back to article Microsoft builds open source document database on PostgreSQL, suggests FerretDB as front end

Microsoft has launched a document database platform constructed on a relational PostgreSQL back end. The Redmond giant is also suggesting users can kick things off by implementing an open-source FerretDB solution on the platform to get started. While observers might be stunned by Microsoft's growing acceptance of the open …

  1. news.bot.5543

    MS SQL Server?

    So, is Microsoft's own SQL server now abandonware?

    1. RAMChYLD

      Re: MS SQL Server?

      Knew I had something going when I decided to make Postgres my new application's backend instead of SQL Express!

    2. deadlockvictim

      Re: MS SQL Server?

      Think of SQL Server as the clever child in a sports-mad family.

      Yes, she gets good results, she is praised frequently by others, she helps her siblings with their homework, she is just, well, not what they want now and they can't really dump her.

      The family is needlessly embarrassed by her.

      So, they quietly promote her, give her what she needs but is rarely praised by her family.

    3. abend0c4 Silver badge

      Re: MS SQL Server?

      PostgreSQL is (very) extensible by design, making it an excellent basis for a project of this nature.

      SQL Server is a pretty competent traditional relational database product. But PostgreSQL is that too, so it definitely has an edge.

      The Microsoft product is probably better integrated with the Windows development ecosystem - and it's probably the future of that which will determine its fate.

    4. daft-key

      Re: MS SQL Server?

      Probably about as "abandonware" as it was 20 years ago when MySQL and postgreSQL replaced it and Oracle everywhere.

      Oh wait, that didn't happen either.

  2. Anonymous Coward
    Anonymous Coward

    .........Actually Began To Surface Thirty Years Ago........

    Quote: "...trend for document databases began to surface more than ten years ago..."

    Really??

    I was working using Interleaf's RDM product.....in 1994.

    RDM - Relational Document Manager

    1994 - That would be 30+ years ago.

    1. Richard 12 Silver badge

      Re: .........Actually Began To Surface Thirty Years Ago........

      Ah, but back then it was still submerged.

      Up periscope!

    2. gosand

      Re: .........Actually Began To Surface Thirty Years Ago........

      This is more than ten, so it's technically correct.

  3. Jou (Mxyzptlk) Silver badge

    So is Ferret DB a kind of cache?

    I am not deep into SQL and databases, even though I have to deal with them from time to time on the administrative side since many SQL-specialists can query faster than I can blink, but are blind on the administrative side.

    So my question: What is the reason for two DBs combined in that way? Is Ferret some kind of data-cache for repeated similar queries to take off the load of Postgres?

    (As for BSON, there is a BSON powerhsll module - maybe worth a try since the PS built-in json does not store/retrieve the datatype like "int32" cleanly like the PS built-in XML)

    1. drankinatty

      Re: So is Ferret DB a kind of cache?

      All pretty new to me as well, but reading the ferretDB docs, ferret looks like the interface layer between you and your NoSQL docs and the DocumentDB extensions to postgres. It also seems ferret provides mongo emulation allowing you to move your collections off mongo and store them in postgres using mongo native libraries to access, update, delete, etc.. the collections, documents, databases, etc..

      Take this with a grain of salt, as this is just what I gleaned from a quick read of https://docs.ferretdb.io/ docs. I'm still trying to wrap my head around what the two MS postres extensions that are DocumentDB do and how that interacts with ferret. It will take a much deeper read and likely actually setting up a test collection through ferret and then looking at the storage, schema, etc. from postgres to get a better feel for just exactly what roles the DocumentDB extensions and then ferretDB play on top of postgres to make it all work. The site says all Mongo libraries work with ferretDB and I'm presuming that includes the PHP library as well. Should be interesting to figure out, and I'm always happy to see another open-source player in any area.

      1. Anonymous Coward
        Anonymous Coward

        Re: So is Ferret DB a kind of cache?

        FerretDB emulates MongoDB, and stores data in Postgres. DocumentDB from Microsoft helps FerretDB to do this a lot more efficiently than if it would be just Postgres.

        MongoDB tools, libraries, frameworks all work with FerretDB, unless they use some obscure MongoDB feature which is not implemented yet.

  4. steelpillow Silver badge
    Boffin

    S1000D, not

    Here in Technical Publications land, aka Tech Pubs, we have something called a Common Source Database (CSDB) loaded with tech docs in XML format, all to the S1000D global standard. It majors on document management. Wanna know the manufacturer, age and batch number of that bent wire that brought down your 15-year-old airliner? Wanna know if it was put in the right way round? Your CSDB has it all, right down to the document version that had the drawing the wrong way round (well, maybe a little help from S2000M too).

    What matters is not speed, it is the metadata management tools. That includes direct manipulation of the character-based XML markup, not some impenetrable bytecoded json.

    Here in my dark little corner we can't afford a full-blown S1000D CSDB, and it's OTT anyway. We get SharePoint libraries - the cloudy collaborative MS Office's document database. Even if it wasn't fucked up by pointy-haired managers who lock us out of fixing their bug-by-diktats, it would still be wall-to-wall shite, utterly unfit for purpose.

    Now, S1000D's carrier-grade rigour (well, aerospace-grade really) and its habit of spitting the final documents out in print formats, such as Word or PDF, do not suit ICT products. But there are lessons to be learned. Document management needs span a yawning space between the coloured-pencil grade and aerospace-grade offerings. You know, the kind of tech pubs you get when you buy a sit'n'ride lawnmower or a bunch of servers, or the kind of digital presentation formats that adapt to the reading pane of a web browser or a Kindle. Something open-source so you can fix it up to match your business' tech pubs needs and not be forced into its straitjacket. Dare I hope that someone is now attempting to move into that market?

    1. Roland6 Silver badge

      Re: S1000D, not

      > Dare I hope that someone is now attempting to move into that market?

      Big space, the last big effort I remember was the MAP/TOP initiative of the 1980s, trouble is it got bogged down on network standards, and many suggested technical document formats were still in their infancy, so best leave it to the market…

      Given these had GM’s and Boeing’s backing and thus weren’t small, it will take substantial effort, including government intervention to make things happen.

      1. steelpillow Silver badge

        Re: S1000D, not

        Tech doc formats haven't changed in all that time, still Word and PDF. Nobody ever prints them off these days, just brings them up on their lappie. This despite the almost-as-ancient arrival of t'interweb, online help, ePub, etc etc etc.

        I have this kerrazy dream. Just define a standard stylesheet for your industry/company and output your native XML formatted CSDB as ePub 2 or 3. An open source authoring tool would be good, so you could contribute tech pubs related productivity features. You could call your authoring tool - oh, I dunno, Calibre maybe.

        >Sigh!< I'm always having kerrazy dreams.

  5. Anonymous Coward
    Anonymous Coward

    We screwed up...

    Years ago we made the mistake of migrating our knowledge base in OneNote. Mighty convenient, until we tried to export it.

    The only export option is a single massive PDF file that can't be split. :(

    1. Jou (Mxyzptlk) Silver badge

      Re: We screwed up...

      One Note is anti-organizational. Nearly every One Note I saw is chaotic beyond measure, the ordered one are rare. Just like cardfile.exe from NT 3.51 - try it! You can download the latest NT 3.51 service pack, open it with 7-zip, and extract cardfile.exe. It will work, and you will recognize similarities...

    2. Anonymous Coward
      Anonymous Coward

      Re: We screwed up...

      > Years ago...

      Bit late now for you, but for anyone else in the same boat the OneNote .one backup files are HTML under the hood so the data is there, so to speak, but you'll need to work to extract it. Depends how handy you are on scraping HTML-like files.

      1. Anonymous Coward
        Anonymous Coward

        Re: We screwed up...

        Years ago we migrated TO OneNote. What makes you think we aren't still stuck???

        But good to know, when we do get fed up I will see if I can write something to parse them. Reverse engineering protocols and file formats is something I have experience with as a hobby, but wasn't something I was asked to do on this instance.

        Personally I'd have gone for a folder full of markdown files with inline images, but this was shortly before I joined the organisation.

    3. Larry D

      Re: We screwed up...

      You can export onenote to .mht files which is basically HTML. Even the corporate version.

      Yes OneNote can be chaotic, particularly when using the mobile version where the app is sub optimal for organising, but at least it's searchable.

      At a new site I dump all my learnings about the site in to my own OneNote. Then people complain I have this resource and why can't I put it into their terribly organised formats (OneNote or whatever). Don't blame OneNote for messy minds.

      Is SQLServer abandonware? Probably. I know corporates moving to Postgres or whatever when SQLServer hits scalability limits. Does Microsoft's Linkedin use SQLServer? https://www.youtube.com/watch?v=2DgU8T69VRo&t=13s

      1. Jou (Mxyzptlk) Silver badge

        Re: We screwed up...

        Microsoft bought LinkedIn June 2016. There is no reason to expect it to change the backend to another database, just 'cause it was bought.

  6. Tim99 Silver badge
    Windows

    We've posted this before, but is it "Web Scale?" Wikipedia, NSFW?.

    1. Henry Wertz 1 Gold badge

      a) ok at first i thought that link was dumb but it is in fact hilarious. /dev/null indeed.

      b). BSON? Gross.

      c) As one commenter points out, document management stuff has been around in some industries for decades. I doubt we'd want the same software that an airplane manufacturer uses but it might be good to see what features it has to avoid 'reinveting the wheel', i.e. having something lacking features not because they aren't implemented yet but due to inexperience, not realizing how useful certain features could be.

  7. Rafferty McDonald

    Ferrets or Martens?

    Looks like it's a mongodb mapping proxy.

    I've also been letting an eye on a different animal that has document db implement on PGSQL.

    https://martendb.io/

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like