Hardware Requirements for Hosting
On another of our guides, we point the prerequisites of a reputable hosting company, and with that knowledge applied you can be pretty comfortable knowing you can choose a hosting company that meets your needs.
One other major factor is the hardware requirements that your site needs. Even the most experienced driver will struggle to get great performance out of a bicycle… poor analogies aside you will want hardware that can serve your needs.
Before going into any detail, it’s worth mentioning that any vanilla website that doesn’t receive large amounts of traffic will typically be absolutely fine on any shared, VPS or dedicated host. I am generally describing requirements for general use of a server for hosting websites, though there’s countless reasons for acquiring server resources, well beyond my imagination.
That aside, the following covers components of hardware and service that will determine the overall serving capacity that your hosting plan can provide to you, in respect to hosting websites and general capacity to just work.
Our other guide outlines of the difference between shared / VPS / dedicated, bear this in mind when talking about the availability of said resources as I’ll tend to avoid outlining those differences here.
Hard Disk Space / Solid State Disk Space
What It Is:
These are the storage devices that are used to keep your data. Data is persistent, meaning that data is not lost after a reboot of the server. Disk space is typcially measured in Gigabytes. As a frame of reference, you could store about 80 thousand 50 Kilobyte web pages per Gigabyte, and more often than not you’ll be offered at least a Gigabyte of storage on even the most economical of hosting plans.
- A hard disk is a rapidly spinning platter that been the goto method of storage since the 1960’s. Hard and on a dollar per Gigabytes basis, is cheaper than solid state storage. Hard disk space tends to be plentiful, but is slower than solid state because the read/write heads have to wait until the disk rotates to the part of the platter to read/write from.
- - Solid state disks have no moving parts and can access data from any part of the disk in roughly the same time without the latency a hard drive has.
Generally, solid state disks are better at random reads and writes on a disk, and have better latency, but cost more per Gigabyte of storage.
What to Look For:
The amount of space you’re allocated in your hosting plan invariably limits to you how much data you can store. Often with shared hosting you will be offered ‘unlimited’ disk space, but in reality you’ll be capped at a particular limit. As long as you have enough disk space for your site, with some wriggle room, you should have more than enough to host simple websites.
RAID (redundant array of independent disks) is also worth considering. A RAID setup involves duplicating data across disks to increase redundancy, mainly to:
- Have a backup of data in the event a disk drive stops working
- Spread the workload over multiple disks so as to avoid one disk being overworked and the rest under utilised
No RAID or RAID 0 (all data on one disk) means that if the disk breaks, your data is gone. In these situations, you would want to use your backup… because of course you have read our other guides and know that having your own backups is the right thing to do :)
On non-dedicated hosting plans, you’ll be sharing disk(s) with others, so the setup a provider has will greatly determine how responsive your site in general and disk allocation is.
Many processes you ask of a computer that include using disk space will inevitably be disk-bound, meaning that the slowest component of doing a workload is limited by the speed of the hard disk.
In summary, a good provider will allocate you enough disk space and have a disk setup that will not result in catastrophic loss of data or slow input/output due to overselling the resource. Memory
What It Is:
The disks store your permanent data, the memory is the region of a server that retains volatile data, meaning data that would be lost on a server reboot. Memory is a superfast storage area that stores data the CPU is going to work with. Access to data in memory is typically around 100,000 times faster than retrieving from a disk.
A modern system will have memory measured in Gigabytes, though you can acquire VPS plans that can have as little as 32MB or 64MB of memory available to you. (In case you’re wondering, you can still do quite a lot with these hosting plans).
Memory is required by processes to run smoothly. When a system runs out of memory, it will start using the disk as ‘swap’ which is many order magnitudes slower, and will become immediately noticeable to users of the system. The operating system may also elect to start killing processes to free up memory.
What to Look For:
Shared hosting plans typically don’t mention any memory designated to you, but there is often an upper limit on how much memory any or all of your processes may consume. Sometimes the limit can be prohibitive.
On VPS and dedicated plans, for the purposes of web hosting you’ll want to ensure there’s enough memory for all your applications to run (a web server at least, perhaps also an email server, perhaps a database). You will also need additional memory that is consumed by processes, for example the web server dealing with a request to a web page.
If you’re tech savvy, you can acquire a small VPS or dedicated plan and optimise the settings of your software to maximise your usage of memory, and can run a busy site with a hosting plan with memory measured in Megabytes, not Gigabytes.
Generally, having more memory means better performance, especially so if it means accessing data cached in memory rather than being cached on disk.
One final caveat to look out for, particular for VPS and dedicated servers is whether the RAM has ECC (error correcting codes). ECC is a technical solution to ensure data integrity, at slightly higher cost and a small performance penalty of a few percent, it gives additional assurance that the data being moved about in your system is sound.
What It Is:
The CPU is the command centre of the server, and its ability to carry out its instructions at a given speed affects how quickly these instructions are exectuted.
CPUs are responsible for taking in data and manipulating it for a desired course of action. Their clock rate is typically measured in Gigahertz and indicates how quickly it can execute instructions.
Modern desktop computers will have at least 4 processors, and high-end dedicated servers can have hundreds of processors. The processors may be threaded, which basically means that one core may have separate streams of instructions that it will process, which increases parellelism between tasks but not the overall speed in which they’re done.
What to Look For:
On shared hosting, it’s difficult to tell how much performance you will get from the CPU, as it’s shared by all the other shared hosting customers. More often than not, the provider will tell you how many cores are on the server you’re hosted on.
For VPS hosting, again, it’s hard to tell how performant the CPU will be for you when it is shared amongst many customers. The
In both cases, there may be a hard rate limit or burstable limit which is applied to each customer.
For dedicated hosting, the cheapest servers will give you a small single or dual core CPU, typically an Atom CPU that can work just fine with small workloads.
Finally, the CPU is typically the most power intensive component of a machine though luckily for you this more often than not is a cost factored into hosting and therefore the power efficiency of a particular model of CPU is not your problem.
IP Addresses – Ipv4 and Ipv6
What It Is:
IP addresses are numerical labels assigned to computers on a network that distinguishes them from one another and provides a system to map connections between one another.
Almost all hosting packages will be accessible via the Internet with an Ipv4 address, which you’ll want if you intend to be serving web pages.
Some providers may offer Ipv6 connectivity too. Use your favourite search engine to see the difference between the two protocols, at the most basic level you’ll be fine knowing that Ipv4 is what the rest of the world can use to connect to your host.
What to Look For:
More advanced usage of a server may require more Ips, though if this fits your use case then most of this document is already second nature to you. Due to the scarcity of Ipv4 addresses, leasing of extra Ips can be quite expensive, ranging from one off fees of a dollar per IP to $2 per month per IP.
Generally you don’t need to concern yourself with Ips other than knowing that the general Internet population is able to access it.
Typically you won’t know what IP address(es) you’ll be assigned before acquiring a hosting package, but with a little searching you can get a rough idea. You may be interesting in doing lookups to see if there’s been any nefarious activity on their network, for instance being listed in a DNSBL.
If you’re using a shared IP, you may also want to know how many other customer’s sites are being hosted on an IP/server. Some budget hosts will cram thousands or tends of thousands of sites onto the one IP, which presents a problem for crawlers and robots who are morally obligated to rate limit requests to a server. This can ultimately mean that your site is slow to get listed on search engines properly, because the search engine crawler has a finite amount of time to fetch pages from your server, but has many other sites to look at too.
See also: Factors to Consider Before Buying a Hosting Service
Much of this topic is covered in the above article, in the general case you’ll want to ensure the hosting provider provides you yourself with enough bandwidth (both in terms of the bitrate of data you can transfer at any moment , and the running total of total data transferred over a given period).
Regardless of any bandwidth cap applied to you, the provider should also have enough of its own bandwidth to satisfy your requirement at any given time. Budget providers can often be guilty of overselling their bandwidth, which can result in periods of the day where the amount and speed of data you can transfer over their network is throttled due to reaching capacity.
You may have heard of a contention ratio in reference to your ISPs residential Internet connection. A contention ratio of 20:1 means a given network pipe of say, 1 Gigabit per second, is shared by 20 customers. During nighttime you may see obvious speed improvements because the other 19 customers are not using their connection, alternatively you may see slowness at peak times like the early evening. The same scenario applied to web hosting, there are busier times of the day where you may notice a degradation in the connectivity of your website (the load on the server may also be at play).
What It Is:
A graphics processing unit is dedicated to processing graphics rendering but can also be used for GPGPU tasks that perform better with parallelism, one example being crytographic operations used in SSL/TLS. In most cases, you would not require nor need a GPU.
What to Look For:
On some VPS and dedicated plans, there is an integrated or dedicated GPU that is available to you.
A GPU dedicated to graphics rendering or used for parallelism ensures that the CPU can allocate its clock cycles to doing jobs more suited to it.
More details on what to look for in GPU hosting will be forthcoming, link will appear here.
I have described the components of hardware that are critical for your hosting needs. The hosting industry and its thousands of providers will offer you all manner of permutations of the above resources, at varying prices and at varying levels of quality.
Ultimately, you should gauge what your minimum requirement for hardware is while having a contingency for additional resources that may be needed at short notice. Knowing your minimum requirements of hardware can at least reduce the mulitude of hosting choices to a more manageable selection.