![]() Of course, the shopping list does not stop here. Shipping was from China so it took some weeks to be delivered to Germany and I had to pay some customs fees, but it was still ok. The biggest disadvantages are, ‘only’ 1GB memory (the Pine64 has 2GB), ‘only’ 100Mbps Ethernet Port (the PineMbps) and Raspbian runs only in 32-bit. ![]() The other candidate was the RaspberryPI 3. If you need Wi-Fi, you have to buy the Wi-Fi 802.11BGN/BLUETOOTH 4.0 MODULE for 10$. The Pine64 does not have a Wi-Fi module build in. ![]() The Pine64 runs with an Allwinner A64 that has a 1.152 GHz 64-bit quad-core ARM Cortex-A53 with 2GB DDR3 Memory, a 1000Mbps Ethernet Port a size of 133mm x 80mm x 19mm, consumes 2.5 Watt and that all for only 29$. Yes, this does not exist but the best trade-off I could find was a Pine64. When I was looking for a suitable compute engine, I was looking for a compute engine that is cheap, has lots of 64-Bit cores, has lots of RAM, fast network connection, does not consume much energy and is small. So first things first - yes, of course, we need a compute engine. In this article I will show you, why I have chosen the Pine64 as a compute engine, how I build it together step by step and what challenges I have been faced. I also wanted to see how performant this cluster will be.īut to be very honest - I built it because I wanted to :) It is more the idea to create a relatively cheap cluster to play around with k8s and to see how k8s behaves if a node in the cluster fails. Of course, the answer is not that I ever want to create a cluster to be used as a production cluster. Why the heck should someone build a Kubernetes (k8s) cluster with six Pine64 when you can just use Google Cloud, Amazon Web Services or Microsoft Azure?
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |