Answer the question
In order to leave comments, you need to log in
Proxmox, does it make sense?
Hello.
There is a server, mother ASUS P8B-X, Xeon E31240, RAM 24GB, 2x1TB WD Black, with Proxmox 5.1 installed on top of Debian 9, on which Raid1 is created. 2 virtual machines are running: Windows 2008 (allocated: 4 cores and 12 GB of RAM) and Asterisk (2 cores and 2 GB of RAM). This configuration has been running on this server for more than a year. Was win2008 on iron earlier.
In windows2008, users (5-10 people) via rdp work in 1s 8.2 and 8.3. (file versions). Working in 8.2 is more or less comfortable (the base is about 7 GB), but in 8.3 (typical empty configuration) the launch is about 1 minute and brakes at work.
Installed server 1c and sql, improvements are almost not noticeable.
So far, with my little experience, I see several solutions:
1) Remove the currently available 2x1TB WD Black (because they have been working continuously since the birth of the server, it’s been about 7 years already), install others, at least the same but new ones, merge them into Raid 1 and install Proxmox on them. Next, install 2 more SSDs, combine them into Raid 1 and install VM images on them.
2) The option is the same as in the first case, only without HDD, Proxmox and VM should be installed directly on the SSD, but as far as I understand, in this case, the ssd operating time will be significantly reduced.
3) Is a hypervisor needed at all in this scenario? It can do everything the old fashioned way, install 2 SSDs in Raid1 on hardware, they have Windows 2008 and related software on them. (Transfer Asterisk to another computer, this is not a problem)
4) 5) 6) .... your option ....
Are the options for using ZFS Raid generally relevant here? Proxmox already has this thing out of the box, will it be good if you use it instead of the usual Raid?
The task is to speed up the work 1s.
At the moment, the VM load is no more than 20-30% in all respects, or even less.
What do you advise?
Please, if it is not difficult, to support the answers with links.
Thank you.
Answer the question
In order to leave comments, you need to log in
The option is the same as in the first case, only without HDD, Proxmox and VM to install immediately on SSD, but as far as I understand, in this case, the ssd operating time will be significantly reduced.
Option 3 seems fine to me. The only thing is that the point in raid1 of two SSDs is zero, since the read speed will no longer be any more than ssd gives, and if they do, then at the same time, you will only lose in space.
On the wheelbarrow where you will deploy the asterisk separately, add hdd raid1 from large disks and add backups of file 1s (well, and something else) on them according to the schedule.
Open the task manager and look at what your 1s rests on, judging by the description, all your actions will lead to nothing
Zfs no need to use it is very slow
What I would do:
1 check the smart of both WD Black - if at least one of the hard drives dies, then the entire array slows down
2 would run the read / write tests - does it correspond to the expectation?
Further, according to the situation.
General tips:
1 For comfortable operation of virtual machines, their virtual hard disks should be on fast disks, the debian + proxmox OS itself can be on a regular, even very slow hard disk - I still have ide segate 160gb working at home and it has an OS - and vmki live on raid5 of 6 WD Black.
2 The easiest option for organizing space in proxmox is regular LVM, yes ZFS works faster but needs a lot of RAM and is more difficult to configure (out of the box LVM works fine and predictably)
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question