Your own server can be cost-effective

Often we’re called into a small to medium business to look at a server issue, because it’s finally become too hard for the guy who ‘knows a bit about computers’ to fix. Really, you’d be amazed how often this happens. Often the server is ten years old and struggling to cope with the modern demands, particularly with the size of data these days.

Usually it’s more cost-effective to replace the server with something small, such as a completely custom unit made from common parts designed for their purpose. In fact, we have two of those custom servers here – they cost less than $1500, have huge RAM and processing power, and large redundant storage. The trick here is that of course, we also maintain them ourselves. When the network card in one of them developed a fault, we grabbed a $20 card from stock and had it running in 20 minutes. That is the risk we choose to undertake there, IF there is a fault, WE have to fix it within a timeframe that suits our needs.

Of course your risk is different with that ten year old server – WHEN there is a fault, YOU need to pay $$$ for it to be fixed, and it will likely take a while, plus response times, you get the picture.

All of this discussion so far has centered on a hardware fault, but what happens when the server just messes up? Well that’s an interesting question because it’s rarely a software fault that causes a problem down the track. It is certainly true that getting a server properly configured can be a time-consuming, annoying, painful, difficult task, but once it’s running properly it’s very rare for the software to fail on its own. It does happen that certain Windows updates might break something, or installing something might cause an issue, or a virus could cause problems, but all of those require an external influence – someone deliberately installed something, or updated something.

We provide specialist personnel to certain clients, and one particular government client has an enormous fleet of servers – hundreds – and they all need to do their jobs. In the years that we have been on site at that client, the number of servers that went down without external influence is less than five. To be honest I can’t think of any. And why is that? Because 99% of those servers are virtualised.

Virtualisation is a technology that allows a server system to run a number of client systems. For example, a host might have 32 processors, 128GB of RAM, and mountains of storage. It could run 32 client servers each with 1 dedicated processor and 4GB of ram with no trouble. More advanced configurations then go into dynamically allocating processors as the client servers get busier, or more RAM where it’s needed, but the ins-and-outs of virtualisation aren’t the point here today.

Adding “the Cloud” to the mix (it is the ‘in’ thing at the moment), you could get what is called a Virtual Private Server. This is where a company sets up a datacenter on the internet with huge virtualisation servers, and then sell people a ‘slice’ of their infrastructure.  For a fixed fee, you are given a set allowance of processor cores, RAM, and storage which you can use however you like (within reason). These generally carry a few huge benefits;

1. Their datacenter is usually located in a very secure location, and physical theft of your data is almost impossible (as is fire or flood)
2. They will provision it with your choice of operating system in minutes
3. The license fee for the operating system is usually included in the cost
4. The new server is connected almost directly to the internet, getting much faster speeds than you probably would usually
5. You don’t own any hardware to maintain
6. Your service provider (eg Simpler Technology) can support you by directly remoting to your server on the cloud
7. You can access it anywhere you have internet access, or secure it to only allow your office to communicate
8. If you need to do more, you can simply add resources to it in minutes – no day long outage to upgrade the RAM and processor
9. If there is a hardware failure, your server will generally not notice and remain available (this magic is called clustering, failover, or high-availability depending on the specifics)

Of course there is a downsides, and we like to give you ALL the information around here;
1. You can’t get to it without an internet connection. If your server is on site and your internet gets cut off, you can still get that local server and its services

So for all of that, the message is this; a cloud VPS provides many benefits especially in the maintenance area while keeping the cost reasonable.

If you are going to consider a VPS, we recommend the team at VPSBlocks – they’re based in Melbourne so the performance for Australian users is fantastic.