing Nutanix Amazon Machine Image (AMI)
    • Deploying & configuring a Nutanix CVM on EC2 ( no limits, deploy one to every region if you want)
    • List Nutanix Instances
    • Start and stop Nutanix instances

    More details to follow in an upcoming Tech Note.