back to article Will the last person at Basho please turn out the lights?

Basho, once a rising star of the NoSQL database world, has faded away to almost nothing, The Register has learned. According to sources, the company, which developed the Riak distributed database, has been shedding engineers for months, and is now operating as a shadow of its former self, as at least one buy-out has fallen …

  1. Anonymous Coward
    Anonymous Coward

    It could never have lasted

    Basho was a crazy but fun place, even despite the largely unearned sense of entitlement and superiority among some of the staff.

    The Valve model didn't work: engineers worked on fun hobby projects that were far more about satisfying intellectual curiosity than solving customer problems. Totally understandable but not sustainable. Quality control was non-existent for a long time: a major feature was accidentally disabled in the production version for a few months.

    Money was spent as though it would always be plentiful. There was very little oversight and when it came, it was soon shot down. Drink and other substances were ever present.

    Senior leadership was inexperienced, certainly after 2012, and largely in-hoc to Georgetown. Some appointments were of questionable value and did more harm than good. Post the 2014 regime change, yet more inexperienced people came in and thought they were rather clever for talking about data gravity ... which always sounded like the sort of thing a less talented person might think sounded smart. Of course, the new CTO bashing the product and previous engineers on Twitter didn't help the new regime to hire or maintain credibility.

    But the product was largely good, if niche. Many of the engineers were great. The support was often outstanding. With proper leadership, Basho could have been so much more.

  2. Anonymous Coward
    Anonymous Coward

    Googling for riak brings up a somewhat readable slideshow from nearly six years ago: Basho and Riak at GOTO Stockholm: "Don't Use My Database."

    (The question whether the customer who emitted the statement "We would not have been able to build this application without node.js and Riak Search" still exists comes to mind. Anyone who finds node.js a necessity has a major problem in the architecture department)

    1. Korev Silver badge
  3. Steve Button

    Great Technology.

    Shame it won't be supported for a while, but if someone starts a new company (and if that was me, I'd just call it Riak forget Basho) then it could well live on.

    I guess this is a lesson to learn when working with Open Source technology. The support can just disappear, but you might still be able to get hold of the "talent" via some other means.

    1. Stuart Whitfield

      Re: Great Technology.

      Support for Riak is available commercially (we at Erlang Solutions are a support provider - and were a Riak reseller) and there is a decent-sized community that will step up and ensure Riak survives and evolves, I'm sure. For instance, there are at least 3 significant Riak users working on their own open source reply action layer ATM...

      1. Stuart Whitfield

        Re: Great Technology.

        *replication layer

  4. imanidiot Silver badge

    That depends

    "Former employees that signed a non-compete contract will also have to wait out those contracts before signing up to work on Riak at another company."

    Non-compete clauses don't caunt if you get laid off or fired. That has been established in court already.

    1. Charlie Clark Silver badge

      Re: That depends

      Or if the company folds.

  5. Anonymous Coward
    Anonymous Coward

    Sorry, who?

    Sorry, who? Never heard of them.

    Whether we like it or not (and often unfortunately), the success of any product is as much down to publicity and awareness as it is any technical merit it may have. We've all heard of MySQL, MariaDB, PostgreSQL, and - shh! - even Oracle and Access. This is the first time I have ever heard of Basho and Riak. As too many have found out the hard way, if you don't have a userbase and a community, you don't have a product, sadly.

    1. Anonymous Coward
      Anonymous Coward

      Re: Sorry, who?

      That probably says as much about your product awareness as anything else. Riak was and is a significant player in the data space, and I'd expect anyone working in a data-intensive role to be at least aware of it and its key characteristics.

      Its big problem is the same as most of the other "NoSQL" systems. It could never answer the question of what it did differently or better than the other systems. Add in these internal corporate dramas and an inability to sell and it should surprise no one that it has ended up on the rubbish heap alongside other such flash-in-the-pan luminaries as RethinkDB.

      I suspect a few others will join it shortly. Their VC cash is running out, they're not profitable and they're not getting any more easy investment. Odds on couchbase being next, anyone?

      1. Charlie Clark Silver badge

        Re: Sorry, who?

        Riak was and is a significant player in the data space

        In the what? Codd must be turning in his grave.

        1. Anonymous Coward
          Anonymous Coward

          Re: Sorry, who?

          "It's data space, Codd, but not as we know it!"

          And whatever happened to VoltDB (NewSQL 2.0)?

    2. a_yank_lurker Silver badge

      Re: Sorry, who?

      Obviously not very aware NoSQL, non-relational databases. NoSQL databases have a tendency to either really suck at the application or be better than sliced bread at the application. So they are more difficult to deploy successfully as there are applications they truly suck at. To make matters worse different types of NoSQL have very different strengths and weaknesses. Relational databases may not the absolute best for an application but they rarely are the absolute bottom dweller. I am aware of Riak but have not really looked into it as what I saw indicated it would not be a good fit for my needs.

  6. John Smith 19 Gold badge
    WTF?

    and the UK NHS thought this was a mature technology to pin the NHS "Spine" on?

    F**k me sideways. And Erlang as a language. Holy s**t. From what I've heard it's a solid environment for secure programming with hard real time constraints, as you'd expect from something designed to develop PBX's in. I just don't know how many people know it.

    That said this demonstrates the best and the worst of these sorts of businesses.

    IRL SW systems have URINE tasks (Uninteresting Research Into Necessary Equipment). The stuff the app needs to do but no one if given the choice wants to do. Comms protocols come to mind for example (tedious and error prone to implement without good support tools).

    And in this case it looks like they finally realized you have to have a balance. Yes a distributed data base is cool (although I seem to recall hearing someone did a Lotus Notes app that was good enough to run a metals trading market about 20 years ago) but a fair share of URINE also has to be done by people.

    But look. The companies virtually dead but the software is still both accessible and maintainable.

    The codebase is still available.

    Basho is dead.

    Long live Riak?

    1. Destroy All Monsters Silver badge

      Re: and the UK NHS thought this was a mature technology to pin the NHS "Spine" on?

      Actually Erlang (or Elixir, also running on the BEAM virtual machine) is very cool, although anytyped. These days, you can get the Actor model on the JVM as Akka too (in Scala).

      I just want to find the time behind a sofa to study the Phoenix web server ....

    2. cdegroot

      Re: and the UK NHS thought this was a mature technology to pin the NHS "Spine" on?

      Erlang is actually a soft real-time system, which makes it an almost perfect fit for "web-scale" systems. See for example Whatsapp. It turns out that its actor model is nice and a great fit for modern systems. A new language on the VM, Elixir, is taking off like crazy and overall the ecosystem seems to be booming.

      Riak needs a ton of maintenance, and I can't wait for a community to pick it up and run away with it - I've only heard and experienced good stuff around it.

      1. John Smith 19 Gold badge
        Unhappy

        Erlang is actually a soft real-time system,

        What I've read of it I liked, but given the scale of the NHS task even a little tweaking, when you're dealing with 10s of millions of records ( most of the UK population + annual births - annual deaths) could get very tricky. How any people in the UK have Erlang skills?

        1. Anonymous Coward
          Anonymous Coward

          Re: Erlang is actually a soft real-time system,

          'How any people in the UK have Erlang skills?' Many more than you'd think. Big Erlang/Elixir teams exist in companies as diverse as bet365, EE, Skybet, Alert Logic, Vocalink (their real time payment switch relies on Erlang and is live in various international markets) etc. Many of these are running systems with demands far in excess of the NHS Riak installation

          There's a vibrant Erlang ecosystem out there which is growing quickly thanks to the publicity that came from WhatsApp a few years ago and the rise in popularity of Elixir (which runs on the Erlang VM)

          Open source Riak has some high profile users with significant installations - Goldman Sachs, the Weather Channel for instance

  7. Anonymous Coward
    Anonymous Coward

    And more DBMSs will follow soon...

    I worked in 3 NoSQL companies so far and they have all the same problem: they are not sustainable, even after years and a lot of money.

    Riak was a big name in the NoSQL space, and $60M is a lot of money. AFAIK no one of the VC funded companies, in the DBMS space, are profitable today. I'm expecting to see other companies disappear in the next months, especially the ones can't acquire big enterprises as clients.

    I think the DBMS business is not compatible with the VC game. The key with this kind of slow business, as the DBMS market, is to don't get funding at all, get a bunch of great developers, give equity and bootstrap + grow organically with real clients.

    If the product is really good and innovative, you don't need expensive sales and pre-sales guys that cost you a fortune. The product should sell itself, like Atlassian Jira.

  8. John Smith 19 Gold badge
    Unhappy

    I'm curious what the downvoters are downvoting.

    Perhaps you don't like the idea that devs sometimes have to work on stuff that's not very interesting but necessary to turn a clever algorithm, or data structure into an actual product that people want to spend money on.

    Boo f**king hoo.

    Clever algorithm <> application.

    While the original dev might excuse the rough edges of their brain children users are unlikely to be so forgiving.

  9. Anonymous Coward
    Anonymous Coward

    Can't say I'm going to miss it

    Back in 2015 I tested Riak along with Consul, Etcd and Zookeeper for an app needing a simple distributed state store. Compared to the others Riak was a beast to configue and half the time the damn thing would never even start. Just trying to connect to a newly started cluster would block for minutes and inevitably report riak.RiakError: 'unavailable'. We were totally bummed having heard great things about Erlang and wanting to give it a shot but Riak sank that ship.

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

Biting the hand that feeds IT © 1998–2019