251 lines
8.3 KiB
Markdown
251 lines
8.3 KiB
Markdown
_discoverable: no
|
|
---
|
|
_hidden: no
|
|
---
|
|
title: Managed Kubernetes Clusters
|
|
---
|
|
subtitle: Scalable and climate friendly
|
|
---
|
|
headline1: Scalable
|
|
---
|
|
headline2: Managed
|
|
---
|
|
headline3: High Performance
|
|
---
|
|
header_background_color: #13410B
|
|
---
|
|
header_text_color: text-light
|
|
---
|
|
nav_classes: navbar-dark
|
|
---
|
|
image:/u/image/cards/django-hosting-best.jpg
|
|
---
|
|
feature1_title: Scalable
|
|
---
|
|
feature1_text:
|
|
|
|
Your Kubernetes cluster can start from 3 nodes and grow to any size. Every
|
|
node that you add to your cluster contributes additional CPU, RAM and
|
|
persistent storage.
|
|
---
|
|
feature2_title: Managed
|
|
---
|
|
feature2_text:
|
|
The Kubernetes cluster and the operating system are maintained by
|
|
ungleich. You can deploy your applications worry free.
|
|
|
|
---
|
|
feature3_title: Dedicated
|
|
---
|
|
feature3_text:
|
|
Your Kubernetes cluster is running on dedicated hardware. No
|
|
interruption, no rescheduling, no peer pressure. All resources are
|
|
100% usable by you.
|
|
---
|
|
feature4_title: High Performance
|
|
---
|
|
feature4_text:
|
|
The kubernetes cluster is based on AMD Ryzen CPUs to provide 12
|
|
threads per node and 64GB RAM. The operating system is stored on an
|
|
SSD and your data is stored on a second SSD.
|
|
---
|
|
feature5_title: Environment friendly
|
|
---
|
|
feature5_text:
|
|
Each Kubernetes node has been tested to consume less than 100W at
|
|
maximum performance. While our data center is using 100% renewable
|
|
energy, we try to even make an extra effort to reduce impact on the
|
|
environment.
|
|
---
|
|
feature6_title: 100% IPv6
|
|
---
|
|
feature6_text:
|
|
The Kubernetes cluster is ready for the IPv6 world: all nodes are
|
|
running on an IPv6-only network. For IPv4 reachability we offer you
|
|
dedicated mapped IPv4 addresses.
|
|
|
|
---
|
|
content1_text:
|
|
|
|
## Hosted Kubernetes Clusters by ungleich
|
|
|
|
While you might be aware of other managed Kubernetes clusters, here at
|
|
ungleich we have quite a different offering. Let's start with the
|
|
obvious ones that you are unlikely to experience anywhere else:
|
|
|
|
### Environment friendly, for real
|
|
|
|
Other providers are dealing with Carbon Certificates, offsetting their
|
|
energy and claiming to be green. None of that makes sense to us, as
|
|
the emissions are produced nonetheless, fossil fuels are used or
|
|
nuclear energy contributes to running your Kubernetes cluster
|
|
elsewhere.
|
|
|
|
Here at ungleich we start from the beginning and continue to improve
|
|
our energy usage every day. Not only are we using a 100% renewable
|
|
energy (locally produced!), but we also research specific hardware to
|
|
optimise the energy usage required by Kubernetes clusters.
|
|
|
|
## Standard building blocks
|
|
|
|
Other providers offer you a complex system of building blocks: this
|
|
CPU, that memory, different storage classes, bandwidth consumption and
|
|
so on and so on. You easily lose the track of what you wanted to do in
|
|
the first place: to run your workload.
|
|
|
|
Here at ungleich we offer you exactly 1 building block to build your
|
|
kubernetes clusters: The ungleich k8s node.
|
|
|
|
## The ungleich k8s node
|
|
|
|
The ungleich k8s node is an AMD based, 12 thread computer with 64 GB
|
|
of main memory and two SSDs. The first SSDs is used for storing the
|
|
operating system and the second SSD is used to store your data.
|
|
|
|
The second SSD has a capacity of 1TB.
|
|
|
|
## The ungleich kubernetes cluster
|
|
|
|
Each Kubernetes cluster consists of 3 or more k8s nodes. While you can
|
|
grow it with adding individual nodes, we recommend to grow it in
|
|
numbers of 3. The reason for this is the underlying storage technology
|
|
[Ceph](https://ceph.com/) that saves each of your data blocks 3 times.
|
|
|
|
## Failure resistent
|
|
|
|
As your workload is running on any of the k8s nodes, maintenance work
|
|
and failovers can be executed without interrupting the services
|
|
(assuming your workload is deployed accordingly).
|
|
|
|
## Storage support included
|
|
|
|
Using a built-in dedicated Ceph cluster we provide you with persistent
|
|
storage that can be accessed using standard Kubernetes PVCs.
|
|
|
|
As every k8s node contains a dedicated SSD, by adding new nodes, your
|
|
storage cluster automatically grows as well.
|
|
|
|
As the Ceph cluster stores every object 3 times, the 1 TB SSD of
|
|
each node contributes about 300GiB usable storage.
|
|
|
|
|
|
## Pricing
|
|
|
|
Our pricing is very simple: Per K8S node we charge 300 CHF/month. This
|
|
includes rental of the hardware, network traffic, electricity and
|
|
maintenance of the kubernetes cluster.
|
|
|
|
This might not be the cheapest kubernetes you can build, but it is
|
|
probably the world's most energy aware and most modern Kubernetes
|
|
cluster out there.
|
|
|
|
---
|
|
offer1_title: 3 Node k8s cluster 900 CHF/month
|
|
---
|
|
offer1_text:
|
|
|
|
* [One time setup fee 150 CHF](https://datacenterlight.ch/product/k8s-setup-3)
|
|
* 3 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total 36 threads, 192 GB RAM, ca. 1 TB persistent storage
|
|
* Suitable for a development use
|
|
|
|
---
|
|
offer1_link: https://datacenterlight.ch/product/k8s-hosting-3
|
|
---
|
|
offer2_title: 6 Node k8s cluster 1800 CHF/month
|
|
---
|
|
offer2_text:
|
|
|
|
* [One time setup fee 300 CHF](https://datacenterlight.ch/product/k8s-setup-6/)
|
|
* 6 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total 72 threads, 384 GB RAM, ca. 2 TB persistent storage
|
|
* Suitable as small production cluster
|
|
|
|
---
|
|
offer2_link: https://datacenterlight.ch/product/k8s-hosting-6
|
|
---
|
|
offer3_title: 9 Node k8s cluster 2700 CHF/month
|
|
---
|
|
offer3_text:
|
|
|
|
* [One time setup fee 450 CHF](https://datacenterlight.ch/product/k8s-setup-9/)
|
|
* 9 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total: 108 threads, 576GB RAM, ca. 3 TB persistent storage
|
|
* Suitable as a production cluster
|
|
|
|
---
|
|
offer3_link: https://datacenterlight.ch/product/k8s-hosting-9
|
|
---
|
|
offer4_title: 12 Node k8s cluster 3600 CHF/month
|
|
---
|
|
offer4_text:
|
|
|
|
* [One time setup fee 600 CHF](https://datacenterlight.ch/product/k8s-setup-12/)
|
|
* 12 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total: 144 threads, 768 GB RAM, ca. 4 TB persistent storage
|
|
* Suitable as a production cluster
|
|
|
|
---
|
|
offer4_link: https://datacenterlight.ch/product/k8s-hosting-12
|
|
---
|
|
offer5_title: 15 Node k8s cluster 4500 CHF/month
|
|
---
|
|
offer5_text:
|
|
|
|
* [One time setup fee 750 CHF](https://datacenterlight.ch/product/k8s-setup-15/)
|
|
* 3 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total: 180 threads, 960 GB RAM, ca. 5 TB persistent storage
|
|
* Suitable as a large production cluster
|
|
|
|
---
|
|
offer5_link: https://datacenterlight.ch/product/k8s-hosting-15
|
|
---
|
|
offer6_title: 18 Node k8s cluster 5400 CHF/month
|
|
---
|
|
offer6_text:
|
|
|
|
* [One time setup fee 900 CHF](https://datacenterlight.ch/product/k8s-setup-18/)
|
|
* 3 nodes with each: 12 Threads, 64GB RAM, 1 TB Storage SSD
|
|
* Total: 216 threads, 1152 GB RAM, ca. 6 TB persistent storage
|
|
* Suitable as a large production cluster
|
|
|
|
---
|
|
offer6_link: https://datacenterlight.ch/product/k8s-hosting-18
|
|
|
|
---
|
|
content2_text:
|
|
|
|
## Tech Details
|
|
|
|
### The ungleich k8s node
|
|
|
|
* CPU: AMD Ryzen 5 5600G
|
|
* RAM: 64 GB DDR4-3200
|
|
* OS SSD: M2, 128 GB or more
|
|
* Data SSD: SATA, 1 TB, usually ca. 500 MB read/write (work load dependent)
|
|
* Connectivity: 1 Gbit/s
|
|
* OS: Alpine Linux
|
|
* Networking: IPv6 only
|
|
|
|
Sample CPU flags:
|
|
|
|
```
|
|
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca fsrm
|
|
```
|
|
|
|
### The ungleich k8s cluster
|
|
|
|
The Kubernetes control plane access is available via IPv6 and
|
|
IPv4. All pods are running in a dedicated /64 IPv6 network.
|
|
Service IPs are IPv6 only and are located in a second /64 IPv6 network.
|
|
|
|
The cluster does not support type=LoadBalancer, because it is not
|
|
necessary in IPv6 only environments. ClusterIPs are already by default
|
|
world reachable.
|
|
|
|
IPv4 reachablity can be configured in two ways:
|
|
|
|
* Via our IPv4-to-IPv6 proxy (http, https based)
|
|
* Via a dedicated IPv4 address
|
|
|
|
The current stack uses Calico as the CNI.
|