Adam's Thoughts

Author: adamfd82

  • From the archives… My Olympic Journey – part 2 of my look back

    11 June 2022

    So as we left off in the last post, I got the change to start working on the London 2012 Olympics. I packed my bags and headed down to London in September 2009, starting as the first network architect on the project, that eventually grew to run a team of 8 direct reports and indirectly look after a team of tens of venue engineers. I never saw myself as a manager, and it wasn’t something that I directly aspired to – however it also seemed like I had a good knack of building fantastic teams and getting the best out of them. The ‘Olympic Experience’ was a hugely formative one for me, working in a critical environment where delays and downtime were not an option. Yet again I was surrounded by hugely talented people I got to learn and grow from, becoming more involved in general infrastructure than just networking, and also helping to bring x86 virtualisation for the first time in an Olympics.

    I think one of the key things for me was how the whole team would work together, from the application developers, infrastructure team, venue IT teams, results teams and more – a whole team working as one to deliver. DevOps was a nascent term in 2010, but perhaps something we were doing before it really took off and became more formalised as it is today.

    Working closely with the telecoms and equipment partners BT, Cisco and Acer, my team built and managed a network spanning something like 50 venues and two core data centres consisting of over one thousand network devices, securely supporting thousands of clients and hundreds of servers to deliver the Games.

    This resulted in me being offered the opportunity to take my knowledge to Brazil and the Rio 2016 Games to design, build and run the IT Infrastructure. Again, this might have been a bit before the term ‘private cloud’ took off, but that I guess is what we ended up building! Working in a brand-new country, building a local team, learning the culture was a fantastic experience, with the added bonus of making a set of great friends who have now spread out across the world, making their own amazing careers.

    Building on top of my previous experiences, we delivered a hugely successful infrastructure project: reducing the data centre server count by something like 80% while improving speed of delivery of new servers through automating end to end deployments using a data driven approach. It was also my first taste of PaaS, looking at how we could take our application stack and run them on a PaaS instead of spinning up VMs and middlewear per application; an experience that would be a key part of the next phase of my career and set the ground for the move in to containerisation and container orchestration. Also worth mentioning is the continual evolution of security, each Games presents new challenges with how we secure it; I think somewhere out there on the net is a video interview I did discussing the growing security threats and how we have to continue to evolve to deal with them.

    Then there was Rio itself! Living in Rio was very influential on me for sure (not least because that’s where I met the person who would become my husband!). Although I had worked with multi-national teams before, being fully immersed in another culture very much was a key step for me and the Olympics really does take multi-national teams to a new level – counting on my hand now I think I had the opportunity to work with people from nearly every continent (Antarctica was missing!) and I’ve lost track of the number of countries. This was a huge opportunity to learn how different cultures work, how successful teams are built and come together, and how to get the very best out of the team.

    As Rio came to an end, I faced a decision of continuing with the Olympics and moving to Tokyo, or returning to the UK. This in the end wasn’t that difficult to decide for a multitude of reasons, but the decision was to move back to London and take up a position in the Cloud Services division, heading up a newly formed Microsoft Cloud Engineering team. But that’s for the next post!

  • From the archives… A look back at over the last 17 years – part 1 – starting as a network engineer

    Originally posted 2 June 2022

    As I mentioned in my last post, I said I’d post a little retrospective of my 17 years in IT, so here is part one!

    Back in 2005, I started to work for a company called, at the time, Atos Origin. I had applied to the Atos Origin Grad scheme in 2004, shortly after I finished Uni. After a multi-stage process I successful in getting a place starting, starting in September 2005. This gave me a bit of a gap, where I put some of my maths knowledge to do private tuition for a year before starting (something that gave me huge respect for teachers!).

    No alt text provided for this image

    I stared my Atos career in networking, working for the network project team in Birmingham. I recall one of my first tasks being given a Cisco 7200 router to build, at the time knowing next to nothing about what a Cisco 7200 was! Thankfully, I had a great team around me to teach me the ropes. This router ended up being part of a shared internet service that, as I recall, terminated a 155Mbps ATM link.  And as one does with a new toy we tried to do some speed testing on this link before hooking it up; although at the time we couldn’t get close to maxing it out from out laptops that only had 100Mbps ethernet ports! And now I am posting this from a 1Gbps symmetric home internet connection…

    The first few years of my career were spent in this team, mostly working on what was a cutting edge shared networking platform, while also having the odd stint of crawling under data centre floors laying cables too! The concept of shared resources of course being the bedrock of today, but in 2005 was still quite novel! I had the opportunity to work across a great range of customers, both on shared platforms and dedicated builds; but more critically the opportunity to learn from some phenomenally talented people, working on cutting edge hardware and highly sophisticated and complex architectures.  There was also a chance to work with some very legacy stuff even back in 2005; there’s a clear memory of plugging a device in to a Token Ring MAU with a warning ‘if it clicks rapidly don’t panic’!

    I think it was early 2008 when I first got involved with what today would be termed private cloud; a project to virtualise an EOL physical estate on to the newly released VMWare ESX 3.5. My role started at looking at the networking but also gave me the chance to take my VCP exam as well.  This was my first ‘taste’ of enterprise virtualisation and helped set the roadmap for the rest of my career; seeing the potential of taking tens of physical server down to a single blade, consolidating racks of devices in to a single blade chassis, while improving performance. It also didn’t go unnoticed that it wasn’t just about reducing servers to reduce servers, but also the carbon footprint – another element that is of course hugely important and increasingly more visible and key in decision making.

    In 2009, the opportunity to work on the London 2012 Olympics came up, so I packed my bags down to London and started the next phase of my journey that would end up taking me half way around the world, but that’s for the next post!

  • From the archives… Microsoft Future Decoded

    Originally posted 31 October 2017

    It’s not that far off two years ago since I wrote a blog post from the Microsoft Cloud Roadshow event in Sao Paulo. Now I’m a bit closer to home, and more directly involved in Azure in my day to day job, heading up Atos’ Managed Azure Cloud engineering and development team. So, what has the best part of 2 years changed in the Microsoft vision at this this two day Microsoft event in London, Future Decoded?

    I took a few key take aways from the keynotes today, and it is still thoroughly exciting to see Microsoft’s vision.  

    Day one had a wider set of themes than just Azure Cloud by itself, including a fascinating keynote on Quantum Computing, and a comedic Q&A with David Walliams. But, we still had some excellent talks on both Cloud and Digital Workplace. The introductory slide below helps to set out that vision, underpinned by the ‘intelligent cloud’.  

    A couple of key messages stood out to me in the talk including one statistic that within 3 years time 50% of the global workforce will be mobile. An amazing stat really, and Microsoft’s vision is of technology becoming just a transparent enabler to allow that work to flow between location and device. It’s a world I already live in, working from between home, office, and indeed Microsoft roadshows, using devices interchangeably to pick up work from my laptop, tablet and phone as most convenient and suitable for the occasion. The future vision sees that going steps further including improvements in Office 365, as well as of course the Surface range.  

    When it comes to Microsoft’s enablement of digital transformation and Microsoft’s view of four towers , the first – Empower Employees – is crucial, and links back to the comment above on transparency.  

    All this work should be to empower employees in the end. I would say the other three points of engaging customers, optimising operations and transforming products all fall out of the first step. Back in my previous role working on the Olympics, a key message that would always be said is the transparency of technology – generally people don’t care about the complex and substantial IT systems that are behind watching a competition, as long as they work!  IT only becomes visible if it fails. In the digital workplace we are heading towards, and in some areas faster and closer than others, technology should take a similar viewpoint. The convenience and simplicity we are often used to at home is coming to the workplace, whether enabling geographically diverse teams to communicate and collaborate better, to an expenses system that is as simple and convenient as many modern apps to use or, as I’ve been doing today, the ability to seamlessly work on a document on my phone, tablet and now laptop to post online. Modern workplace and business applications form two of the four pillars Microsoft discussed.

    They were joined by Applications and Infrastructure, the backbone of Azure Cloud, and Data and AI. These are transformed by the move from virtualisation to PaaS services (specifically mentioned as microservices an containers in the below slide), as well as the convergence of disparate data to a connected data estate. A key point here was with the commonly viewed point of data as a currency.  

    The convergence of disparate data is a fairly standard message, but the capabilities of Azure and Microsoft’s view of AI being an enabler give a significant next step in Big Data analytics.  

    I had a conversation earlier discussing the growth of AI in our day to day lives over the last few years, whether it be interacting with Alexa, Cortana or Siri to the ever improving translations online. It has become such a natural day to day part of our lives without really that much realisation that we are living with tools that only twenty years ago were more the realm of Star Trek than our living rooms. Who knows what the future can hold!

    And one last thing to mention, Project Emma. A genuinely amazing and emotional piece of work.

    https://blogs.microsoft.com/transform/feature/emma-can-write-again-thanks-to-prototype-watch-raising-hope-for-parkinsons-disease

    Tomorrow Future Decoded continues, looking more specifically on Azure.

  • From the archives… Microsoft Cloud Roadshow São Paulo

    Originally posted 24 January 2016

    On Tuesday and Wednesday last week I spent two days at Microsoft’s 2016 Cloud Roadshow.

    Through a series of sessions, Microsoft shared their vision of Cloud and Windows Server 2016 (as well as Windows 10, Office 365 etc).  And I must say I left with a fantastic impression of both the latest improvements either live today or coming soon to Azure, and a real desire to start playing with Server 2016.

    I’d like to share some of my thoughts from some of the sessions (and if you happen to follow me on Twitter you will have had a live blog of some of the sessions almost!)

    For those in Europe, the roadshow will be coming to London on the 29th February, and more details can be found online: https://www.microsoftcloudroadshow.com/cities

    Azure:

    • Azure is running from over 100 data centers worldwide, supported by almost 2 million servers
    • 20% of Azure’s VMs are Linux
    • An Azure data center has a minimum of 30 racks of storage
    • 1,400,000 SQL databases running in Azure
    • 425,000,000 AD users in Azure

    Microsoft Azure Stack:

    Coming with Server 2016 is ‘Azure Stack’.  Instead of just Hyper-V and SCVMM, it will be possible to run the same software stack as runs Azure.  This will bring the same web based management tools that you use on Azure to your own private cloud, as well as what looks to be very seemless integration for a Hybrid Cloud environment.

    Microsoft Operations Server Manager:

    The new OSM web based server management tool definitely looks to have aspirations to replace your SIEM to me – it pulls in event data from all Windows servers as well as firewalls and can present a view of all potential performance, operations and security issues.

    Microsoft Advanced Threat Analytics:

    Another Security Tool, focused specifically on combining data from Windows Servers (via your existing SIEM tool) and providing out the box correlation of Windows events looking for suspicious behaviour.  In a demo shown at the event it detected a potential hack via stolen credentials and un-authorised escalation of privileges.  It can then feed this data back in to your SIEM – effectively providing an in depth level of intelligence behind your event logs more than your average SIEM can do.

    Advancements in Active Directory:

    AD (at least Azure AD for now) will be able to provide built in two factor authentication, in theory making it easier to bring two factor authentication to your own applications, directly via AD.

    ‘JEA’ – Just Enough Administrator rights.  Based through Powershell only, a way to provide highly granular restricted administration rights to users.  MS presented a clear vision of not having people log in to servers, with only remote management.  Right now JEA just applies to Powershell, but could scale out to graphical tools in the future.  Linked with this, but with wider scope is Just in Time rights, based on a web portal that will allow users to request their limited admin rights to  a server, that then has an approval workflow and grants the user rights for a specific time-frame only.

    Server 2016:

    A big topic here!  There’s a lot of new stuff coming here, but here are some of my highlights:

    Nano Server: Nano Server is a new highly cut down version, that can run in a footprint of around 400MB.  This smaller surface has resulted in internal MS testing of far less patches required, and considerably less reboots.

    Nano Server has limited scope, and isn’t designed to be able to run all your legacy apps.  However, it does partner very well with Microsoft’s new Containers, or running infrastructure features such as Hyper-V, DNS Servers, and hopefully by release time AD servers.

    It is a 100% remote management only, via Powershell, Server Manager, or most interestingly the new (currently in internal release only in MS) Remote Server Management Tool.  Indeed, to look locally, there is only a basic diagnostic system to look at configured IP addresses in an almost Linux appliance like shell.

    Containers: Microsoft does containers now!  Now running in 2016 TP4, MS Containers are supported on all versions from Nano to Full (although really targetting Nano-Server as the ideal platform).  Docker is supported for management and deployment as well.   There is also something that will be coming to Azure (currently in technical preview only as well) that to me looked like ‘Containers as a Service’ – in Azure you can deploy a fixed set of servers to run your Containers, currently based on Ubuntu Linux but with Windows 2016 planned, that deploys a stepping stone server, three servers running management tools, and a dynamic number of container host VMs.

    Micro-Services: linked to a clear MS vision here with Nano Server and Containers are a drive towards Micro-Services.  MS has a new management platform / framework, Azure Service Fabric for managing and deploying Micro Service based applications.  They have some great plans here, including zero down-time rolling updates for stateless and stateful Micro Service applications.  

    RSMT: a new web based management tool that can manage all your servers from Azure to internal, Nano Server to full GUI versions.  Through the tool users can access all the normal sort of features they would expect, opening a Powershell window, Services, Cluster Manager, AD etc.

    Software Defined Network: this is an area where it looks like a lot of improvements have been made to the network stack, but it still has a way to go I would say – it didn’t strike me as mature as NSX for instance, and even now still doesn’t provide the functionality we have in our own environment with the Cisco Nexus 1000v virtual distributed switch add in for 2012 R2.

    Azure Load Balancing is now available in 2016, although this is limited to L3 load balancing still and doesn’t offer any L7 or even L4 features.  Interesting MS themselves are using the A10 virtual load balancer for all Microsoft Live ID authentications.  

    The new software defined FW is a big step forwards though, and although MS say it won’t replace a HW edge firewall all together for big enterprises, I can see it actually doing this for some internal private clouds – a comprehensive set of features were presented, along with some impressive performance figures.

    VM Security: a few new features are now introduced, including Shielded Virtual Machines.  This is mainly applicable in a multi-tenant environment, and allows a tenant admin to prevent access to the VMs, or the data on them, from the overall environment administrator.  It utilises a new Azure Key Store to store key material, Bit Locker for encrypted disks, and a new Host feature to also allow VMs to be restricted to specific hosts only.  This should be supported in both Azure, Azure Stack and Hyper-V in Server 2016.

    Software Defined Storage: Some noticeable improvements here, some features were appearing in 2012 R2 but far more mature in 2016.  Storage Spaces already exists in 2012 R2, but has a lot of improvements in Azure Stack.  It allows you to use local disks in each server to build your CSVs and distribute your VMs across, removing a SAN or NAS.  And the performance figures shown are very impressive – a 4 node cluster with each node having 2 SSDs and 4 HDDs was returning 650,000 IOPs, split 70/30 reads and writes.  2016 brings improved Storage QoS as well.

    There are also improvements in SMB Security, Deduplication, ReFS performance (I haven’t used ReFS in 2012 R2, but now I’m definitely tempted with some drastic performance increases on some operations).

    Storage Replica is another interesting technology introduced, with block-wise replication of volumes, with a demonstrating showing the automated recovery of VMs on to a different host, with the underlying VHD volume having a Storage Replica replication to a different server.

    StorSimple is also an interesting product, allowing an easily deployable Hybrid Cloud storage approach, with a local appliance with replication to cloud storage.  Based on iSCSI on your local private cloud, with internet or ExpressRoute (the dedicated link to Azure data centres) connectivity to replicate all traffic to the cloud for backup, or indeed a full DR environment cloud based.  They demonstrated some compelling price comparisons based on a 60TB array.  The devices themselves are being made by Seagate.

    I’ve spent my weekend playing with some VMs, getting a nano server working etc and having fun being a bit geeky!  And I’m left with a very positive impression, especially considering this is still only TP4 with months to go before the production release!

  • From the archives… Securing the Olympic Games in Rio de Janeiro

    Originally posted 23rd September 2016 (which also happened to be the day I left Rio after living there for 4 years!)

    One of the most ever present concerns, not just during the Olympic Games, but for the 2 years preceding them since our first systems went live for the Volunteer Portal, is ensuring the security of our systems. To make the technology of the Olympics run smoothly, we bring to the table over 25 years of experience of delivery excellence to the Olympic and Paralympic Games, some of the most visible events in the world, but also from doing this for our clients day in day out.

    The Olympic Games are first and foremost about sport and bringing the world together. Our work should be invisible, silently working away behind the scenes, but always ensuring that everyone can keep enjoying the big sport event. Every day hacker’s work to come up with new ways to disrupt IT systems, and in return corporations have to keep one step ahead to ensure their systems and data remain secure.

    For Rio we have been building on our experience to securely provide the most connected Olympics ever. Our systems processed and delivered more data than any previous ones that reached and impacted more people than ever before. The Atos team, in conjunction with our partners, worked tirelessly to insure that this information gets delivered successfully, allowing the world to share in real time in the most connected way yet. 200,000 of hours of testing have taken place, testing thousands of different scenarios to ensure that when the event started on the 5th August 2016 with the eyes of the world watching, we were ready.

    But it is not just the results that matter. Our Games Management Systems processed 430,000 accreditations; set-up effectively in the Rio 2016 partners cloud set-up. For the Olympic Games, the pressing IT challenges are to further secure operations, contain costs, and leverage experience and investment across multiple Olympic Games. To meet those challenges, the IOC is committed to continuous improvement and innovation, and delivering greater benefits from the evolution of technologies and emergence of new services.

    These accreditation passes not only act as the person’s credentials to access Olympic venues, but also act as visa waiver for entry in to the country. Other systems managed the 50,000 volunteers and their working schedules. Security was paramount, and for the first time the accreditation systems were running – in the cloud, delivered together with fellow Rio 2016 technology partners.

    Our team of experts, based in the Technical Operations Center in Rio de Janeiro, worked 24/7 throughout the event, keeping a close eye on everything flowing through the network. We anticipated collecting and analyzing more data than ever, building on the over 400 million IT security events that were analyzed during the Rio 2016 Olympic Games and 120 million IT security events during the Paralympic Games.

    Using the latest in real time data analytics we worked to sort through these millions of IT security events, looking for behavior that really were suspicious, filtering down to ensure that our team of Security Experts got fed what they needed to see, and made the human call on what really was a risk to our systems, and what wasn’t. We brought in skills in data analytics to crunch through the vast amounts of data we gathered to bring out knowledge and patterns to help ensure we keep learning and improving to stay that one step ahead of the game.

    So when you were watching the excitement of the opening ceremony, hearing the roar of the crowd as the athletes go for gold, our team were there behind the scenes ensuring that results were securely delivered to the world’s media and that the huge workforce got to the right places at the right times with the right access.