Former unicorn MapR desperately seeking cash as threat of closure looms

MapR Technologies, once a shining star of the emerging big data analytics set with a valuation of more than $1 billion, has two weeks to find an investor. Failure to do will see the software developer shutter its HQ and terminate 122 staff, including founder and CEO John Schroeder. The company says it is in discussions with …

[Ian Johnston]

So, basically, anyone who wants the company's IP can either pay megabucks now and take on 122 staff, or pick it up for next to nothing in the firesale next month? Ooooh, tricky.

[Korev]

You're correct, but I can't help but feel sorry for the 122 staff...

[EmilPer.]

the 122 staff ...

... will need like 2 weeks to find new jobs

[Anonymous Coward]

@Ian Johnston .... not quite...

There are stories that Schroeder turned down 800 million recently.

There are rumors that he hid their financial situation from the employees and painted a rosy picture that everything was OK.

The truth is that there is a letter of intent. The company that is doing their due diligence is a good fit because they own a bunch of smaller tech companies where there is synergy.

You are correct that if the deal falls through... its lights out and the IP will be sold for a percentage of what its worth. You can bet that there are vultures out there. The key thing is that they will need to have a really good sales force. There are a couple of large tech players who passed on the chance to purchase, could do so if the price was right.

MapR is a diamond in the rough. Their MapR-FS (File System) outperforms HDFS and fits the need for a distributed file system to make ML, and other advanced 'big data' techniques work and work well. As an example... Mesosphere uses Cloudera's HDFS (Its free for them to do this...) and could use MapR-FS instead. Of course Mesosphere has to also migrate to Kubernetes too. (But that's a different story)

The problem is that you need the staff. At least the ones still on board. (Support, Engineering, and Consultants) The IP without the staff who is knowledgeable is worth less than if you have the staff.

Posted Anon for the obvious reasons.

[Anonymous Coward]

Re: @Ian Johnston .... not quite...

"Their MapR-FS (File System) outperforms HDFS and fits the need for a distributed file system to make ML, and other advanced 'big data' techniques work and work well."

This is a myth MapR convinced themselves to make everyone feel happier about selling a "technically superior" product even while getting caned in deal after deal by the competition. If we want to get down to speeds-and-feeds the reality is HDFS has been at performance parity with MapRFS since the M3 days, and either one is more than capable of saturating every bit of disk IO you can throw at it. Part of the reason MapR have failed is that they organisationally learned this mantra and convinced themselves not just that it was true, but worse, that it mattered.

The reality is the market moved on from boring speeds-and-feeds years ago. MapR, as a company, have repeatedly failed to recognise this. At every turn they've missed where the market is going. And that's sad. There are smart people there and they did once upon a time build some very clever technology (MapRFS is still a great scalable NFS store), but they were hamstrung by idiotic management and awful strategy.

[Frumious Bandersnatch]

But, but ...

Unicorns!?!

[CheesyTheClown]

Burned $300 million?

$200,000/year times 2 is $400,000 for an inflated cost or employing one overpaid SV employee. Multiply that by 200 employees. That’s $80 million a year for 200 employees... to develop and market a product.

Now... let’s assume that the company actually received $300 million in investments.

Was there even one person in the whole company actually doing their job? And was that job spending money with no actual consideration for return on investment?

[DataChicken]

Re: Burned $300 million?

Except that's not how this works, that's not how any of this works. :)

You have to hand it to MapR that their product is fantastic. I think they will be around for a long time. AWS is fine for smaller projects but not for bigger ones.

[Anonymous Coward]

@CheesyTheClown ... Re: Burned $300 million?

The company has been in existence for 10 years.

You think of burning thru cash in terms of salary only. There's a lot more going on.

Marketing and other things like office space, etc. ...

There is definitely more to the story than what has been reported.

In the letter, Schroeder said the company simply couldn’t issue the notice earlier, since that would “significantly jeopardize” chances of success with respect to its pursuit of financing.

This says a lot. Some have said that Schroeder wasn't being totally honest with most of the employees.

Yes, the sales team did miss. It was a massive miss. In part, some would say that the target was unrealistic, and in part, some of the sales team were good, others not so much. Selling MapR isn't an easy task. It takes a strong sales rep and a good SE.

There's a lot more I could say... but of course I can't...

I was amazed that it took this long for the Press to catch on. If anyone had friends who worked for MapR, they would have found out from their LinkedIn news feed a few days after Schroeder cut the sales/marketing team.

Posted Anon for the Obvious reasons.

[CheesyTheClown]

Re: @CheesyTheClown ... Burned $300 million?

Thanks for commenting.

I honestly had no idea how MapR would sell in the first place. The problem is... it was a great product. But it was also expensive. And I don't really care how good a sales team you have is, the website is designed to scare away developers.

I just visited there and I'm pretty sure that I've been in multiple situations where I could have seen the technologies as interesting, but the website makes it look like it's too expensive for me to use in my projects. I can use tools that cost $10,000 or less without asking anyone. But they have to be purchasable without having to spend another $10,000 on meetings where people show Gartner magic quadrants.

I can't use any tools where I can't just pop to the web site and buy a copy on my AMEX in the web shop and expense it. When we scale, we'll send it to procurement and scale, but we're not going to waste a ton of money and hours or days on meetings and telephone conferences with sales people who dress in suits... hell I run away without looking back when I see sport jackets and jeans.

Marketing failed because MapR is not an end user program and developers can't make the purchasing decisions. The entire front end of the company is VERY VERY developer unfriendly. Somehow, someone thought that companies all start off big and fancy. My company is a top-400 and we start projects as grass-roots and once we prove it works, we sell the projects at internal expos and the management chooses whether to invest more in it or not. MapR looks expensive and scary and difficult to do business with.

This is why we do things like always grow everything ourselves instead of buying stuff that would do it better. Everyone is trying to sell to our bosses and not selling to the people who actually know what it is and what it does.

I wish you luck in the future.. now that I've looked a little more at you guys, I'll check the website occasionally when I go to start projects. If the company starts trying to sell to the people who will actually buy it (people like me) instead of to our bosses... maybe I'll buy something :)

[Anonymous Coward]

@Cheesy Re: @CheesyTheClown ... Burned $300 million?

The web site scary? Not really.

In terms of price... MapR revamped their pricing by the amount of storage available.

But while it may be a premium to HDP/Cloudera, you have to consider that MapR outperforms HDP/Cloudera. And then there's the issue of how good your procurement group is at negotiating a deal.

There is a community version which has most of the features. So you can build a non production PoC for free and then if you want to go to production... pay the fee get a new license and apply.

You are correct. MapR is not for end users. None of the Hadoop / Big Data platforms are. You have to think of Hadoop/MapR as a enterprise wide decision. A multi tenant platform unless you have enterprise level departments that can afford to use them.

Most 'big data' projects fail because the enterprise doesn't understand the tasks at hand and do not have staff capable of delivery.

To your point, the reason you see a push to the cloud is that you can rent time and start smaller with less risk. However, if you do a TCO over a 3+ year time frame you'll see that going to an external cloud provider is more expensive. Taking a hybrid approach.

As I've said in other posts, MapR is a well engineered product. There are flaws, but its better than the competition. The problem that all of the Big Data tool providers have is trying to explain the complexities of the platform and to level set the expectations for those new to the platform.

MapR has had a lot of missed opportunities. And unfortunately mismanagement has hurt them.

There is more that I could say... but then I wouldn't be anonymous and already I'm sure there are a couple of people who can already guess my real name.

[Anonymous Coward]

The software is quite good

I really think MapR's product is superior in this space. We used it at a large manufacturer at my last gig. Their platform stays up forever and can handle any scale you throw at it. Someone would be smart to gobble them up.

[Gordon 10]

Re: The software is quite good

unfortunately just being superior is has been no guarantee of survival since at least Betamax if not well before.

[CheesyTheClown]

Re: The software is quite good

Everyone always talks about Betamax as if it was infinitely better than VHS. As someone who thoroughly understands the physics, mechanics, electronics, etc... of both Betamax and VHS from the tape all the way through the circuitry up to the phosphors, I'll make it clear... yes Betamax was better... but the difference was negligible. The two formats were so close to being the same that it barely mattered... and when transmitting the signal over composite from the player to the TV which ... well to be honest was 1950s technology (late 1970s TV was still 1950s tech... just bigger)... it was impossible to tell.

S-Video and SCART (in Europe) made a slightly noticeable difference. Using actual component cabling could have mattered but neither Betamax or VHS could take advantage of that.

The end result was simple... when playing a movie recorded on Betamax on a high end 1970s or early 1980s TV next to the same movie recorded on a VHS tape, you had one big ugly player next to another and the only possible difference you could give the consumer was "Beta is more expensive because the quality is better" and of course... it wasn't... at least not enough to notice. Often you could sell the consumer on audio quality, but on 1970/1980s era speakers and hifi, you wouldn't notice until you were far past the average consumer threshold.

Betacam SP was actually substantially better, but by then it no longer mattered.

I used to have 400 Betamax decks and 600 VHS decks in my office... all commercial grade duplicators with automatic tape changers. The Betamax decks existed for collecting dust. The VHS decks were constantly being serviced because they were running 24/7. I spent 10 years of my career in video technology development (I am a codec developer at heart, but I know analog too). In 10 years of working with studio/commercial grade broadcast and duplication equipment, and knowing what I know about the technology, if I saw Betamax for $120 and VHS for $110, I'd still by VHS.

[Anonymous Coward]

Re: The software is quite good

It was only superior for a brief window around 2010. MapR's software stack is a proprietary clone of the upstream components they pretend to sell - they simply could not keep up with the pace of development across the dozens of projects they were copying from so rapidly fell behind. Today it's a disjointed, half-baked mess of features without a single management or monitoring plane that doesn't integrate easily with anything else on the market. Awful user and operational experience.

[Anonymous Coward]

@AC ... Re: The software is quite good

@AC, I think you either work as an Apache committer or one of MapR's competitors.

FD. I have worked w all three platforms and have 10 years in this space. I do not work for MapR but have supported their customers.

All three vendors have different tools for cluster management. Cloudera's is also proprietary and like MCS (MapR) has its quirks. Still far superior to HDP's use of Ambari.

I agree that MCS could be better designed. Still its usable.

That said, MapR has always had the lead when it came to performance and features.

MapR-DB is superior to HBase, however their marketing had a big push on MapR-DB JSON tables which has some good features if you want to store documents and not binary data, yet using the binary tables is also viable.

In terms of MapR-ES (Streams) while not 100% Kafka compatible, their latest release is close... you don't need it. If you want, use StreamSets and you can apply it to both Kafka and MapR-ES. makes life easier.

In terms of tying in to existing corporate monitoring systems... YMMV. if they support REST you can do it. You can use MapR's maprcli to pull alerts and then push them to the other system in a simple shell script. Or email them. You can also listen to Stream[0] to get all sorts of information and then filter for the stuff you want.

Sorry but clearly you don't know the platforms as well as you think.

Posted Anon for the obvious reasons.

[Anonymous Coward]

Re: @AC ... The software is quite good

You're kind of making the point. This was MapR's sales pitch time and time again. They'd tell customers you don't want HDFS, you want MapRFS, because it's "technically superior". They'd tell customers you don't want HBase, you want MapRDB, because it's "technically superior". They'd tell you you don't really need those Kafka features anyway, but here's yet-another-also-ran vendor to use if you think you do.

Meanwhile you can go to AWS or Amazon or Google (or Snowflake or Databricks or Cloudera) and have a complete replacement stack spun up in about fifteen minutes, exactly to spec, charged by the minute.

No one, and definitely not customers, cared about technical superiority. That's why MapR lost to HDP and CDH on premises and why all three are finally losing to the cloud vendors. Customers care about time to value, ease of use and security/visibility in approximately that order. If you're still sitting there tossing yourself off over how "technically superior" your product is, you're going to be left in the dust, and like MapR you're going to go bust.

This is particularly the case when the bit about technical superiority is just a lie the company has fooled itself with. Recognising your own inferiorities is the first step to success.

[Ian Michael Gumby]

Re: @AC ... The software is quite good

Clearly you don't know what you're talking about.

Have you worked with all three vendors? Add to that AWS?

I'm not an SE these days but I could easily close a deal shooting down what you just said.

Snowflake? Yeah lets put a RDBMS in the cloud and claim its the greatest thing since sliced bread.

[Anonymous Coward]

"MapR was established in 2009 to commercialize then-fashionable open source Hadoop technology."

This is a common misconception. MapR's principal product is MapRFS, a distributed filesystem that is pretty neat for storage use cases and sells well to storage people buying storage things. MapR's "Hadoop" business was mostly marketing. In that respect MapR were in the business of selling proprietary, closed source clones of Hadoop ecosystem components that were less capable and several versions behind in compatibility terms. Did you want some HBase? Well they'd rather you bought some MapR-DB, which is similar but not quite the same. Did you want some Kafka? Well they'd rather you bought some MapR-ES, which is similar but not quite the same. Want some Kerberos for security? Well you can have some MapR tokens which are pretty much the same but don't integrate with anything else on the planet, not even the rest of the MapR stack.

I really feel sorry for MapR. Go back in time to 2010/2011 and they were undeniably ahead of the game, shipping technically awesome features. The quality of (some) of their people is beyond reproach. But the company, strategically, just went from insanity to insanity. Whether it was trying to compete in the on-premises storage array market, or trying to sell proprietary clones of Apache projects or most recently attempting to rebrand themselves a Kubernetes management company without any Kubernetes products or experience, every single strategic initiative has been absolutely, positively, bonkers.

MapR were sustained by a small number of huge customers. American Express, SAP, and a couple of Auto and Industrial firms had significant contracts or agreements, but not much else. Unfortunately those mega-customers are now all looking towards the cloud where MapR just cannot compete. They have run out of money and run out of road.

[Anonymous Coward]

@AC sorry had to downvote you.

The issue is that you really don't know as much about MapR as your think. MapR had more customers than your realize. In a lot of different areas. You would be surprised at some of the companies that run MapR.

There's more to running a software company than just building a good product. Hopefully MC Srivas learned some valuable lessons. (He was too conservatives at times)

MapR built a better file system and their MapR-DB is an extension of that. HBase was removed, but was supposed to be coming back because customers wanted it so they could land their existing infrastructure and then migrate to MapR-DB.

There were a couple of large miscues when it came to product development and management. This definitely hurt them but they actually had the edge when it came to Kubernetes and storage. Some of their ideas were definitely misguided. MapR was further along w Kubernetes than either HDP or Cloudera. In fact their MapR-FS made it easier to work w containers.

There's more that I could say about Ted Dunning and John Schroeder among others.

But the product isn't dead. If the company that has been talked about does buy MapR, there is definitely some synergy that would be interesting.

Again, posted Anon for the obvious reasons.

[Anonymous Coward]

MapR will sell

MapR will sell -- they have a lot of ARR and a captive customer base they can milk for 5-10 years. Question is -- how much will they go for? MapR fired all their sales staff a few months ago to prep for sale. They realized their technology, like Hadoop overall, is a dead end and the only thing to do is milk your captive customers for whatever you can with as few employees as possible. Perfect for some form of PE vulture capitalist.

$800 mil is a lot for MapR IMO, I guess JS was holding out for that magical billion dollar mark. I think he's going to wish he had taken it when he had the chance.