Nice tutorial on Linux kernel.
Nice reference on how to create swap space on an SD card.
vm.min_free_kbytes setting allows kernel to compute a minimum value in the low memmory zone and reserve it for its own use.
To read the current value
[Fri Mar 24 15:18:05 root@system:~]$sysctl vm.min_free_kbytes
vm.min_free_kbytes = 1888
To set new value of 4MB. Based on your system requirement configure the value. Ideally 1.8% is good enough.
echo 4096 > /proc/sys/vm/min_free_kbytes
To configure permenantly on every reboot add the below entry to /etc/sysctl.conf
A Quick reference guide for C and CPP programmers for the operator precedence.
Scalability is a system’s ability to swiftly enlarge or reduce the power or size of computing, storage, or networking infrastructure thereby adapting to resource demands, optimizing costs, and improving the operations team efficiency.
Scaling up is adding further resources, like hard drives and memory, to increase the computing capacity of physical servers.
Scaling out is adding more servers to your architecture to spread the workload across more machines.
A good indicator of when to scale up is when your workloads start reaching performance limits, resulting in increased latency. Only the throughput can be increased, but when you hit any hardware bottleneck to extend the hardware, we need to scale out.
Based on use case , operational cost, rack space and forecasted workload demands, customer can choose to prefer scale up or scale out.
Source: Scale Up vs Scale Out | Data Center Infrastructure | ServerWatch
Form Factor: U.2 2.5″ 15mm
Storage Interface: PCIe Gen3 x4 NVMe
NAND: Micron 64 layer 3D TLC NAND
PRO is read intensive, available in 3.84 TB, 7.68 TB and 15.36 TB
MAX is Mixed use, available in 3.2 TB, 6.4 TB and 12.8 TB.
2 million device hours
Seq. read 14W Max
Seq. write 21W Max
Seq. Read(MB/s) – 3500
Seq. Write (MB/s) – 3100-3500
Rand. Read(K IOPS) – 835-850
Rand Write(K IOPS) – 105-310
Average Latency – 86us for rand read and 11us for rand write.
Workload optimized for high speed random reads and writes. Good for Tier1 caching drives.
Target Market segments
Best fit for below applications.
AI/ML/DL training and caching.
OLTP and HFT
No SQL database and large object database
Some basics on Linux Software RAID. There are good references on the same in the same wiki.