John Hayes
Senior Product Marketing Manager, SquaredUp
Senior Product Marketing Manager, SquaredUp
There was a lot going on at November’s Gartner Infrastructure, Operations and Cloud Strategies Conference, but the central theme was inevitably the transformational impact of the AI revolution. IOCS is a major event bringing together leading vendors and thousands of practitioners for a mix of vendor-led sessions, expert presentations, keynotes, roundtables and one-to-one consultations.
Naturally, as you might expect at an IT conference, the air was ringing with the trill of vendor jargon and sound bytes. As well as the usual suspects such as “hypermodal” and “multi-cloud”, a new buzzword made its debut “Autonomic IT” – i.e. systems operating without human intervention.
Being an observability specialist, Matt Crossley’s talk on the Gartner Magic Quadrant for Observability was the first session I added to my agenda. Matt had the unenviable task of summarizing the increasingly diverse and complex observability market in the space of just 20 minutes. One of the most important points he made about the essence of observability is that it should align with business objectives.
Observability is a fast-moving space, experiencing intensive innovation on a number of fronts. The key trends that Matt highlighted were:
Another subject close to my heart is SRE, so I also checked into Harran Ennaciri’s excellent (and very well attended) session on the principles and practices of SRE. He began by narrating the story of a nightmare release that took a business down for 48 hours.
I think that the longest outage I was ever involved in was five hours or so. I can only begin to imagine what the carnage in the War Room Slack Channel of a 48-hour meltdown must look like.
Harran worked through each of the process (or lack of process) failures that caused – and then prolonged – the outage and showed how applying SRE principles could have prevented a whole world of unnecessary pain.
The topic of LLM cost management could easily become highly technical and abstruse, but Cameron Haigt did an admirable job of communicating the subject clearly to a non-specialist audience.
A lot of emphasis is placed on the API costs of LLMs (i.e. the costs involved in sending and receiving tokens) – Cameron made the very sobering point that these costs are actually only the tip of the iceberg. A point made even more sobering when allied to the observation that most AI projects actually fail. The session also included some really useful advice on tooling for Prompt Optimization as well as strategies such as caching and LLM Routing.
Verizon was highlighted as an exemplar of corporate AI strategy. As well as building a Gen AI Centre of Excellence, they have instituted a mandatory training program for the entire company – from CEO down.
LogicMonitor is an observability platform that has been on my radar for quite some time, so I was delighted to accept the offer of a system demo.
Notwithstanding the usual cautions about vendor-led demos, I came away really impressed. Whilst the system is capable of APM, its real strength lies in networking and infrastructure, and it really stood out as a sophisticated and polished platform. Two aspects which really stood out were its agentless architecture and its extensibility model, which allows users to develop their own complex workflows and custom resources.
AIRIA was also a company that was new to me, but they have built a very impressive platform for building and managing AI applications. AI development entails a multi-faceted workflow and touches on an array of concerns such as prompt evaluation, LLM routing, orchestration, model guardrails, cost optimisation and more. The AIRIA platform has a rich feature set which addresses each of these needs.
The spectre of the VMWare licensing shock loomed large over a number of sessions. In Tuesday afternoon’s keynote speech, Gartner Vice President Dennis Smith actually called out the VMWare crisis as being on a par with the Y2K bug in terms of business impact.
In a session on Top Trends Impacting I&O, Gartner Analyst Chris Saunderson noted that some companies are now even contemplating de-virtualization – returning to bare metal.
FinOps was quite a pervasive theme at the event, and one of the best attended sessions was a presentation by two engineers from Booking.com, detailing how they had embedded a FinOps culture across their IT function. This was a strategic undertaking underpinned by a philosophy of including all stakeholders.
Although they used Apptio as the principal solution for managing cloud spend, their FinOps strategy embraced a number of tools including ServiceNow for forecasting and Slack and Gmail for alerting and internal comms. They cited some impressive stats illustrating the success of the initiative. For example, they achieved a 94% reduction in projects running without a budget.
The event may have only lasted two days, but as Chris Saunderson pointed out, the Gartner Method for identifying important trends is a technique that every business can (and probably should) apply all year round. The model is based on common sense principles and consists of three stages:
Who knows – maybe there is an AI that can do it for you – although I need to figure out whether it would be agentic or autonomic!