Servers power the software that will help your business thrive and your team to collaborate, but what do you need? We decipher the jargon and highlight the key features you need to look for.
Lifehacker’s Business Tech Guide 2015 is presented by our ongoing IT Pro coverage, offering practical advice for deploying tech in the workplace.
Sharing files, running important applications for your team, managing printers – these are all tasks that can waste your business a lot of time if you haven’t put the right tools in place.
Chances are, when your business started there were only a small number of people and sharing files was pretty easy. You’d either email them to each other or pop them on a USB stick and hand them across the office – what we used to call SneakerNet in the old days. If you got a little more tech savvy you might have even shared folders on each other’s PCs so you could access files.
But as the number of people in the business grows, the job of sharing information gets harder. At some point it makes sense to designate one computer as a hub from which everyone else accesses shared information. That’s what a server does – it acts as a central point that can be accessed by lots of people for accessing IT services.
Key Server Terminology Explained
Like any technical area, buying servers means that you need to get your head around a bunch of jargon.
Racks and Towers: If you’re planning to invest in several pieces of business technology, such as storage, network switches and storage then you’ll need to decide between either a tower or a rack-mounted server. Towers look like big desktop PCs. Rack-mounted servers are designed to be held by a couple of screws and a mounting system in a standard computer rack.
For a small business starting out, a tower is probably a slightly less expensive option as you can probably get away with non-rack-mounted network and storage equipment. But if your business is growing a rack-mounted server is a good option as it will take up less space and can be stored in a secure rack, away from prying eyes and hands.
Hot-Swappable Parts: One of the things the separates servers from regular computers is that many of the components are designed to be removed and replaced without turning the server off and interrupting your business. The main components that are made with this in mind are disk drives and power supplies. This capability is called “hot swappable”. It works as the server is made with multiple hard drives that are configured so that data is automatically written to multiple places. If one drive fails, your data is safe on another.
Similarly, having multiple power supplies means that if one power supply fails the load can be picked up by the second one. Being hot-swappable means you can switch a bad unit for a good one without impacting operations.
How Much Memory And Storage Does A Business Server Need?
Calculating how much memory and storage your server will need can be a daunting task. For the purpose of this exercise, we’re going to assume that the server will be carrying out several functions. These are:
- File server (for sharing documents and other files)
- Print server (for managing a couple of networked office printers)
- Finance system
- Key customer management system (this is the main system that your business uses for running the business)
- Email server (for corporate email)
For each of these, you’ll need to allocate three main things:
So, you might end up with a matrix like this:
|File server||4TB||2 cores||4GB|
|Print server||50GB||1 core||2GB|
|Finance system||1TB||2 cores||2GB|
|Customer management system||2TB||2 cores||4GB|
|Email server||2TB||2 cores||2GB|
|Totals (with some overhead)||10TB||10 cores||16GB|
(We’ll be covering storage in more detail in Chapter 2.)
With processors, it’s important to understand that it comes down to much more than just the clock speed — the number usually tacked onto the processor name that’s given in GHz.
The modern processor is really a number of CPUs joined together. So, a quad-core processor is really like four CPUs that have been melded together to act as one. So, your buying decision will be based on the number of cores and the clock speed together.
[related title=”Business Tech Guide 2015″ tag=”BusinessTechGuide2015″ items=”5″]
So, if you decide that you need 12 processor cores to run all the applications and services your business needs then what you’re looking for is a server that holds enough servers to cover the number of processor cores you need.
Although it’s possible to add processors to a server later, to accommodate future needs, you’re probably better off allowing for some overhead now. So, although out current need is for ten processor cores, we’d most likely, look for a server with 12 or 16 cores.
When you’re adding a new processor, you’ll need to ensure that the second, or subsequent processor is the same specification as the one you already have.
Or, you’ll need to remove the existing processor and add new, alike processors to the system.
With memory, upgrading later is easier but you still need to ensure that you have enough for your current needs and anything you anticipate in the immediate future.
The good news is that most vendors maintain stock of common upgradable components so you can buy equipment today with confidence that you’ll be able to upgrade in future as needed rather than investing in hardware before you need it.
In the old days, you’d probably end up buying separate servers for each of those functions, or combine a couple of the less demanding applications on one box in order to save a few dollars.
Today, you can buy a single server to do all those tasks. They key to this is virtualisation.
Operating System And Virtualisation Options
Virtualisation isn’t a new technology but it is one that has matured and reached commercial acceptance over the last decade or so. Here’s how it works.
In the past, the accepted best practice was that each server was allocated a specific task. For example, you would run email from its own server. You wouldn’t double up functions so that the one server would run your email and finance system.
There were two reasons for this. Firstly, if the server failed, you’d lose access to two critical functions. Secondly, the amount of processing power available wasn’t sufficient for both functions to run effectively on the one server.
Both of these problems have been overcome through better, more reliable hardware, and by Moore’s Law. Processing power is no longer a bottleneck for most applications.
Virtualisation takes one powerful computer and, through the use of software called a hypervisor, allows you to create virtualised computers that are allocated their own slice of the available memory, processor and storage capacity.
Virtualisation and Hypervisors
Think of it like this. Imagine your server is made of a series of layers.
At the bottom is the hardware – the processors, memory and storage. Sitting above those is the hypervisor. It acts as a broker that distributes the hardware resources to the virtual servers that sit in the next layer up. Each of those servers, although running on the same server, is isolated from all the others. So, if one virtual server needs to be restarted or has a problem, it won’t affect the others.
The hypervisor is a barebones operating system that carries out a single task – the allocation of hardware to virtual servers. This can be dynamic.
Let’s say you decided to set up a server with three virtual servers running on it. Each separate virtual server required two CPU cores. That means you need access to six CPU cores. However, the reality is that none of the applications will be using 100% of their allocated server cores at the same time. The hypervisor can allocate the resources, to a maximum of two cores per server, dynamically. If this is done efficiently, you might be able to get away with a physical server that has just four CPU cores.
It also means you can run multiple operating systems on the one physical server. So, if you have applications that require Windows Server, you can run those alongside applications that require a Linux server.
As well as being a really efficient way to get the most out of your server hardware, virtualisation offers you lots of flexibility without having to invest in multiple physical servers.
Choosing a hypervisor for your server can be a confusing task. The two main players in the hypervisor market are VMware’s vSphere and Microsoft’s Hyper-V. There are also some other options such as Citrix XenServer and Oracle VM.
The two main players, vSphere and Hyper-V, have different strengths and weaknesses.
If you are running Microsoft server applications exclusively and have an existing licensing agreement with Microsoft then the scales will probably tip in Hyper-V’s favour. As well as being optimised for Windows Server, Hyper-V licensing costs may be included in your current server licensing arrangements, or added at a reasonable cost.
If you’re running a more diverse set of applications and need to support different server operating systems such as a Linux-based OS, then you may find vSphere is a better fit. However, it’s possible this will cost more.
Another thing to bear in mind is that if you decide to run all your applications from a single physical server you are creating a single point of failure in your systems. One of the great features of modern hypervisors is that you can replicate entire virtual servers across different physical hosts easily. Although the hardware cost goes up, you can increase the confidence that a single hardware failure won’t stop the business.
Within businesses, the most popular server operating system is Windows Server. Now over a decade old and in its sixth iteration (if we ignore the old days of Windows NT Server), Windows Server is increasingly popular and is the most use operating system in businesses.
Access to skilled professionals and technicians when it comes to deploying Windows-based servers and applications is plentiful. All of the major hardware manufacturers and suppliers have established relationships with channel partners to assist with not only setting up a server and your preferred hypervisor but with deploying applications and getting your business productive as quickly as possible.
If you prefer to take the open source road, then you’re most likely going to be looking at Red Hat, SUSE or Ubuntu for the operating systems you run.
Checklist: Buying A Server
- Make a list of what will need to run on the server
- Determine the operating system, processor, memory and storage needs for each application
Choose which hypervisor will best suit your environment
- Design a solution that adequately mitigates the risk of bringing your business to a halt of a single component or device fails
- Talk to a hardware suppliers to determine the best match of hardware and software for your budget
You should now have a pretty solid overview of servers and how best to deploy them. Check in at the same time and place on Wednesday for an overview of networking.