From df5fbd9e08d3c0e6782ff556186b1f75c7c91af7 Mon Sep 17 00:00:00 2001 From: schmidt-scaled <82834682+schmidt-scaled@users.noreply.github.com> Date: Fri, 19 Jun 2026 16:03:21 +0300 Subject: [PATCH 1/2] Update cloud-instance-recommendations.md --- .../cloud-instance-recommendations.md | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/docs/deployments/deployment-preparation/cloud-instance-recommendations.md b/docs/deployments/deployment-preparation/cloud-instance-recommendations.md index 741814fd..b74ce6fd 100644 --- a/docs/deployments/deployment-preparation/cloud-instance-recommendations.md +++ b/docs/deployments/deployment-preparation/cloud-instance-recommendations.md @@ -12,9 +12,9 @@ other instance types as long as they fulfill the system requirements. Simplyblock can work with local instance storage (local NVMe devices) and Amazon EBS volumes. For performance reasons, Amazon EBS is not recommended for high-performance clusters. -!!! critical +!!! important If local NVMe devices are chosen, make sure that the nodes in the cluster are provisioned into a placement group of type - _Spread_! + _Spread_! Otherwise you have no guarantee that multiple nodes are not located in the same failure domain. Generally, with AWS, there are three considerations when selecting virtual machine types: @@ -22,7 +22,8 @@ Generally, with AWS, there are three considerations when selecting virtual machi - Locally attached NVMe devices - Network performance (dedicated and "up to") -Based on those criteria, simplyblock commonly recommends the following virtual machine types for storage nodes: +Based on those criteria, simplyblock commonly recommends the following virtual machine types for storage nodes +***in disaggregated clusters***: | VM Type | vCPU(s) | RAM | Locally Attached Storage | Network Performance | |-----------------|---------|--------|--------------------------|---------------------| @@ -35,6 +36,10 @@ Based on those criteria, simplyblock commonly recommends the following virtual m | _i4i.8xlarge_ | 32 | 256 GB | 2x 3750 GB | 18.75 GBit/s | | _i4i.12xlarge_ | 48 | 384 GB | 3x 3750 GB | 28.12 GBit/s | +For hyper-converged EC2 kubernetes worker nodes (EKS, k3s or other), any instance with a local SSD can be chosen. +Keep the [minimum system requirements](./hardware-requirements.md)in mind. At least 8 vcpu are required for Simplyblock. + + ## Google Compute Engine Recommendations In GCP, physical hosts are highly shared and sliced into virtual machines. This is true not only for CPU, RAM, and @@ -45,7 +50,6 @@ simplyblock. !!! critical GCP deployments can currently not be used in production as GCP does not guarantee 4K write atomicity on their SSDs! - Therefore, a "sudden power off" is not safe, data corruption may occur! Generally, with GCP, there are three considerations when selecting virtual machine types: @@ -54,10 +58,6 @@ Generally, with GCP, there are three considerations when selecting virtual machi - The size of the locally attached NVMe devices (SSD Storage) - Network performance -!!! critical - If local NVMe devices are chosen, make sure that the nodes in the cluster are provisioned into a placement group of - type _spread_! - Based on those criteria, simplyblock commonly recommends the following virtual machine types for storage nodes: | VM Type | vCPU(s) | RAM | Additional Local SSD Storage | Network Performance | From e19a3096dcb1f7b4dcfd67dcd4d9d4e1eca01113 Mon Sep 17 00:00:00 2001 From: noctarius aka Christoph Engelbert Date: Fri, 19 Jun 2026 15:06:54 +0200 Subject: [PATCH 2/2] Update cloud-instance-recommendations.md --- .../cloud-instance-recommendations.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/deployments/deployment-preparation/cloud-instance-recommendations.md b/docs/deployments/deployment-preparation/cloud-instance-recommendations.md index b74ce6fd..86bdf039 100644 --- a/docs/deployments/deployment-preparation/cloud-instance-recommendations.md +++ b/docs/deployments/deployment-preparation/cloud-instance-recommendations.md @@ -14,7 +14,7 @@ Amazon EBS is not recommended for high-performance clusters. !!! important If local NVMe devices are chosen, make sure that the nodes in the cluster are provisioned into a placement group of type - _Spread_! Otherwise you have no guarantee that multiple nodes are not located in the same failure domain. + _Spread_! Otherwise there is no guarantee that multiple storage nodes won't be located in the same failure domain. Generally, with AWS, there are three considerations when selecting virtual machine types: @@ -36,9 +36,9 @@ Based on those criteria, simplyblock commonly recommends the following virtual m | _i4i.8xlarge_ | 32 | 256 GB | 2x 3750 GB | 18.75 GBit/s | | _i4i.12xlarge_ | 48 | 384 GB | 3x 3750 GB | 28.12 GBit/s | -For hyper-converged EC2 kubernetes worker nodes (EKS, k3s or other), any instance with a local SSD can be chosen. -Keep the [minimum system requirements](./hardware-requirements.md)in mind. At least 8 vcpu are required for Simplyblock. +For hyper-converged AWS EC2-based Kubernetes workers (EKS, k3s or other), any instance with a local SSD can be chosen. +Keep the [minimum system requirements](./hardware-requirements.md)in mind. At least 8 vCPUs are required for simplyblock. ## Google Compute Engine Recommendations