For those who don’t know what Kickfire was, here’s a quick explanation: a dedicated appliance designed to run a modified version of MySQL, with massive amounts of RAM (the unit I had the pleasure to test had 500GB of RAM, and it was the “smallest” in the family). So, you can imagine one of its key features was loading the entire database into memory. Another striking feature (or at least it used to be) was that it came with an SQL chipset. That’s right! A chipset responsible for optimizing SQL statements (commands or instructions, if you prefer).
My first contact with Kickfire was in 2009, when I got to know the product at the MySQL UC conference and visited the company’s facilities (same name) for a more in-depth lab.
The product itself was excellent. And even beautiful! An impressive, respectable cabinet. The performance was like: “Are you kidding me?” I’d say Houdini had something to do with that performance. Just for a hypothetical comparison, a query that took 15 minutes on a DL385, 2 Xeon 3GHz, 16GB RAM, 15K RPM disks, took just over 40 seconds on Kickfire. That’s a hell of a difference! This was on a database of about 300GB. Of course, I made sure the databases were identical and the MySQL tuning was at its best.
The two engineers who accompanied me knew all the ins and outs of the appliance, MySQL, quantum physics, and astrology. I mean, the technical quality of the people involved in the project was fabulous.
What was Kickfire’s target? Data Warehouse. OLAP.
It’s worth mentioning that Kickfire would be just another player in a very competitive market with a bunch of solutions, as diverse as: software-based, hardware-based, hardware and software combined, plus a dose of “prayers”.
The truth is, as much as I liked the product, the “pretty” cabinet, and the logo… The company collapsed in August this year, and even the website was taken down. I tried contacting by email and phone, but all I got was: “the company is restructuring.” Sometimes I struggle with English, just like with Portuguese. There are so many words, rules, ordered sentences… subordinate clauses… Whew, that’s why I prefer “binary speak,” dude — it’s either 0 or 1, no middle ground. I kept wondering: “What the heck does restructuring mean?” I have no idea!
The fact is the company took a “shutdown +F”… hahaha. Please don’t look for the +F option in the shutdown man page. Use the force to figure out what it means.
Anyway, it’s a shame. I saw rumors that Teradata either bought or was negotiating to buy the brand, the engineering team, and the technology. All that through the usual “fire sale” model.
What makes such a good product fail so spectacularly? Here are the guesses of a professional guesser:
a) We need to accept that we are technicians. Not all of us are Jobs, Sergeys, Pages, Elissons, or Zuckerbergs. We brush bits & bytes, and if we want to undertake a killer idea, in the galaxy algorithm, we need to partner with competent people in marketing & management. And that was missing at Kickfire;
b) Price. At the end of the day, companies buy price. Quality, resilience, performance, and support must be implicit in products of this nature. Price is the detail. Do you think a device like this costing USD 50K in the United States is reasonable? With that money, you could buy two Mercedes in Uncle Sam’s land. It’s not affordable for most medium companies. There are other competing solutions at much more competitive prices;
c) MPP (massively parallel processing) was a missing feature in Kickfire. Devices for processing large amounts of data must have this as a “feature zero,” meaning it shouldn’t even be listed as a feature, it should be standard;
d) MySQL not so mysql after all: The MySQL running on Kickfire was a proprietary, closed, transformed version of MySQL. They turned Sakila into a pink dolphin. Can you believe that? Needless to say, their MySQL was not 100% compatible with our MySQL.
e) Last but not least, sales and customer service. Has anyone tried to buy or get commercial info about the product? No? Lucky you! Total collapse. There was no defined sales department. Like any Silicon Valley startup, maybe the cleaning lady was better with customers than the sales team, but she didn’t know bits & bytes.
Okay, but these are just guesses… The fact is: it’s a shame. It was a great product! There are other excellent options such as Teradata, Netezza, Dataupia, etc., and others more related to our ecosystem like InfoBright, Virident, InfiniDB, etc.
Some might wonder: “Why didn’t Alexandre include Gates in item a)?” The answer is simple: he made a fortune with IT but isn’t necessarily an expert on the subject. Right? After all, who likes the blue screen?
