Reviewing 2016, Previewing 2017
Year-end reviews, and year-beginning previews, offer us an opportunity for sober reflection on the three fundamental questions of any cooperative, rational human endeavor: What do we know? What should we do? What can we hope?
I’ve written already about raising first institutional capital in 2016. That post is background for this one.
We made significant progress in Stardog qua product in 2016. Let’s review.
- 3 Stardog Engineers
- Released 4.0.3
- Evren Sirin made the first commit, 2 January.
- 10 Stardog Engineers
- Released 4.2.2
- Mike Grove made the last commit, 29 December.
In between January and December:
- 9 releases total (low by our historical standards)
- 1,143 commits, which is about 8% of all total commits to Stardog
- 2,811 files were changed
- 170,673 insertions
- 135,148 deletions
- Evren Sirin has the distinction of being the only Stardog engineer who removed more code than he added, about 6,600 lines. Not bad for the CTO!
These statistics don’t include anything that was started in 2016 but hasn’t yet been merged, including outstanding branches on memory management, cluster, and core server rewrite. There’s some chance they would move our total code for the year from positive to deficit, which we would clearly prefer.
In other words, we only increased the Stardog codebase by about 35,000 lines, most but not all of which is code. That includes some non-trivial highlights:
- reduced memory usage
- join-order optimizations and other query planner improvements
- hardened cluster
- native SPARQL parser
- stored queries
- improve Virtual Graph performance
- and 220 bug fixes
As of this blog post, the Stardog codebase is about 575,000 lines of code, a number that’s been reasonably fixed for the past few years, while we’ve added significant new user-facing capability to the system.
Okay, how did we do that? Seven new engineering hires, who collectively raise the sheer amount of engineering awesome around Stardog to new heights:
- Jess Balint (previously MySQL, Oracle) delivered BITES and is working on Virtual Graphs
- Stephen Nowell (previously BAH), who joins us from our services group, where he worked on NASA, to lock down tech support and work on the Stardog Web Console
- Alex Toktarev (previously NetCracker, Deutsche Bank, Hazelcast) is working on native memory management in Stardog
- John Bresnahan (previously Argonne, Red Hat, Dell) is working on cloud deployment and integration, including cluster management, as well as an unreleased (coming soon!) Stardog virtual appliance solution (codename “Starman”) built with Golang
- Paul Marshall (previously Argonne, Rackspace, Dell) is also working on cloud integration and testing, including a Stardog-specific Blockade/Chaos Monkey testing framework
- Pedro Oliveira (previously Stardog (!), Technicolor, Disqus) is working on search and spatial performance and was one of the first Stardog engineers originally before doing the “San Francisco thing” for a few years
- Scott Fines (previously NISC, Splice Machine) is working on distributed Stardog 6; he started on 2 January and got a PR accepted by 4 January…clearly a slacker!
These great developers join an engineering team and technical staff of which I could not be more proud:
- Mike Grove (founder), Stardog tech lead since the first commit; my first hire, ever, and that sure worked the fuck out!
- Evren Sirin (founder), a deep source of knowledge and insight who keeps everyone else running fast just to keep up–my second hire, ever.
- Pavel Klinov, whose integrity is exceeded only by his skill, ingenuity, and excellence of craft
- Al Baker, an experienced team lead and project manager
- Michael Soren, the most maintenance-free developer I’ve known
- Greg Coluni, comes to us from working on DOD systems, and who has everything he touches on lockdown
- Kate Belisle, our data scientist of the frozen north, who’s working next-gen R&D
- Kevin Long, a stalwart project manager who did amazing work at NASA and in the oil & gas industry before coming to us
We also made big progress on the revenue side of the business. We moved from perpetual licensing to annual subscription model in early 2016 and that transition went much more smoothly than I could have guessed.
While it is somewhat artificial, given the new baseline, our MRR increased by about 4,500% this year. Two hires on the sales team were key here:
- Jonathan Doan (previously Visual Sciences, Omniture, Limelight, eXelate, Tealium) is running global sales and has brought a new level of experience and professionalism to our enterprise sales game
- Mark Wood (previously Visual Sciences, Limelight, Tealium) is running EMEA enterprise sales and is, or so he claims, distantly related to British monarchy
We’re working on improving Stardog every day. From a software engineering perspective, we intend to return to the previous release pace, which was our historic norm, of about 2 releases per month.
From a product management perspective, we are working on improving the user experience and the power of Stardog by focusing on
- query performance (always)
- stability, especially in memory pressure situations, and in the cluster
- transacted writes performance
- scalability; look for big changes in 2017 here
- Stardog Studio (codename “Magnetar”), a native app for using, managing, and implementing Stardog
We will also be experimenting with our go-to market strategy. Maybe 30-day evaluations should be 45-day evaluations. What is the right mix of features for Stardog Developer, Enterprise, and Community? Should we have a non-commercial variant? We need to be very deliberate this year about finding an increasingly more efficient set of strategies and processes around how people evaluate and learn Stardog.
More New Hires
We’re making more new hires this year including another East Coast sales rep, as well as two front-end engineers to focus on Magnetar. Please get in touch if you are interested.
Marketing? Yeah, Marketing!
To date, we’ve never spent a penny on marketing, advertising, etc. I don’t say that to brag, because it’s surely nothing to brag about. I say it because, first, it’s true and, second, it’s interesting. All of our revenue growth to date has been organic, inbound growth.
The goal in 2017 is twofold: to increase organic, inbound growth dramatically by actively marketing Stardog as the solution to enterprise data unification; and to learn how to tell the Stardog story in an outbound sales process that complements our organic growth.
What do we know? We improve Stardog every day, both as a product and as a business, and we can attribute that growth to three facts:
- it solves a real need–data unification–that every big organization in the world has in spades
- it solves that problem in a way unlike any other enterprise offering
- the people responsible for the first two facts are very good at what they do and very passionate about doing it
What should we do? Insist on excellence from ourselves and help to elicit it from others, too. Improve the product and the user experience every single day. No zero days here. Kaizen forever, my people!
What can we hope? That markets are rational? Hmm, no, scratch that. We can hope for more good breaks than bad and that political instability around the globe diminishes instead of increases. And we can hope for opportunities every day to tell people the Stardog story.
Download Stardog today to start your free 30-day evaluation.
10 January 2017