Right Workload | Right Platform
We talked about Hype Cycles in the August edition of OCTO Retrospective and how following each new wave of innovation without closing out the previous has led, in part, to the current levels of technical debt organisations are trying to manage. The first step to undoing this cyclic approach is to understand the root of the problem and the capabilities of all technology options on the market today. Moving from a technology-led process to one that is designed to meet the organisation's outcomes over time.
Build YOUR cloud.
One of the initial challenges is the definition of the ‘Cloud’ and other locations where we run workloads. The Edge, Core, and Cloud terms have been used for years now but they suggest hard divides between each, as well as a lack of granularity to support modern application requirements.
My thought is we need to break these terms down to a more granular level, defined as:
Far Edge: True edge deployment, sensors, IoT etc
Near Edge: Small environments like retail or telco edge, smart cities, or AI factories.
Private: IaaS and Container platforms built into your facilities or colocation, not to be mistaken for Enterprise IT
Partner: Private clouds provided by partner and managed service providers, like CDW ServiceWorks.
Vendor: hosted in the hyperscaler facilities but operated by the vendors, think VMConAWS (VMware) or NC2 (Nutanix)
Native: Services from the hyperscaler players like PaaS or native functions
This breakdown allows us to go on and understand the unique benefits and define our placement strategy. One point to note is that a virtualised infrastructure is not, alone, enough to be considered a private cloud, this is what I still refer to as Enterprise IT. A true private cloud needs to exhibit some of the core cloud operating concepts (self-service, charge/show back, automation, scale) to move beyond being a traditional Enterprise IT deployment. This does not mean that Enterprise IT is not the right option for many organisations :)
While it is key, we understand the granular breakdown, as this allows us to make the right decisions on where to run any given workload. We should not consider them as isolated components, but more as execution environments for workloads, each bringing their unique advantages and disadvantages. These environments should be placed under an overall management framework (technical and process) that creates a single logical environment that could be called YOUR CLOUD (insert any name of choice).
Cloud as an Operating Model
One thing that has become apparent in talking to customers over the last few months is the lack of knowledge about the capabilities of technology in the Private Cloud space. When the questions of owning and operating infrastructure arise, or how we support our development teams, the natural and only considered option seems to be the public cloud.?
For many workloads, this is the perfect execution environment, but it is not the nirvana outcome for everything. If it was, we would not be having cloud repatriation conversations regularly. Modern technology can deliver the 'cloud' experience from many different locations and with smart workload placement we can direct new resource requests to the optimal execution environment.
The missing link in most approaches is a lack of definition around the placement rules for workload requests, defining a clear workload placement strategy and the surrounding governance will help align technical outcomes to the organisation's outcome. Consider elements such as cost, sovereignty, latency, performance, scale, length of service and ecosystem when defining your strategy.
Right workload | Right platform
Once we understand the capabilities and functions of each execution environment, we can start to take a more informed and outcome-based decision process. We should be looking at each workload's requirements and ensuring they are placed on the right platform to deliver the required business value. This is the concept of our Right Workload | Right Platform mantra and one that is approached from a position of independence and without bias.
Smart Workload Placement
Consider the use case for a new workload request, 10 instances required for 3 days of testing, public cloud is likely the best environment taking full advantage of the billing model, while that same 10 instances being needed 24x7 for the next 24 months could mean a private execution environment delivers a better TCO. We must remain aware of the scale and demand being placed on the overall environment, in contrast to the effort needed to create the varied execution locations, avoiding the trap of over-engineering the solution.?
Leveraging the cloud management platforms (CMP) with integration into your ITSM tooling can, today, provide a self-service, cross-platform experience to all that need to consume services. On top of this, a CMP can centralise some of your data visibility and orchestration for automation activities that happen downstream.
Strategy before Technology.
The most important message to take away from all this is that strategy must come before technology. When I review strategy documents and have 'futures' workshops with customers the common theme is the strategy is reverse-engineered from a technology outcome. This might be based on current technology deployments, available skills and capabilities, perceived budget constraints or influence from so-called agnostic outside sources. Whatever the influence, this bias can lead to a journey and path being set that is truly aligned to what the organisation needs.
I met with a customer recently who had suffered through 6 changes in IT leadership over 6 years, each new person bringing a new strategy and a forced change as that was the perceived way to prove their value in a new role. This is one of the best values we, the CDW Office of the CTO, can bring to the table, we are not bound by sales cycles, external vendor influences or internal bias for what has gone before. We can truly challenge your thinking, by asking a simple question many times - why?
Customer conversations
The Right Workload | Right platform discussion above is still one of the primary conversations we have with customers and the last month has been no different. Balancing the need for cost-effective long-term 'homes' for traditional workloads, network adjacent development environments that avoid cross-location latency challenges, with modern cloud benefits and native cloud services. It can seem like an impossible journey but with the right approach and thought there is light to be found at the end of the tunnel.?
?Ensuring we start with the applications and data requirements we can work forward to the right platform for each workload. The challenge here is avoiding the assumption that everyone is at the same level of knowledge, ensuring we bring each customer on a unique journey (following a framework) that recognises each customer's maturity level.
?Artificial Intelligence (I know money in the overused term jar) continues to dominate every tech conference and vendor launch at the moment. What I do find interesting is the maturity level of conversations with customers. I met with two customers this month who had shelved AI plans until mid-2024 as they understood more pressing basics needed to be addressed. One customer bought 100's of GPUs (no servers just the H100s from NVidia) just in case they had supply chain challenges next year. Several others have started to leverage more off-the-shelf AI solutions, the ones built into existing software solutions (think Microsoft Co-pilot, Chatbots etc) and are now looking at how they can build a sustainable AI strategy to ensure continued investment.?
Two things have become clear in the AI conversations over the last month. Firstly, the importance of choosing the correct type of AI and foundation models to build upon, there are some significant performance differences and cost implications around making the less than optimal choice. Second is the importance of the software stack used as part of the AI workflow, its impact on governance and compliance but crucially on performance and therefore cost!
Our view is that the use cases, business case, and strategy need to be clearly defined otherwise there is a risk of jumping on the AI hype cycle. Wasting budget that 'might' have delivered better value elsewhere in the organisation.
领英推荐
OCTO Content Round-Up
Our thoughts and opinions as the CDW office of the CTO (OCTO) continue to flow into the world via our social sites and the main CDW Insights page (https://news.uk.cdw.com/ ). Please follow each of the Chief Technologists and CDW on LinkedIn to make sure you don’t miss anything.
In October I had the pleasure of attending and presenting at the Pure//Accelerate event. We have the pleasure of being the first partner to sponsor the event and pick up the Elite Partner of the Year award. Have a read of my summary here:
?Looking forward to the start of November when we have VMware Explore, I thought it would be fun to take a look at the last 20+ years of VMware software-defined innovations. Parts 1 and 2 are live and dive into Networking over the last decade.
?
Our leader Kyle Davies spent a week in Las Vegas attending the NetApp Insight event, exploring all things new in data management, have a read of his article here:
Or check out the video (that includes demos) here:
??
Tim Russell has been busy this last month, multiple parts of his great AI series have continued to be published, here is Part 2 for those that have already picked up Part 1 from last months OCTO Retrospective
?Continuing the AI theme, Tim also provided his thoughts on the Frontier AI report along with the Prime Minister's speech on the subject:
?Tim also has a great read on the impact of Windows 11 and how to approach your migration with the best chance of success.
?
Jaro Tomik has published the first of his series on Experience Management and XLAS taking a unique and engaging approach by exploring the topic. New to XLAs and Experience Management? Don’t worry – this is Ron’s story...
?
Greg van der Gaast has also started to pen his thoughts on how we need to challenge the status quo in the world of security,
?Hope you enjoy the content from the team :)
Random personal thought
Simple one this month, your mental health is important, please look after yourself and pay attention to those around you! You never know what is going on.
Hopefully, elephants playing in water can brighten your day, taken on the Chobe River in Botswana.
Stay Safe and look out for next month’s recap and please reach out if you want to discuss anything in more detail.?
?Rob Sims?
Chief Technologist (Hybrid Platforms)?
CDW UK?