Service statistics collection with Flussonic
Hello everyone, I want to tell you about our first service that we are going to provide to our customers: collection of statistics and reports .
For many years we have only sold software, now we have prepared a service for collecting statistics and providing reports for launch. We take Flussonic, launch it on their servers, subscribers watch videos from these Flussonic instances and create recordings of viewing sessions.
These sessions just merge into our service and we will show the reports made on them in your account .
Turns on easily: there is a large “Enable statistics” button in your account. After clicking in a minute, Flussonic will start sending us sessions and they will be stored securely.
')
At the moment we will give only the most basic reports: distribution by country, by browser (and other devices), by channels. We can give both a raw list of sessions and aggregate similar ones:
For two weeks of closed testing, we have a modest billion records so far, but we are sure that we can handle 100 times more: we can very quickly show even unique sessions, and this is a very expensive report. Specifically, in this screenshot, I showed how gluing similar sessions into unique ones, i.e. if from the same address, with the same User Agent, they tried to watch the same thing for a narrow period of time, then most likely this is one session, which is simply poorly accounted for.
We have already faced a very unpleasant problem of parsing the User Agent header: after all, we have a lot of set-top boxes and other similar pieces, but they are not in common browser databases.
An unexpected problem with MaxMind also surfaced: you can't just buy a full base from them and show you the search results for it, it requires a special license from $ 30,000.
But all this is so, the problems to be solved and we are now working with all this.
A separate moment with a javascript script: this is a SPA React application that lives in a separate microservice along with the service itself. We are not the first component to do this: javascript, css live and develop with the backend of the service itself and the user is connected by simply inserting javascript. There are tricks with authorization, because we want not only to report the account, but at the same time its access rights, so the authorization session opens when a user logs in to his personal account through server-to-server interaction.
We have a lot of plans for the future:
If someone wants to try, feel, but has not yet started Flussonic, then we are happy to give a free trial .
And, by the way, it is important: we are going to do the collection of statistics and reports for free!
For many years we have only sold software, now we have prepared a service for collecting statistics and providing reports for launch. We take Flussonic, launch it on their servers, subscribers watch videos from these Flussonic instances and create recordings of viewing sessions.
These sessions just merge into our service and we will show the reports made on them in your account .
Turns on easily: there is a large “Enable statistics” button in your account. After clicking in a minute, Flussonic will start sending us sessions and they will be stored securely.
')
At the moment we will give only the most basic reports: distribution by country, by browser (and other devices), by channels. We can give both a raw list of sessions and aggregate similar ones:
For two weeks of closed testing, we have a modest billion records so far, but we are sure that we can handle 100 times more: we can very quickly show even unique sessions, and this is a very expensive report. Specifically, in this screenshot, I showed how gluing similar sessions into unique ones, i.e. if from the same address, with the same User Agent, they tried to watch the same thing for a narrow period of time, then most likely this is one session, which is simply poorly accounted for.
Few details
We have already faced a very unpleasant problem of parsing the User Agent header: after all, we have a lot of set-top boxes and other similar pieces, but they are not in common browser databases.
An unexpected problem with MaxMind also surfaced: you can't just buy a full base from them and show you the search results for it, it requires a special license from $ 30,000.
But all this is so, the problems to be solved and we are now working with all this.
A separate moment with a javascript script: this is a SPA React application that lives in a separate microservice along with the service itself. We are not the first component to do this: javascript, css live and develop with the backend of the service itself and the user is connected by simply inserting javascript. There are tricks with authorization, because we want not only to report the account, but at the same time its access rights, so the authorization session opens when a user logs in to his personal account through server-to-server interaction.
Plans
We have a lot of plans for the future:
- make a separation in the statistics for cross-server traffic and traffic to clients
- to help catch those who fumble the account to friends, as well as those who are engaged in restoring
- add information about the depth of viewing the archive, etc.
If someone wants to try, feel, but has not yet started Flussonic, then we are happy to give a free trial .
And, by the way, it is important: we are going to do the collection of statistics and reports for free!
Source: https://habr.com/ru/post/330898/
All Articles