‘Data Mesh is Dead — It Just Doesn’t Know It’

Data mesh is currently in the stage of ‘innovation trigger’ and is expected to be ‘obsolete before plateau’ soon.
Listen to this story

At a recently concluded tech conference in Bengaluru, the discussions centred around how data mesh is helping democratise data. Although it may have sounded like a new concept to amateur attendees at the conference, it is far from that. In fact, in social media circles, talks are on about data mesh already being dead!

The then principal consultant at ThoughtWorks, Zhamak Dehghani, first defined the concept of data mesh in 2019. Data mesh is a new approach for analytical data management based on a distributed architecture. It enables end-users to access and query data where it lives without needing to be transported to a centralised architecture like a data lake or data warehouse. In this approach, data is treated as a product and owned by the teams that most intimately know and consume the data. 

The idea was to provide organisations struggling with centralised data platform architectures with a solution enabling them to make informed decisions promptly. Data mesh helps to eliminate the challenges of data availability and accessibility at scale and allows business users and data scientists to access, analyse, and operationalize business insights from virtually any data source, at any location, without intervention from expert data teams.


Sign up for your weekly dose of what's up in emerging technology.

In terms of origin, data mesh isn’t very old. Yet conversations about it being dead have already started doing rounds. Let’s analyse why.

The truth behind the “dead” story

In June 2022, Gartner brought out the Hype Cycle Data Management 2022. A Hype cycle usually measures the maturity of technologies based on the current level of adoption and the number of years to mainstream adoption. Based on the hype cycle, data and analytics leaders can identify promising technologies and practices and decide when it is appropriate to evaluate them for adoption.

According to the Hype Cycle Data Management 2022, data mesh is currently at the stage of ‘innovation trigger’ and yet to enter the stage of ‘peak of inflated expectations’. However, it is expected to get ‘obsolete before plateau’.

Gartner analysts Mark Beyer, Ehtisham Zaidi and Robert Thanaraj, quantified the benefit of data mesh to be low. According to them, its market penetration is also low — between 1 to 5 per cent of the target audience. The hype around data mesh stems from assertions that it remediates difficulties in approaches like centralised data warehouses, data lakes and data hubs.

Shortly after the report was published, industry experts began voicing their opinions for and against the observation. Scott Hirleman, host at   

Data Mesh Radio condemned Gartner for its “vendor-first, technology-first view of the world” and opined that it will likely never be obsolete. 

Former Gartner analyst and present head of data strategy at Profisee Malcolm Hawker defended Gartner’s observation. “I don’t think anyone at Gartner believes that data mesh is currently obsolete. The chart says it will be obsolete in the future, but it’s not obsolete now. It’s my opinion, Gartner believes the data fabric will become the dominant data management architectural pattern and that it will be the driving force behind the obsolescence of the data mesh.” 

But is data mesh really dead?

Christian Kaul, data modelling aficionado & data structure designer at Obaysch, posted on LinkedIn, “Data mesh is dead; it just doesn’t know it yet. A few years down the line, there will be so many failed implementations of what people claimed to be data meshes that the term will just be too burned to carry on.” 

Rogier Werschkull, data warehouse architect-engineer at 6G, seconded Kaul by citing the signs of “zombification” of data mesh. According to him, all those claiming to implement data mesh are selling the idea that they are building a mesh with tool X or Y. However, if asked how they implement data mesh principle X or Y in real life, there’s no answer. 

Does that mean the data mesh is really dead? 

Well, implementing data mesh definitely faces some challenges. “This doesn’t mean that data mesh can’t work at all; it’s challenging but definitely not impossible,” clarifies Kaul. 

Mature data management is required for data governance at the business domain level. Also, the data sources should provide complete data. Very often, data quality, data provenance and system architectures create obstacles in the way of implementing data mesh. Furthermore, being decentralised, it is important to ensure that the quality of data products owned by different teams is consistent.

“If you want to implement something like data mesh, then you need to organise it. IMHO you need rules, guidelines and processes, including role definitions. Then you need the willingness of the management to implement it. For sure, you need tools as well. But the disruption is not introducing tools, it is organisational, especially mind change,” comments Andreas Neumann, IT architect at Siemens Energy.

According to Michael Ryan, managing principal consultant and head of architecture at Amdocs, for data mesh to succeed, the organisation needs to ensure that it follows the DATSIS (discoverable, addressable, trustworthy, self-describing, interoperable and secure) principle. Inappropriate identification of either the data detail or lack of integrity to combine them can lead to data product proliferation, resulting in increased management and maintenance and eventual collapse of the mesh.

Deploying mesh is time consuming 

Roberto Coluccio, big data architect & project lead at Agile Lab, feels that the utility of data mesh cannot be assessed within such a short period of time, especially when its implementation calls for social, organisational and cultural shift. “For those who expected a 6-months-long journey towards data mesh and failed, well, I think the failure was in their expectation as well as, probably, in understanding what data mesh is. It’s indeed a journey, you can start quick and dirty but you’ll need to iterate and rework, or you can do things straight in the first place and iterate less but you won’t jump start,” notes Coluccio. 

What we think

Well, it is a matter of time to see how data mesh evolves in the near future. At the same time, given it is at the “embryonic” stage of maturity (at least, the Gartner analysis says so), it is too early to impose the “obsolete” tag on it. Also, it is too extreme to discard data mesh as another hyped term doomed to be forgotten. Some of the fundamentals of data mesh are indeed valuable. Organisations shouldn’t rush either ways. They need to be observant and pay attention to how to utilise the best parts of data mesh.

More Great AIM Stories

Zinnia Banerjee
Zinnia loves writing and it is this love that has brought her to the field of tech journalism.

Our Upcoming Events

Conference, in-person (Bangalore)
Machine Learning Developers Summit (MLDS) 2023
19-20th Jan, 2023

Conference, in-person (Bangalore)
Rising 2023 | Women in Tech Conference
16-17th Mar, 2023

Conference, in-person (Bangalore)
Data Engineering Summit (DES) 2023
27-28th Apr, 2023

Conference, in-person (Bangalore)
MachineCon 2023
23rd Jun, 2023

Conference, in-person (Bangalore)
Cypher 2023
20-22nd Sep, 2023

3 Ways to Join our Community

Whatsapp group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our newsletter

Get the latest updates from AIM