Servers and Drives FAQ

What are drives and servers and how do I use them?

Drives are just like physical hard disks or CDs. They store your operating system, applications and data, and are persistent across server reboots.

Servers are virtual machine instances, and may have one or more drives attached to them. They are either running or powered off. When a server is running, the cloud provides VNC access. When a server is powered off, you can configure the CPU, memory, drives and IP addresses which it uses.

Some of our competitors have less-flexible products in which every server has exactly one drive. Our approach is more general, allowing you to attach several drives to a single server, or building several alternative configurations which boot from the same drive. (However, only one running server can access a given drive at any one time.)

How are the sizes of drives and servers measured?

Like a physical hard drive, drives are measured in gigabytes (GB). Servers have two adjustable sizes: the amount of memory (in MB) and the amount of CPU bandwidth (in core-MHz).

How large a server and drive do I need?

This will depend entirely on what you plan to use your server for: everybody’s requirements vary. However, a good starting point is to configure with the same sizes as you would use if you were purchasing a physical server for your application. You can then tune up or down the specification as you require. Bear in mind that most physical servers are well over-specified for their actual requirements and you can probably get away with a lot less than you were expecting :)

How do I access and control my server?

The cloud provides basic VNC access to servers, which works right down to the BIOS level, and allows you to install, configure and recover your operating system, even if nothing is running inside your server. In normal use, you should access your server by a native method (e.g. SSH, Windows Remote Desktop etc) which will provide superior performance.

What VNC software do you recommend?

Our VNC service requires a reasonably recent client. TightVNC works well on Windows, and clients based on gtk-vnc such as Vinagre work well on Linux. The RealVNC Enterprise Edition Viewer is good on Linux, Windows and MacOS if you select “Always use best available colour quality”. The Linux and MacOS versions of this are available as a free download under ‘Enterprise Edition Viewer’ on the RealVNC web site.

Neither the Apple Remote Desktop VNC client, Chicken of the VNC, nor JollysFastVNC work correctly on MacOS as they don’t support some of the newer protocol features which we use.

What is the maximum size of a single server or drive?

The largest single server you can create is 20000 core-MHz CPU, 8192 MB RAM, with one or more 500GB disk drives.

What is the minimum size of a single server or drive?

The smallest single server you can create is 1000 core-MHz CPU, 512 MB RAM, with a 5 GB disk drive. This is typically sufficient for a Linux applicance such as a DNS server or a small web server.

How many servers and drives can I have?

There is no limit to the number of servers and drives that you can have in your account.

What physical infrastructure is Serverlove based on?

We use exclusively Dell hardware, with R710 and R410 servers used in the cluster. The hosts have the following specifications:

  • Dual quad-core Xeon E5520 or E5620 CPUs (running at 2.26GHz and 2.4GHz respectively),
  • Either 32GB or 48GB of RAM.
  • Either 4x or 6x Enterprise-Class 1TB disks with battery-backed cache.

They run a customised cloud hosting platform, which is built on open source technologies including Linux, qemu-kvm, LVM and iSCSI. The disks are arranged into RAID 1 pairs, and virtual disks are allocated out of this pool of RAID 1 arrays.

How are my servers and drives distributed between physical hosts?

When you create a drive, it is allocated on a randomly chosen host with sufficient free space.

When you start a server, our system prefers to place it on the same physical host as its drive(s). If this is possible, the server can directly access the underlying physical disk. Otherwise, if the host containing the drive is already too busy, the server will be started on a nearby host and it will access its storage by peer-to-peer iSCSI over gigabit ethernet.

One of my drives is a backup of another. Can I specify that the two drives should be allocated on different physical hosts for better resilience?

Yes, although this is a new feature and only possible through the API at present. When you create the second drive, pass in ‘avoid DRIVE1′ where DRIVE1 is the UUID of the first drive.

What do the Shutdown, Hard Power Off and Hard Reset buttons do?

The shutdown button sends an ACPI power button signal to the server operating system, exactly as the ‘soft power-off’ button does when pressed on a physical machine. Normally operating systems are configured to shutdown or hibernate when this is done.

The hard reset button sends a non-maskable interrupt to the server CPU in exactly the same way the reset button does on a physical machine. You should only use this if your server has crashed; otherwise you should reboot from inside your operating system.

The hard power off button behaves exactly as if you have switched off a physical machine by turning off the power. You should only use this if your server has crashed; otherwise you should shutdown using the button or from inside your operating system.

 
21 queries. 0.164 seconds.