update Nov 7th 2015
In this NEW series of posts I’ll walk you through the steps, issues and frustrations of designing, building and configuring a low-power performance virtualisation cluster based on three nodes. It’s a write-as-I-go story which means I’ll cleanup the text when I can find the time 🙂
The cluster is designed around the following principles:
- Low power consumption (equal or less than my current single host)
- Be as small as possible
- Silent running
- Enable high availability for VM’s
Based completely on Microsoft technology(been there, done that) VMWare vSphere- Must be easily scalable and feel “enterprisy” 🙂
- Low cost solution!
The first couple of posts will be easy (for me) as I’m re-using the existing three nodes from the earlier Binsfeldius Project builds. For your convenience I will also include the stuff that I am using but haven’t changed since the original project so you don’t have to jump around.
Low power consumption
To achieve this I’m using one of the latest Intel mainboards, the DQ67EP (link) with an Core i3-2100T (link). This CPU is the low power version and has a TDP of 45W. The Intel board also doesn’t draw as much power as the competition. The idea was to use a PicoPSU (150W) however my current selected case comes with a 200W PSU builtin. I’ll see what this does first, can always switch as this case has the drill holes for an external powerbrick for the PicoPSU.
I’m not putting a drive in the nodes but I’m going to boot from my NAS. I have a new QNAP TVS871 which will provide the necessary iSCSI LUN’s. The NAS is always on anyway so this saves on power consumption for each node.
Additionally the challenge is to have the three node cluster use about the same or less power than my current single host which runs around 90 Watts under load.
Be as small as possible
No need anymore for lumpy hardware. The Intel board is a mini-ITX form factor and by using small PSU and no drives, the case can be really small too. As a cluster needs several physical networks and the DQ67EP has only one physical nic, I’ve added an Intel 1000/PT Pro Dual Nic card. The bulk package includes a half-height bracket which enables using a small case. Took me a while to decide on the case, first looking at the M300-LCD (link). Very pretty and very small however with quite a few drawbacks. It has the lowest height which would mean using a risercard and flipping the nic on it’s side. Pictures of the case, the mainboard, the nic and the cpu with cooler attached could not tell me if it would fit (appearently no-one could tell me as no-one has tried this combination of components before). It had a looooooong leadtime on delivery so I needed another case. After a lot of mixing and matching I’ve chosen the Antec ISK 300-150 (link).
“As small as possible” also means no additional hardware other than the clusternodes.
Silent running
Even though I have a separate room for my hardware, it needs to be as quiet as possible. The Core i3 doesn’t generate much heat (currently it runs at 44 degrees C) and the TriCool fan in the Antec case at low setting is very quiet.
Enable high availability for VM’s
A three-node-cluster with the ability to have zero downtime for VM’s which live in the cluster
VMWare
The virtual layer is based on the spiffy new VMWare vSphere 6.0 platform meaning the nodes will run on ESXi 6.0 and management will be done through vCenter Server 6.0 appliance. I’ll give the new webclient a whirl as well. Keep in mind that VM created on version 9 and later cannot be managed by the vSphere fat client anymore.
Must be easily scalable and feel “enterprisy” 🙂
It must be easy to add more nodes without going through complex design/configuration stuff.
Low cost solution!
Speaks for itself