UK VMUG 2015 Report – part 3

Following on from parts 1 and 2, this covers the last 2 sessions I attended and the closing keynote.

The mid-afternoon session I chose to attend was “Hitchhiker’s Guide to Migrating from vCloud Director to vRealize Automation” by Ricky El-Qasem (@rickyelqasem) and Alec Dunn (@LegoYoda) of Canopy Cloud.

Canopy’s private cloud was originally on vCD, with vCAC used as a user catalog frontend. Due to VMware’s stance that vCD is for Service Providers, not Enterprise usage, Canopy had chosen to migrate from vCD to vRA. The session walked through the required functionality which would be required in vRA to replace vCD

  • Standard User – Build VM from scratch, not from blueprint
    • Start with blank blueprint and use Advanced Service Designer to orchestrate the process
  • Upload an ISO
    • Create a web portal/ftp site
  • Build own catalog
    • Use a blueprint to request a blueprint be added to custom catalog
  • Create an organisation – dedicated groups of users, with separation
    • Series of catalog entries to request new Tenant, Add user, Change Tenant, etc. Reservations used for Separation .
  • Dynamically spin up a vDC
    • Currently a work in progress, but should be achievable
  • Spin up dynamic network
    • Using NSX – vRA is integrated out of the box, workflow to build new edge when a VM blueprint is executed
  • Active/Active Stretched Metro Cluster
    • Wasn’t possible with vCD!
    • Another WIP – need to define which DC to run a VM on, use affinity to keep it there except in DR
  • Migration of existing workload from vCD
    • Difficult as vCD vApp ≠ vSphere vApp ≠ vRA “vApp”
    • Manual process, but VMware PSO working on bulk export/import method

It was an interesting session for me, as I’ve only had the most superficial exposure to vCD/vCAC/vRA, so this was great exposure to some in-depth issues and highlighted just what can be done to bend the products to fit requirements.

The final session I chose was “vRealise Automation – Lessons Learned” with Frank Buechsel (@fbusechsel).
Whether it was just me being ‘punch-drunk’ from all the previous sessions, I don’t know, but I found this one quite difficult to take in. The main thing I took away was that a vRA managed infrastructure is quite difficult to shutdown and start up! Sorry Frank…

And so on to the closing keynote “Architecting your IT career” with John Troyer (@jtroyer).

I notice that at this point, the video and slide deck from this presentation aren’t available – I do hope that changes as it was another great session.

John began by talking about what ‘Architect’ means in relation to IT, then moved on to what’s termed ‘T-shaped’ experts – people who have one strong skill area, and a wider but lower general skill base (an inverted T). His suggestion was that as those skill areas can change rapidly, and skills become stale, we should also have an ‘Interest’ skill, to make us π shaped experts.
The next topic was Convergence and Divergence in architecture, where we’re seeing the end of separate technology silos, new solid-state storage arriving, and a Convergence on platforms (hyperconverged) as well as dispersion (rack-scale).

Moving on to Cloud, automation and scripting are key, and an observation that while there are a large number of platforms, AWS is 10x bigger than it’s competitors combined.
No talk about the future of IT would be complete without a mention of DevOps, and John talked about ‘The Phoenix Project’ which a significant proportion of the audience had read, and the importance of time-to-value. He also mentioned the importance of Cloud-native apps and Containers, Cloudfoundry vs Docker, the choice of single cloud (mainly AWS) vs multi-cloud (the stance of everyone else but AWS), and how this affects operations and infrastructure requirements.

How do you learn all this? John suggested side projects, and to gather a team – mentors, peers, meetups and open source projects.

His final recommendations:

  • Shift from operator to creator mindset
  • Do things that charge you up
  • Look Up! ie don’t keep your head down doing what you’re doing now.
  • Be a communicator and collaborator
  • To get the job you want, kick ass at the job you have

It was another great presentation from an accomplished and knowledgeable speaker that certainly left me buoyed up for the future of IT.

UK VMUG leader Alaric Davies then took the stage, with the 4x VMworld ticket giveaway. Congratulations to all who won, I know you’ll have a great time. Finally there was the shock announcement that 3 of the 4 UK VMUG leaders would be stepping down, and an invitation to apply for the roles.
A slightly sad note to finish on, but they’ve done a fabulous job for a number of years and I can’t begin to imagine the time and effort it takes to put on an event of this quality as volunteers. Whoever takes over will have pretty big shoes to fill…


UK VMUG 2015 Report – part 2

This is the second part of my report on the UK VMUG 2015 conference held as usual at the National Motorcycle Museum in Birmingham.

The second session I chose to attend was “How Healthy is your Disaster Recovery Plan?” run by Zerto.

I was almost 10 minutes late arriving for this as a result of a detailed conversation with VMTurbo in the Exhibitor Hall, so unfortunately missed the introduction and a significant part of the session. Note to self, save these sort of discussions for the longer breaks next year!

What I took away from the remainder of the session was that Zerto provide a DR product which is Hypervisor agnostic, allowing it to be used for migrations between Hyper-V and VMware (for example) or even between Cloud vendors. I know that we use Zerto at CSC for precisely this use-case, so it was interesting to hear more about it.

Positives: Cloud/Hypervisor agnostic, 2 license models (time limited or perpetual), Non-disruptive testing
Negatives: Async only, n-d testing requires manual config of networking


For the pre-lunchtime session I attended “Cloud Automation: Finding the Silver Lining” run by two people from Xtravirt and one formerly from the same firm.

I scribbled copious notes during this session as automation is an area I’m heavily involved in but not specifically with vRealize Automation/vCAC, but I will summarise my take-aways as best I can.

  • vRA/vCAC has breakout stubs for integration, but for more in-depth integration CDK is required.
  • Using built-in blueprints lack flexibility, but employing the Advanced Service Designer can increase project scope.
  • Adding rollback capability to a workflow can significantly increase the complexity – for an Orchestrator/UCS/ServiceNow project, it took 4 workflows up to 123.
  • Projects using vRA usually end up requiring vRealize Orchestrator.
  • Don’t assume that the availability of a plugin means things will be easy – they might not cover what you want to do, or be any good, or have version issues.
  • To use vRO you need to know JavaScript, even if it’s just the basics of IF, FOR, Variables, Properties, as well as datatypes and endpoints such as SQL, xml, json, REST and SOAP.
  • Planning:
    • Requirements – have to get these right, make sure Specific, Unambiguous, Comprehensive
    • Produce a Conceptual Design
    • Design the process flow
    • Agree Agile vs Waterfall delivery
    • *Expect* changes – agree the change process before you need it
  • Testing – As soon as you write code, you’re a coder, even “just for scripts”. Need to have some way of testing scripts before implementing into production, so have a representative test environment.
  • Ideally work on “Idempotence”- an operation that can be applied multiple times without changing the result. Example: job that creates a Logical Switch, only create it the first time, make sure that re-running doesn’t result in multiple identically name Logical Switches.
  • Error handing:
    • Plan this up front – it’s massively harder to add in at the end
    • Validate inputs
    • Control errors – remember to provide feedback, automate handling where possible, include diagnostic info if possible)

Lunch was then provided by Webroot, and I have to say it was pretty  good for a conference buffet. Again there was ample opportunity to wander round the Exhibitor Hall and chat to the vendors there.

First up after lunch was “Automating the Next Generation Data Centre” with Josh Atwell

This was a discussion of the tools, methodologies and new paradigms that are in play for automating modern highly abstracted datacentres.

  • Policy Based Management – reduces manual effort, is consistent and flexible, and can control and monitor configuration drift.
  • VASA Provider evolution:
    • 5.5 – Basic capability profile, Block devices
    • 6.0 – Add storage DRS, IO Stats, Events, Alarms
    • 6.0 VVOLs – Add VVOL Profile, VVOL Management, SPBM v2, Unmap capability
      – These improving capabilities improve automation potential.
  • VMware Tags – Great for automation. Stop using “Notes”
  • Desired State Tools –  Many available including Puppet, Chef, Powershell DSC, Salt. Each have own strengths and weaknesses.
  • Software Defined – this is IT as a service:
    • Needs infrastructure extensibility
    • Policy based management
    • Automatic actions
  • Scripts are code and need version control
  • Next Generation Skills
    • Understand all that the App needs
    • Work to enable Self-Service
    • DevOps is culture, not tools
    • Rise of the Infrastructure Developer

This was another compelling presentation that has given me a lot to think about, and DO!

UK VMUG 2015 Report – part 1

I missed VMworld this year due to a number of reasons, so was determined to attend this years’ UK VMUG to catch up with the VMware community.

As usual, there was the community vCurry night, prior to the event (note: one year can we please have a change from Lamb curry and Lamb koftas? ) and the associated vQuiz, ably hosted by VMUG leader Stuart Thompson in his unique style.

The following morning I arrived bright and early for the complimentary breakfast, before making my way up to the main hall for the introduction and keynote.

Slide decks are being made available here and I will update with links when the videos are posted.


UK VMUG leader Alaric Davies (@alaricdavies) opened the show with a high energy welcome, including congratulations (and Tshirts) for the new UK VCDX awardees Gregg Robertson (@GreggRobertson5) and Sam McGeown (@sammcgeown), as well as news of the 4 VMworld 2016 ticket and travel costs giveaway at the end of the conference.

Opening keynote – Joe Baguley (@joebaguley)

The day was worth attending for the keynote alone. VMware UK CTO Joe Baguley is an outstanding and inspiring speaker, and if we were from the other side of the Atlantic I’m pretty sure we’d be whooping and high-fiving at the end of his presentations.
I will bullet point his main points as it would be definitely worth your while viewing the full presentation if you missed it.

  • Digital accelerators: VMware viewed as #5 in the list of vendors most able to accelerate a companies move to digital.
  • Bridging Client-Server to Mobile-Cloud: This is the move from rigid structures to digital, or slow and predictable, to rapid and changing
  • Pat Gelsinger quote: “Riskiest thing in business is to take no risks”, the world is changing due to the move to the App Culture. Joe referenced his 15yo daughter who has moved banks 3 times to move to better banking apps.
  • App Economy: Business works on a cycle of producing an app, which produces data, which we analyse to improve app-cycle.pngthe app, and so on.
    Enterprise IT moves around this cycle slowly. If you move around it quicker, you win.
    Things like microservices, big data, and containers are designed to help companies go round the cycle quicker.
  • RAID -> SDDC: In the 1980’s 1 disk wasn’t big or fast enough, so RAID controllers were born to abstract this. Also provided “design for failure”, so were a “Cloud” of storage. SDDC provides this but for datacenters.
  • Unicorns (Google, AWS, FB) have the scale to run custom apps, on custom platforms, on commodity x86/storage/IP. We don’t have this luxury, but SDDC allows standard apps, on SDDC, on commodity x86/storage/IP
    – Infrastructure as code makes everything possible
  • Hardware – becoming a world of white boxes: The end isn’t nigh for hardware vendors, but the future is different – higher volume, smaller margins. Jevons paradox
    The future is racked commodity hardware.
  • Software stack: more layers of virtualisation (turtles all the way down)
  • Docker: virtualisation at the OS level, like Terminal Services, BSD jails. Provides shipping containers for code.
  • Farming: Most people in the room can name all their servers, and know where they are in the DC, we should treat servers like chickens not kittens (cattle, not pets). Developers wanted an API to create/destroy 1000s VMs per hour. Ops couldn’t deliver, so instead they asked for 1 big VM and used containerisation to deliver it.
    Now we’re rethinking infrastructure in terms of microservices, containers, Continuous Integration/Delivery.
  • Fragmented Ecosystems: have hidden costs, tool sprawl, governance issues.
    vSphere Integrated Containers – extension to vSphere to allow transition. Found that performance on VIC is better than bare metal due to hypervisor optimisations.
    Photon platform, new subscription based platform, optimised for containers, large scale API.
    Project Xenon, about dealing with the billions of containers/microservices of the future. See blog
  • Need to evolve skills, teams: the future is about delivering APIs not servers.
  • Simon Wardley reference – Pioneers/Settlers/Town Planners
    Pioneers get bored quickly, Settlers needed to take new tech and make it work, TPs needed to commoditise the working product.
    P/S/TP is a cycle, Pioneers use utility from Town Planners to build new things.

Joe finished with a cryptic mention of the future being Unikernels. It was a thoroughly engaging and inspiring presentation, and I hope you will take the opportunity to watch the video when it’s made available.

Now read part 2