Recherche avancée

Sur d’autres sites (3410)

• Handling high volume traffic and traffic peaks with Matomo just got easier

  16 avril 2018, par Matomo Core Team

  When you use the self-hosted version of Matomo on-premise instead of the Matomo cloud-hosted solution, you may experience some traffic peaks on your Matomo server when the traffic volume on your websites increases. For example, every day at a certain time you might receive two or three times the amount of traffic that usually visits your website. This can have many negative impacts, including :

  • Slow loading time for your JavaScript tracker (piwik.js) which in turn may slow down your website giving your users a poor experience. Also you may see less page views in Matomo because by the time the tracker is loaded on your website, the user has already moved on to another page.
  • Some tracking requests might be simply ignored at some point because your server might not be able to handle any tracking requests anymore which results in many untracked visits and page views.
  • You may need additional servers only to handle traffic peaks which results in increased server costs, maintenance work and maintenance costs.

  The solution

  Handling traffic peaks has been possible with Matomo for years using the Queued Tracking plugin. When this feature is enabled, tracking requests are put into a queue instead of being processed immediately. Then when a job is running separately it takes the requests out of the queue and processes them. This brings various benefits.

  Faster tracking

  It improves the tracking speed on your server by a factor of 5 to 15. So for example, instead of a tracking request taking 50ms, it takes only 5ms. This means your server will be able to handle a lot more concurrent requests compared to the traditional tracking and is likely to survive traffics peaks much more likely without any trouble at all.

  Faster processing

  When a request is queued, the request still needs to be processed eventually. Because the Queued Tracking solution can take multiple tracking requests out of the queue at once and process them in one go, the processing speed increases massively as well. This is because by default each tracking request has to bootstrap Matomo and do a lot of things again and again which takes quite a bit of time (you’d be surprised). Instead, many things can now be cached and don’t have to be done multiple times. As a result, your server can process tracking requests much faster and needs less resources overall which in turn reduces cost and trouble.

  Queued Tracking is now easier to set up

  In the background, Queued Tracking has been using Redis, an in-memory database. While Redis is very fast, it’s not simple to setup and maintain it. Especially when it comes to making Redis “highly available” and when you need to scale your Redis. Also, your servers will need a lot more memory for Redis as all queued tracking requests are stored in memory.

  One click setup

  We have now added support for a MySQL database so you can activate Queued Tracking with a simple click. What used to take hours or maybe even weeks to set up and a lot of maintenance, can now be cut down to seconds. Queued Tracking will then simply reuse the database that you have been using all along for storing all your visits. A side benefit is that your server won’t need more memory and all queued tracking requests even survive a server reboot.

  Both Redis and MySQL are now supported in Queued Tracking. If you do have experience with managing Redis, we still recommend using this solution as it’s likely a bit faster. However, in most cases the MySQL solution should work just as well.

  Further improvements

  We have made various other improvements for Queued Tracking that increases the performance and you can now be notified when the number of queued tracking requests reaches a certain threshold. View the changelog for a list of all changes.

  Learn more

  We have been setting up Queued Tracking multiple times when it comes to high volume traffic or dealing with peaks and are amazed by the results. Often, we can even reduce the overall amount of needed servers.

  If this sounds like something that could be beneficial to you, we recommend you have a look at the Queued Tracking page and also check out the FAQ. You might be also interested in learning how to configure Matomo for speed.

  Need help with setting up, maintaining, or scaling Matomo ? Get in touch now.

  The post Handling high volume traffic and traffic peaks with Matomo just got easier appeared first on Analytics Platform - Matomo.

• Adding watermark bitmap over video in android : 4.3's MediaMuxer or ffmpeg

  24 novembre 2018, par Alin

  Here is my scenario :

  • Download an avi movie from the web
  • Open a bitmap resource
  • Overlay this bitmap at the bottom of the movie on all frames in the background
  • Save the video on extarnal storage
  • The video length is 15 seconds usually

  Is this possible to achieve using MediaMuxer ? Any info on the matter is gladly received

  I’ve been looking to http://bigflake.com/mediacodec/#DecodeEditEncodeTest (Thanks @fadden) and it says there :

  "Decoding the frame and copying it into a ByteBuffer with
  glReadPixels() takes about 8ms on the Nexus 5, easily fast enough to
  keep pace with 30fps input, but the additional steps required to save
  it to disk as a PNG are expensive (about half a second)"

  So having almost 1 sec/frame is not acceptable. From what I am thinking one way would be to save each frame as PNG, open it, add the bitmap overlay on it and then save it. However this would take an enormous time to accomplish.

  I wonder if there is a way to do things like this :

  1. Open video file from external storage
  2. Start decoding it
  3. Each decoded frame will be altered with the bitmap overlay in memory
  4. The frame is sent to an encoder.

  On iOS I saw that there a way to take the original audio + original video + an image and add them in a container and then just encode the whole thing...

  Should I switch to ffmpeg ? How stable and compatible is ffmpeg ? Am I risking compatibility issues with android 4.0+ devices ? Is there a way to use ffmpeg to acomplish this ? I am new to this domain and still doing research.

  ---

  Years later edit :
  Years have passed since the question and ffmpeg isn’t really easy to add to a commercial software in terms of license. How did this evolved ? Newer versions of android are more capable on this with the default sdk ?

  ---

  Some more time later edit

  I got some negative votes for posting info as an answer so I’ll edit the original question. Here is a great library which, from my testing does apply watermark to video and does it with progress callback making it a lot easier to show progress to the user and also uses the default android sdks. https://github.com/MasayukiSuda/Mp4Composer-android

  This library generate an Mp4 movie using Android MediaCodec API and apply filter, scale, and rotate Mp4.

  Sample code, could look like :

  new mp4Composer(sourcePath, destinationPath)
  
          .filter(new GlWatermarkFilter(watermarkBitmap)
  
          .listener(){
  
                @Override 
  
                private void onProgress(double value){}
  
  
  
                @Override 
  
                private void onCompleted(double value){
  
                    runOnUiThread( () ->{
  
                       showSneakbar
  
                    }
  
                }
  
  
  
                @Override 
  
                private void onCancelled(double value){}
  
  
  
                @Override 
  
                private void onFailed(Exception e){}
  
  
  
          }).start();

  Testing on emulator, seems to work fine on android 8+ while on older generates a black video file.However, when testing on real device seems to work.

• MP3 (MPEG I) chunk decoder for Python

  3 juillet 2022, par Clement

  I've been searching for a few days and trying many different libraries including PyDub, python_mp3_decoder (segmentation faults), pymad, but have had basically no luck in finding a library for Python that would allow me to decode a MP3 Stream (from a internet radio ; icecast) on the fly and treat it like microphone input (e.g. PyAudio stream).

  



  I am trying to get a stream of decoded audio to use with PyAudio for an acoustic fingerprinting project. The other catch is I cannot use PyMedia which has been suggest here on stackoverflow before since it is not supported on the Mac, nor has it been updated in more than 12 years.

  



  Any input on how I could decode a MP3 Stream in real time in python would be much appreciated ! Thanks !