# Diskover finds and eliminates ROT data > Source: > Published: 2026-07-08 14:04:48+00:00 # Diskover finds and eliminates ROT data [Diskover](https://www.blocksandfiles.com/ai-ml/2025/04/24/discovering-diskovers-data-management-discipline/1607647?_gl=1*5kbpl9*_ga*MzkxNDQyMTIwLjE3NzcwMzc0NTc.*_ga_NSDTXHMMN0*czE3ODM1MTU3MDkkbzE4MyRnMCR0MTc4MzUxNTcxMyRqNTgkbDAkaDA.), the vast scale file and object data management supplier, has a way of increasing effective flash capacity; get rid of the useless data polluting it. The company’s software provides a wide and deep lens through which customers can find, monitor and manage their data estate using direct and derived file and object metadata to set up data pipelines to put data in the right place, select, and filter it for application use. Think if it as providing a single virtual abstraction layer over multi-supplier and distributed file and object data silos so customers can get a single view of their data assets. CEO Will Hall said: ”We're this abstraction layer above the storage, whether it's cloud or on- prem. We've got one customer that's got 13 different storage vendors in one business unit. And so being this abstraction layer above it we have this logical representation of what the business cares about. … AI is the most ecosystem-dependent workload on the planet. So by definition, if you're in a silo, you're not giving the customer the full benefit of that federated view. And then, if I'm Snowflake, I can just connect to that virtual layer, that abstraction layer of Diskover, and I get access to everything down below.” There’s no need for specific connectors from Snowflake to individual storage vendors. The data discovery part can be eye-opening. Hall tells me: “We've gone into a massive global pharma company where at the initial front end of the engagement, we asked them how many files they thought they had. They thought they had 200 million worldwide. In our first scan, we found eight billion in one cluster in one specific data centre and they obviously had four or five more. Not uncommon at all. It's like going from darkness to light. It's extremely illuminating for these customers.” Diskover acquired CloudSoda in June 2025 and gained technology focussed on data movement and orchestration. That combined data delivery with data discovery. With it, Diskover can give customers data sprawl visibility, move data to optimize its placement, lower storage [costs](https://diskoverdata.com/resources/cost-of-data/), help to maintain regulatory compliance, and select and clean data for AI pipeline use. There can be a lot of redundant, obsolete, and trivial (ROT) data squirrelled away on storage systems; temporary file by-products of application runs that didn’t get deleted for example, and other kinds of data clutter; scratch files left behind by rendering jobs, for example. There can be, Diskover estimates, up to a third of storage capacity wasted by holding ROT data when organizations don’t have any data management system. Hall said: “We've got customers where we're … scanning 600 projects for them on a daily basis and there's insane amounts of ephemeral data that's created in these workflows. … a lot of that data are rough draughts and you can go clean those back up if you understand the workflow,” which Diskover says it does and has an AutoClean facility to find and cut ROT data out. The Diskover idea is that customers should scan their data estates to understand what they have, building a catalog, get rid of the data clutter, and then regularly re-scan to keep it clutter-free; “Get your equipment fit and then stay fit” as Hall expresses it. AutoClean involves finding ephemeral data, tagging it and then automatically getting rid if it. To do that, you need to understand the workflows that created it, separate the signal from the noise as it were, tag them both, and then pass the good stuff up the pipeline, to Snowflake for example, and sweep away the noise using the tags. Diskover VP of Product Olivier Rivard said: “You need to integrate with business context and have additional context and more than just file system information. You want to be able to understand potentially projects that are stored in Jira. You want to be able to link this data in other platform like Splunk if you have logs, or even vertical specific application like in media there's a project management solution called, Autodesk Shotgrid. Another example in media, they have LSF (Legal, Support and Finance) jobs that are ways to keep track of their work. But sometimes a failed job will just generate a bunch of data that, if nothing is looking at this data set, will stay on the file system maybe forever. So you need a way to keep track of that data and potentially automatically purge this data.” Another example: “If you leverage [Aspera](https://www.ibm.com/products/aspera), they have a partial extension when there's a failed transfer. If this file has not been touched in more than 15 days, it means that it's a failed transfer. Why do we keep that on storage? That's the kind of knowledge we already have. It's already defined and we can automate and purge this kind of data out of your system.” LSF is also relevant in the EDA field as well. In fact, customers in many industry sectors have common workflows. Hall said: “It kind of starts with M&E, but then you look at life science, you look at healthcare, you look at pharma, … you look at EDA, oil and gas, automotive manufacturing. They all have this kind of common workflow. There's different nomenclature across, but it's typically some kind of database driving the creation of this data. And then, by hooking into that application, because we're so extensible, adding that context to the file data and then using our auto tag feature, we can correlate those two.” What’s happening is that Diskover adds this business context to the file and object metadata it’s capturing. Typically it does that through an API in the application and sometimes with a JSON file. That means it can better identify ROT data left behind by applications when they run jobs. Diskover can use its tags to understand the condition of data items and rate them in terms of value to the customer; Active crown jewels - high value, in active use Cold crown jewels - high value, gone cold, archive or protect Scratch/churn - low value, transient ROT, reclaim - low value, cold, regenerable Hall said: ”More now than ever because of the cost of storage and how hard it is to buy new storage, you need to get fit and stay fit. It's how we automatically purge this data based on those conditions. Understanding all those different industry specific workflows and automating that, that's the value [we add]." Diskover says it can declutter your data estate, no matter how big. It may well be worth doing, so you can make better use of limited SSD - and HDD - capacity.