Published on the 26/10/2023 | Written by Heather Wright
And a critical factor in user experience…
Too many organisations have spent too much money on traditional IT monitoring for too little return and are still seeing outages and severe customer impacts.
That’s according to Padraig Byrne, Gartner senior director analyst and research director for the analyst firm’s IT systems, security and risk group, who is calling on organisations to think differently.
“We monitor everything, but we see nothing.”
Observability tooling has seen a surge popularity since the pandemic lockdowns, as enterprises realised they were digital organisations, and looked to manage and improve the way end-users experienced their applications and services and provide a better digital experience for them.
But Byrne says the traditional approach has been driven by technical teams in response to specific pain points, leading to a plethora of different monitoring tools within organisations – he says 10-15 tools is the global average, though one Australian client fessed up to having 56 monitoring tools – with no enterprise wide approach to building out a coherent observability platform.
“That has led to major gaps in visibility and means for a lot of issues that arise we simply have no way of knowing the issue has happened until customers call complaining about issues logging in to your system, or even worse you go on X and see #yourcompanynameFail,” he says.
He cites the example of a client spending millions on monitoring tools, but who told him ‘it feels like we monitor everything, but we see nothing’.
“I can think of very few areas that we work on in enterprise and IT where we see historically such poor return on these types of software.”
Byrne is urging organisations to look beyond mere monitoring and view observability as something integral to applications and enterprise, and the central nervous system for applications.
“It’s not something external or added later on, you need to make sure observability is a core function of your applications and is integrated into everything you do.”
It’s a message clients seem to be hearing, with Byrne telling iStart that while large organisations such as the big four banks and other financial organisations have been traditional users and have accelerated their application of the tools, non-traditional industry, including healthcare and government are also ramping up their use in the wake of increasing digital interaction with customers.
“This is an area that has existed for 30-40 years, but we need to start to think about it more holistically and stop thinking in terms of a siloed approach. Don’t look at this simply as a point solution.
“Focus on the client and the digital experience and build everything out from there, rather than starting at the infrastructure and building up. Start at the customer and build out.”
Observability enables organisations to identify areas of friction and improve customer experience.
“You’re looking to find out about problems before your clients do.”
But Byrne admits it isn’t a cheap option.
“When I’m speaking with clients who are doing this at scale successfully, the expense for the monitoring observability platforms can be as much as 10 percent of their cloud infrastructure costs.”
He’s quick to add that not everyone needs to be at that level, however, with local retailers, particularly those without a strong online presence, for example, ‘perhaps spending less’.
Another key caveat is that most clients under-appreciate the volume of data they need to deal with and the costs associated with it.
“Once you go out there and look inside your environment for all the log files and metrics etc, it’s a tsunami of telemetry, which is quite capable of absolutely engulfing and overwhelming your IT operations team,” he says.
Not all data is equal though, and all that data needs to be filtered to enable teams to concentrate on what is important.
That’s where telemetry pipelines come into play. It’s another hot area – a relatively new concept also known as an observability pipeline.
Essentially a system that sits between the origin of data and the observability tool, it ingests data and filters and analyses it.
“So basically it does a reduction on the volume of data coming through, but even more than that, it has the capability to do enrichment of that data.
“Using these tools we’re able to take data, tag it by geolocation, business unit, function, whatever else, that enables us later on to make sense of it, rather than throwing everything in to one big bucket.”
It’s a critical point in the current AI-obsessed world, where enrichment is a key step in getting data AI ready.
Byrne says the area of telemetry pipelines, which didn’t exist three years ago, has ‘exploded massively’.
“One key reason is it has a huge impact on cost. Straight away customers see a 10-30 percent reduction in cost, maybe more.
“It’s a very very important area as we are struggling to deal with this tidal wave of info coming towards us.
Byrne has some straightforward advice for organisations looking to get started on the journey: Start from where you are.
“This is not something where you have a green field where you are free to build something new. You probably have 40 years of legacy investment in this area and it makes no sense to rip that out overnight,” he says.
“Very rarely will I ever recommend going out to market to buy a completely new tool because here’s what happens: You start with 12 tools, go to market to buy a tool to solve your problem and you end up with 13 tools.
“A new tool is not going to solve all your problems.”
Instead, he says organisations need to look at what is already in their environment and how they’re using those tools internally, and where their capabilities can be expanded, which tools are ripe for decommissioning because they’re no longer being maintained or added to by vendors, and which tools have a modern pathway.
“Twenty years ago we had IBM, HP, CA and BMC as the big four players. Some of those tools, like BMC, do have a modern pathway, but many clients aren’t aware of that.”
Others however, such as the IBM suite of tools, ahve been sold off and might be an area where organisations want to replace with a modern tool or substitute for somethign else in their existing portfolio.
Byrne warns that migration is never easy, however.
“The level of stickiness is very high. Once these tools are in, they’re in for decades.”
Organisations need to consider the viability in terms of replacements of tools, whether they fully understand what they do and whether they have the skills in house required.
Expanding the use of existing tools is also a valid option, he says.
“If you are using a tool exclusively for log files, see if that tool has capacity to ingest other types of telemetry, such as metrics, traces or events. Expand that utilisation and you are going to centralise that and get benefits from having coherency of data in a single location, and efficiencies,” he says.
“We’ve seen in the last few years that tools that did infrastructure now do application, and tools that did application often do logs. Slowly each tool has been expanding their domain of expertise to the point that it makes sense to talk about a converged solution of an observability platform.”
Just don’t go looking for the one platform to rule them all.
“There will never be one tool that solves all your problems.
“My advice is you need to start now, but it doesn’t mean you need to go out and buy a flashy new tool. Determine where you are, what your objectives as a business are, what you want to do and drill in to what you have available and scale that out.
“As we have ever increasing complexity in our it environments and as we move towards microservices, modern architectures etc, using your existing methodologies for digital experience and application performance monitoring simply will not be sufficient and you will be left behind and that will impact your client experience, revenue and brand overall.”