1. In the Cloud menu, go to Kubernetes, find the required cluster and click its name.
2. Go to the Pools section and click Add pool.
3. The new window will expand.
Enter the pool name, set the minimum nodes and maximum nodes for autoscaling. Autoscaling will regulate the number of nodes within the specified limits by removing unnecessary nodes during low load and deploying new ones during high loads.
For Type, select the type of a worker node: a virtual instance or a bare metal server.
For virtual instances, you should select its flavor, disk size in GiB and the disk type.
Standard. These are machines on which the amount of memory in GB is 2–4 times higher than that of vCPUs.
vCPU. These are machines on which the amount of vCPUs equals the amount of memory in GB.
Memory. These are machines on which the amount of memory in GB is much higher (up to 8 times) than that of vCPUs.
High Frequency. These are machines with a high CPU clock speed (3.7 GHz in the basic configuration).
SGX. These are machines that support Intel SGX technology.
High IOPS SSD. Network SSD block storage designed for latency-sensitive transactional workloads (60 IOPS per 1 GiB; 2.5 MB/s per 1 GiB). The IOPS performance limit is 9,000. The bandwidth limit is 500 MB/s.
Standard. High-performance network SSD block storage (6 IOPS per 1 GiB; 0.4 MB/s per 1 GiB). The IOPS performance limit is 4,500. The bandwidth limit is 300 MB/s.
Cold. Network HDD block storage. The maximum number of IOPS is 1,000. The bandwidth limit is 100 MB/s.
Ultra. Network HDD block storage, similar to Cold disks. The maximum number of IOPS is 1,000. The bandwidth limit is 100 MB/s.
For bare metal servers, you also should select a flavor:
High-frequency. These servers are single-socket servers equipped with 2288G/2388 CPUs, suitable for hosting applications that require high processor frequency.
Infrastructure. These servers are multi-core, multi-socket configurations designed for hosting applications that demand a high number of cores and are optimized for multithreading.
Make sure the Autohealing nodes toggle is on to enable automatic recovery of failed nodes. The option monitors node statuses. When it detects a non-working node, the autohealer initiates replacement. If one of the machines fails, the application will not stand idle: the node will be replaced, and the app will keep working.
The pool has been created.
Was this article helpful?
Discover our offerings, including virtual instances starting from 3.7 euro/mo, bare metal servers, AI Infrastructure, load balancers, Managed Kubernetes, Function as a Service, and Centralized Logging solutions.