ORIGINAL REDDIT POST
Any recommendations for an SMB friendly, off-the-shelf, industry agnostic, on-prem DMS that can index a few million searchable PDFs?
I have a handful of cloud adverse clients that are digitizing all their documents. They have 365 for email but don’t want all their documents stored in the cloud. I have a couple of clients with on-prem M-Files which checks most of the boxes but it’s glitchy…
I have a handful of cloud adverse clients that are digitizing all their documents. They have 365 for email but don’t want all their documents stored in the cloud. I have a couple of clients with on-prem M-Files which checks most of the boxes but it’s glitchy and indexing frequently breaks. Is there a similar DMS solution that is more reliable?
Collected discussion
What is your purpose here. To manage documents from creation to archive? Or to make it easy to find things?
Just an easy way to find historical documents with a good audit trail (who accessed what and when).
Yeah this works, I've used this for a client that had a ton of PDFs with incorrect names. Worked well.
did you read the post at all?
It was a joke
Ive never used it at scale, but Paperless-NGX does well with thousands at the news sites I've put it up. Its a pretty simple docker project thay can be put behind a proxy for certs/access. Can't really prov8de you with any other info, but something you can read it to! Happy Hunting!
Using Paperless-NGX for customers, largest has a couple of hundred thousands in it, works fine, I'm using a custom made vue frontend tho as customization within paperless is subpar but since everything is API first it's easy.
For SMB - Therefore by Cannon. Hands down winner from someone who used to implement solutions like Docuware, Laserfiche, Documentum and a bunch of others over a period of 15 years.
Thats the problem with M-Files, the indexing engine just decides to die for no reason. Had a client call last week because search stopped working after a Windows update Paperless-ngx handles million+ documents but need to tune the OCR settings or it chokes. Default config wont cut it for 2 million PDFs trust me
At "a few million PDFs," the thing that makes or breaks you is the search backend, not the DMS front end. That's almost certainly why your M-Files indexing keeps falling over — full-text indexes at that scale are resource-hungry and fragile when they're under-provisioned. Paperless-ngx (already suggested) is great for SMB, but its default Whoosh index bogs down well before millions of docs, so I'd be cautious there at your scale. Options that hold up at millions because they're backed by Solr/Elasticsearch: Mayan EDMS — open source, on-prem, Docker, can use Elasticsearch as the search backend. The most scalable self-hosted pick and a closer M-Files replacement than Paperless. LogicalDoc — community + commercial, Lucene/Solr full-text, on-prem, industry-agnostic. DocuWare — commercial, SMB-friendly, on-prem option, mature OCR/indexing with real vendor support. Costs money, but you're paying for the stability you're missing now. Two things that matter regardless of product: Do the OCR/text layer outside the DMS (e.g., OCRmyPDF in the ingest pipeline) so it only ever indexes already-searchable PDFs. Cleaner and far less likely to choke. Size the box for the index, not the files: millions of docs means real RAM and fast NVMe for the search engine. A lot of "indexing breaks" problems are just IO/RAM starvation. Before you rip M-Files out, worth confirming whether a full index rebuild plus more resources fixes it — the same root cause will bite any replacement if the hardware's undersized.
I've got a client who's using Infodynamics Inbooks for their DMS. I have a few others that need to implement something & was going to dig in a bit more.
https://www.folderit.com/pricing/enterprise/on-premises/ This one might be worth a try
For a few million PDFs, I’d be careful with anything that markets itself as “simple.” At that size, the search/indexing engine matters more than the nice UI. I’d look at FileHold, DocuWare, and Laserfiche. FileHold seems closer to SMB/on-prem DMS, DocuWare is worth a look if they want something more polished, and Laserfiche is solid but may be heavier/more expensive depending on the client. OpenKM might be worth testing too, but I wouldn’t bet production on it at that scale without a serious pilot. Whatever you pick, I’d make the vendor prove it with your actual data before signing anything. Give them a sample: scanned PDFs, OCR’d docs, large files, weird filenames, old documents. Then test search speed, failed indexing reports, reindexing, backup/restore, and what happens when the index breaks. That’s where you find out if it’s actually better.
At a few million PDFs, I would weight indexing reliability and recovery from a bad reindex higher than the feature list. The product that looks boring but gives you predictable OCR/index jobs and clean export paths will age better than the slick one that needs babysitting every month.
Ive been using square 9 as of late and its honestly pretty powerful for documents.
Windows explorer?