Hands-On Azure for Developers

Performance

One important feature of many storage solutions is their performance. In general, we expect that our databases will work without a problem whether the load is low or high and a single record is big or small. When it comes to ADLS, you have to take into account the following factors:

• Parallelism: As stated in the documentation, it is important to ensure, that you provide a certain level of parallelism when performing reads/writes. The ideal number of threads per one core is defined as 8-12.
• File size: While different data analytics solutions may work differently with different file sizes, it is also important to know that ADLS also has an optimal file size to work with. As it is based on HDFS and leverages the POSIX model for ...