6 comments

[ 4.4 ms ] story [ 31.6 ms ] thread
They are good machines. As long as you don't run out of CPU credits you won't really notice poor CPU performance for normal workloads.
*for high workloads. Yep. When they run out of credits (which happens a lot less often than our clients usually predict), the system handles the workload with the baseline performance. Depending on how well the software running on the server is designed, that can mean simply a few delays on the requests of the users; it doesn't have to break functionality.
The only time we use other instances is when we really do need that all-day CPU performance. Thank you for your article!
Glad you liked it. If you share it as well, you may have five internet thanks-money from me. Actually I'm making this offer available to all the HN users.
I use t2 instances for everything I can. Only on-demand build slaves and job processors get M or C types, and those are killed as soon as their work is done to save money.
very informative, thanks! let's just say I'm changing servers ;)