This week, in our ‘Startup Spotlight‘ series, which features relatively under-the-radar but interesting European tech startups on a weekly basis, we highlight Estonian startup Plumbr, a Java performance monitoring tool that detects root causes of app performance issues.
Founded in 2011, Plumbr's beginnings stem from research that co-founder Vladimir Šor was working on for his PhD thesis. Since then, the startup has picked up some well-known ex-Skype backers and recently expanded its presence to the US.
Here’s a Q&A with co-founder Priit Potter on how Plumbr started, what makes it different from other performance monitoring services and his thoughts on the Estonian tech scene.
tech.eu: How would you describe Plumbr in a couple of sentences?
Priit Potter: Plumbr is a software monitoring solution that finds performance issues in Java applications. Plumbr can uniquely link these issues to their root cause in source code, which means that developers and administrators don’t have to manually troubleshoot slow applications any more.
Where did the idea for Plumbr originally come from?
The idea grew out of personal frustration with tracking down software performance errors. At Nortal, we were often faced with technical problems that in the end were maybe only caused by a couple of erroneous lines in the source code, but effectively brought big IT systems to their knees. These situations are always stressful – the business demands a quick solution because customers suffer, while the IT has no way to predict if they would solve the problem in a couple hours or weeks.
Our technical co-founders Vladimir Šor and Nikita Salnivkov-Tarnovski were often helping different Nortal customers who were wrestling with technical problems. They developed different scripts and hacks to help troubleshoot applications faster. At some point Šor started researching the field for his PhD thesis, which resulted in an algorithm that could automatically detect root causes of Java memory leaks – the first one of its kind.
That’s when we decided to build a product that would be able to analyse any performance problem that an IT system could have. Today, Plumbr covers all the hardest-to-troubleshoot performance problems, like locked threads, long Garbage Collection pauses and slow database operations.
How did you meet your co-founders? What were you each doing before starting the company?
We know each other well and worked together for more than seven years before founding Plumbr. We were all early employees at Nortal, now the largest IT services company in the Baltics. We had a unique chance to grow with the company, work with most of the enterprise IT customers in the regions (telecoms, banks, health sector, government) and intimately understand the challenges of running business critical IT solutions.
Who are your customers? Can you share some numbers?
Our customers are mostly the IT departments of companies from various sectors. Our paying customers include NASA, HBO, VMware, Dell and many others, in total more than a hundred companies.
Can you expand on the technology behind the product and how it works?
Plumbr monitors how IT applications internally behave under load. We have taught Plumbr to recognize patterns of faulty behaviour and then analyse the evidence to detect the root cause in source code. When the error is a common one and we have a known solution, Plumbr fetches step-by-step instructions from its built-in knowledge base of performance fixes.
As such, it contains numerous inventions, some of which we have even patented.
The source of these inventions however is real-world data. Starting from the first days when we launched the product, we have been analysing the performance data that Plumbr collects. The data is anonymous and doesn’t contain anything sensitive, but it’s a goldmine for our engineers.
We have spent many many years studying the statistics of tens of thousands of Java applications. Probably more than anyone in the world.
In retrospect, the combination of three main components allowed us create Plumbr:
1) Deep knowledge of the Java platform from our previous life in the IT services business 2) Data analysis and machine learning skills from the PhD research projects 3) Availability of data. Three months after initial launch we already had 1,000 registered users and were collecting Java performance data from all over the world
What makes Plumbr different from similar services?
Plumbr is able to automatically tell you what you need to change in your Java application to fix its performance issues.
There are many performance monitoring products that collect data from the live system and aggregate it on dashboards. Troubleshooting an actual problem however is still a manual task and usually calls for mastering tens of specialist tools. Plumbr automatically troubleshoots the problem and explains you its roots in source code.
How big is the Plumbr team at the moment? Where does the company have offices?
We have 20 people in total. Most of them are in our Tallinn and Tartu offices in Estonia.
Two months ago, we opened an overseas sales office in Boston that currently hosts one team member but is destined to grow.
How would you characterise the Estonian tech scene? What are some strengths and weaknesses?
It is great to run a tech company in Estonia – up to a certain size. At some point when the team grows to say more than 200 people, you will start to have trouble finding good people because the population is just so small. But for starting a company I think Estonia is a perfect location.
There is a lot of engineering talent with a 'let’s just give it a shot'-attitude. Also, the environment is very nerd-friendly with all the perks usually attributed to Estonia (nationwide cheap 4G coverage, free wifi almost everywhere, fully electronic communication with government agencies, etc). There are already three foreign engineers in our Estonian offices, two of whom moved to the country solely to join our team.
Of course, there are also disadvantages as are everywhere. One of the aspects that affects Plumbr is that we lack experienced tech product marketing and sales people. Many successful Estonian tech startups have built their sales and marketing teams outside the country, which makes it even slower for their experience to infuse into the local community. Until then, you can get around the problem by building an international team too.
Also read: Tallinn tech hotspots: a look inside Garage48 HUB and the Latitude59 conference (photo essay)
How much funding have you raised so far? Any plans to raise more? We have raised €1.5 million and currently have enough runway to our tastes. We are carefully focusing on product development and sales at the moment.
Can you talk about your business model? Has your thinking on monetisation evolved since the start? Our customers benefit from having an overview of their Java servers and being able to fix each of them fast when something happens. We charge a subscription fee of $139 per month or $999 per year for each Java server (JVM) that Plumbr monitors. We also have a separate product for companies who cannot use SaaS services because of security concerns that starts at $10,000 per year.
We have always known that we wanted to build a subscription business, but the pricing model itself has changed quite a bit throughout our history. When we launched the first version of Plumbr, it was actually targeted towards developers and marketed as a troubleshooting tool. Back then we used to charge for single developer and team use. We have also tested per-incident pricing.
But as Plumbr evolved into a performance monitoring solution we finally settled with monthly and yearly subscriptions that is more common in the monitoring field.
What's the company's plan for the next 12 months?
Our vision for the current product is to automatically detect the reasons for every performance hiccup or decline. While Plumbr already covers the ones that are the hardest to detect manually, there is a number of possible technical problems that Plumbr still misses.
We will expand the functionality in two dimensions: 1) Add more types of performance problems that Plumbr detects 2) Closely watch the quality of existing incident alerts and improve if we can make them more useful.
Our core metric is: Can the Plumbr user solve the performance issue with the information that they get from Plumbr incident report?
This interview has been edited and condensed.
Also read:
Republic of Estonia invites the world to apply for Estonian 'e-Residency' online
Weekdone, the Estonian startup that wants to help teams work better together
Estonia’s Pipedrive raises $9 million to bolster its online CRM software
How Fortumo built a global payments business out of Estonia
For more from the Startup Spotlight series:
- Meet Radiant Games, the Icelandic startup aiming to spark children's interest in coding
- Finland’s Meetin.gs wants to fix scheduling meetings with ‘SwipeToMeet’
- Toothpick, the UK startup that wants to take the pain out of booking a dentist
- Blinkist summarises books so you don’t have to read the whole thing
Images courtesy of Plumbr.
Would you like to write the first comment?
Login to post comments