Deployment of Mirantis OpenStack and Fuel system configuration requirements Description

FuelWith Mirantis OpenStack, from a commercial point of view, it is still necessary to run to a verified branded device. However, from the point of view of learning and experimentation, there are other alternatives that can be used to complete learning with minimal hardware.


There are several things to note when choosing hardware:


Depending on the performance requirements, the server is required to have sufficient hardware configuration and network card configuration to circumvent resource contention and to match software and hardware requirements according to the actual situation; a fuel Master node can manage multiple environments at the same time. If you need to run a different hardware environment, you need to consider optimizing the relevant environment, separating the data section and the calculation part, selecting the appropriate external hardware device to host the OS release, and the new hardware drivers can be assembled into the Mirantis OpenStack release, but this action is a bit too complicated. It is also important to consider that some drivers may only be able to run in the corresponding Linux kernel version and not compatible with one or the other Kernle versions, so this part needs attention;



Fuel Master Node Hardware recommendations
To install fuel Master Node, you need to prepare the hosted hardware, including CPU, RAM, Disk, network, and the minimum configuration as follows:


4 Core CPU;4GB memory, 1G network, 128GB SSD disk;



Minimum configuration of the test environment limit:
Dual-core CPU;2GB memory, 1G network, 50GB disk;



Target Node Server Hardware recommendations
In the process of preparing the appropriate target nodes hardware, we can help to clarify which soft experience is running on this. The OpenStack community documentation has some suggestions for this section, so here's a look at the following address, which has the relevant guide:

Compute and Image System Requirements



The OpenStack Operations Guide contains something to do with planning this section, as you can see in the OpenStack Operations guide:


Install guidesoperations and administration Guidesuser guidesconfiguration gudies

Target NodesExample of hardware configuration
The following configuration is a general hardware configuration recommendation for target nodes, but this does not mean that it is the best choice, so it is for reference only:


12 servers, respectively, represents: 1 fuel Master node,3 Console controller nodes,3 storage nodes,5 Station compute nodes;high availibility configuration; Neutron Networking,vlan or GRE topology, used as the rear carrier ceph for Cinder,glance and Nova, and if required, more than 3 servers should be configured as MongoDB nodes for Ceilometer/mongodb;



Controller Node (3 units)
Controllers requires sufficient resources to run the Ceph monitors and MySQL that supports OpenStack core components (Nova,neuron,cinder and glance). The configuration recommendations for each controller are as follows: 6 CORES,2CPUS;64GB RAM (can support 1000+vms), minimum 24GB (2GB is enough for testing), more than 1TB of hard drives, as host OS disk and database, log and other messages ; 2 sheets of NICs, can be 1G or 10G;




Storage Nodes (3 units)
It is recommended to use a separate Ceph node to ensure resiliency and robustness. Hardware requires 0.5 cores per Ceph-osd and 1GB of memory per ceph-osd space. All Ceph storage and journal disks can be configured as JBOD mode. In the production environment, the CEPH replication factor configuration requires more than 3, the machine is configured as follows: Single socket 4CORES;24GB RAM;RAID1,500GB above the system disk; 2nics,1g or 10G; 18TB Ceph Storage (3tb*6), 1-2 SSD disks, 64GB or more, used as Ceph Journal;




Compute Nodes (5 units)
The virtual machine is hosted by the compute nodes, by default, the fuel adjusts the CPU's overcommitment ratio to 8:1, and if the load on the VMS is not clear, it is recommended that you set it to 1:1 to avoid resource contention conflicts. Each compute node configuration is recommended as follows:
Dual CPU,4CORES/SOCKET;64GB ram;raid1 array 500GB is used as system disk; 2nics,1g or 10G is OK;






In general, a dual CPU server is recommended for balancing I/O, CPU, Memory, disk and other messages such as the following micro, HP and Lenovo machines:
Supermicro SuperServer 5037mr-h8rtf; HP ProLiant dl380p Gen8 e5-2609v2 1P 8gb-r p420i/zm 460W PS server/s-buy; HP Smart Array P222 Controller; Lenovo thinkserver RD540 Rack Server;

more about VMware , OpenStack information to the virtual person.

Attention: vmanager_forum650) this.width=650, "title=" @ ' y5r1hgu9%7y2t3_@g}x) 2.png "src=" http://s3.51cto.com/wyfs02/M00/ 72/86/wkiol1xluigsh_g-aabkluis7hq792.jpg "alt=" wkiol1xluigsh_g-aabkluis7hq792.jpg "/>

This article from "Virtual person" blog, declined reprint!

Deployment of Mirantis OpenStack and Fuel system configuration requirements Description