April is a wonderful month when everywhere in Europe — from Hungary (earlier) to Sweden (later) you can see green spots of a newly born flora. These picturesque landscapes are a balm for the soul after a totally white winter. Unfortunately this spring is poisoned with the war that is now happening in Ukraine with hundreds of people including civilians dying each day. The active phase of the Battle for Donbass is now on and flespi is also participating in it.
We do not have soldiers but we have technology and we generate some revenue flows. Being clearly on the Ukrainian side we did two things in order to win the battle.
First, we are providing flespi for free for companies from Ukraine. And if for the very first month, in March, there was almost zero activity from their side, now I can clearly see that economics is still struggling and our services are in growing demand. Especially in the agriculture sphere and later, I hope soon, it will be a construction sphere. For us it is not that difficult to provide great service and technology for free. For Ukrainian companies this is a good support and more money in-house for any business activity. And it really works.
Second, we moved out of Russia and closed contracts with all Russian companies on April 30. For us this resulted in a loss of approximately 30,000 connected devices (this equals to our growth in connected devices for two months). The same with the datacenter — it took us a year to open a dedicated flespi region in Russia and a couple of months to close it. But now there are no direct revenue streams from Russia and flespi technology on the commercial level is unavailable in Russia except when delivered via the 3rd party — we do not limit services regionally and do not have any plans to do so.
Of course this is small pressure. But I believe in the cumulative effect — if thousands of businesses and millions of people worldwide impose such a pressure against Russia and such an assistance towards Ukraine — together we will help the good beat the evil.
OK, now back to flespi. In April our uptime was 99.9934%. Everything was great until the very end of the month and on the night of April 29th our uplink provider had some troubles with the connectivity due the failed link between routers and invalid configuration in the primary/secondary link switching system. You may read a detailed report in the NOC.
By the way, our uplink provider Zylon with whom we have been working since 2008 is now establishing a second datacenter for our purposes that will allow us to create an infrastructure datacenter level redundancy service. As any major change this may slightly affect the network stability until we debug all the issues with hardware and its configuration, but the result is definitely worth it. We plan to launch a second datacenter in production by the end of this year.
Among our long running projects we finally finished in April our migration from seconds to microseconds granularity in device messages. Development started in July, most systems migrated in September and only analytics remained on seconds granularity. Now analytics is also operating with timestamps and durations in microseconds.
One more new feature for analytics is the “inactive” interval selector that is used to simplify the detection of inactive devices in realtime and historical mode.
We integrated two new protocols — ibag and teltonika-mobility-mqtt.
We also published a couple of articles that can inspire you to develop your own scripts for data post-processing using PVM technology: hex string parsing sample and GPGGA NMEA sentence parsing sample.
One new feature that you can not touch but which is an important user experience extension is a “soft” restart of our REST API services. Now when we are installing upgrades (which we usually do a few times per week) our REST API service is trying to finish processing of active requests before the shutdown within a certain amount of time. For your REST API consumers it just means that you should not notice our upgrades anymore except for very specific extreme cases.
And last but not least are the new debug features in the flespi panel — now it is possible to access logs for a specific token, MQTT session or subaccount.
In May we will continue to improve internal components of flespi and prepare the infrastructure to operate in multiple datacenters.