Databases – How do you architect your databases?
Many DBA’s will architect their databases in two ways, one for Performance and one for resiliency. More than likely you are playing Tetris with your database configuration and architecture. What do I mean by this, are configuring your data, log, temp, drives/mounts on your host? Below is a “MS Windows example, but the same applies to Linux and other OS’s by use of mounts….”
The goal is to separate the very random nature of your data vs the very sequential nature of your logs, this also allows you focus on ramping the number spindles behind each for performance. Your setup will look something like this, this could be ultra fast NVMe disks (DAS) or even legacy high performance storage.
Let’s simplify this on Pure Storage ….and get some massive performance gains!
Above – you can see how database administrators are playing Tetris with there data, log, and temp files. There is a simpler way to to this with a massive performance boost! Yup – Pure Storage uncomplicates this natively. Check this out.
Take the above diagram and let’s migrate that data to a Pure Storage array. When you setup a Pure Storage FlashArray, you do not have to build it with the traditional disk assignments i.e. all DFMs on the array are used by the array. UNCOMPLICATED! What’s better – Pure does the hard work a DBA used to do to plan for performance and resiliency natively on the array.
With this configuration you get the performance capacity of the entire array to support this workload. This change has proven to be a 40x increase in performance.
Credit goes to Anthony Nocentino and Marsha Pierce: Architecting Databases for Performance
Be the first to comment