Web Development And Hosting Requirements
Posted By Josie Moran20/07/21
We recently sat down with Carl Gamble, our Development Director, to answer the hosting requirement side of things. And we mean a solid run-through, so if you’re looking for a better understanding, this is it!
Take a look at what hosting is, how it can affect a company website, and importantly- what you need to consider when choosing a hosting provider.
Let’s start with an easy one. What is hosting?
The terminology used for platforming a website so that people can access it.
Are there different types with hosting, and what are the differences?
There’s no difference in terms of the websites themselves. Hosting is the foundation on which a website is built, and this means you can have different servers that provide this. (And receive different packages based on the provider.)
There are different connecting methods to said servers, HTTP and HTTPS (secure). These connecting methods are what you use when connecting from your device to the hosting server. It’s either encrypted or not. There’s less risk with a secured (encrypted) server.
It’s like when you connect to wifi, and with HTTP because it’s unsecured, the information on this can be intercepted. That includes login information, so it can be risky.
With secured HTTPS, people can’t see the data or passwords on these public wifi servers. For example, banking apps and services use secured sites to protect you in the first instance. At every stage, they are encrypting the site to stop that data from being shared.
With EasyTots, can you talk us through the process that you went through to get to the end point? Or at least, so far with hosting.
There’s nothing special about the general process or platform, really (it’s more the matter of choosing the best approach and method). It’s an E-commerce site, and the requirements depend on variables like the volume of traffic. EasyTots anticipated that they would get a lot of traffic when they debuted on Dragon’s Den.
The website is on WordPress and woo-commerce, which means it’s flexible, but it can be quite slow, to be blunt. It’s transaction heavy, and the speed can vary because many processes include checking that the buyer’s card details are correct. That there is balance in the accounts to make the purchase. And it is making sure that it has cleared, and all those sorts of things.
All of these are demand and resource-heavy, and it’s fine if you’ve got 500-page visitors let’s say. But if 500 people try to check out simultaneously, it will affect page speed and the overall functions of the site.
The goal was to have the web hosting set up like an efficient traffic system- perfectly capable and smooth running!
With EasyTots, they wanted to make sure that the website wasn’t affected badly. Such as having long wait times and having the servers battered by the traffic.
We had to evaluate the current server resources before launch. Evaluate the base level of traffic that the website got without that exposure, and then make some best guesses on the amount of traffic expected. Then run assimilations based on what we thought the server would peak at in different scenarios. And finally, decide on whether the current hosting platform would be suitable.
So, what happened?
We determined that the old one wouldn’t be suitable and would cap out at roughly 300 simultaneous users. We identified a new platform that would support and be compatible with what was needed. Like the plugins and specific modules that may be needed, and made sure that it worked!
It meant as little downtime as possible, which was crucial.
DNS propagation was a big part, as the data had to migrate properly from the original site to the new one. When you swap the IP address, the website automatically directs you to the new site. Memory storage can be a problem because if a user has visited the old site recently, it stores that data as being the page to go to. Then if they try to make a purchase, it can cause problems. Suddenly there are two stores (the old and new ones.) And will be putting conflicting data out as there are now two inventories and checkouts.
Before it becomes an issue, it should be stopped. We managed this by making the switch from the old site to the new one out of hours when the traffic was at its lowest (3 am), and then we terminated the purchasing features on the old site so that in the worst case, if anyone was stuck on the old site, they couldn’t use the basket and cause mass confusion.
What made this approach the best one for hosting?
All problems were looked at, and issues were predicted so we could manage more optimisations. And also put in a more scalable solution on the new hosting site.
I went an extra step and started the dialogue with the hosting site team. The traffic WAS increased massively, so it was a good thing we did! This is because we predicted that there was going to be a LOT of traffic at the launch.
They, in turn, made a ticket for the infrastructure team, so they were aware of the traffic levels to expect. Also, to measure the baseline traffic at the same time, for us to compare.
This would have also involved threat management. Giving you stronger protection against security threats but can also cause problems if unmonitored. Meaning “typical” visitors could experience page speed delays and just a general lag.
What sort of thing would need close monitoring, or reporting on with the hosting?
Traffic is just one metric, but it doesn’t mean all that much from an infrastructure perspective. It’s more important that you look at the natural utilisation of what you’re allowed to do. Or you look at what’s been allocated for the site.
In this instance, let’s look at the 0-100% analogy. 100% is the “ceiling” or the maximum level of what can be done. 0% is like there’s no website traffic, 50% is the halfway capacity, and you could double what you’re currently using.
The new hosting provider had a 200% natural burst. Which meant you had an expected utilisation of up to 100%, but in the event of heavy traffic, the website can deal with it. It’s important to note that this isn’t sustainable long-term and is only a short term solution!
What does that mean?
It doubles your resources for the short term and allows the site to cope. This is due to the CPU and RAM- which is essentially the memory of the site. So you can use extra to cope with it. It basically allows more space for these new users that are part of that burst.
The CPU utilisation during the launch period burst way past 200% in all actuality and capped out at around 1250%! And the memory went from being a pretty standard 2GB memory utilisation to 22GB- a ten-fold increase pretty much.
Like clockwork, it started to scale down as the traffic died off and was overall a pretty seamless scenario.
Had the conversation with the hosting provider not happened prior to the launch, it would have hit limits. Meaning people would start to, at best, slow down with page speed. But at a worst-case scenario, hit the timeout requests/this page isn’t responding, and potentially lose EasyTots the sales.
What do YOU think the future of hosting will be? Do you have a hunch on if major changes might come into play?
At the crux, a flexible approach.
Typically, old money hosting was a case of you buying a server based on its specifications. It’s going to have this much CPU, this much memory, it’s a flat fee and will cost you XYZ per month. And that’s it!
Businesses can be busy all day, though, especially if they’re serving worldwide customers. There can also be businesses that are busy 9-5 if they’re an SEO tool, let’s say. These won’t get the same traffic, and the old money approach doesn’t make the most sense. Flexible, pay as you go models are by far more appropriate. It’s better for your bank (unless you get that burst of traffic, and you may get extra charges!). Unhappy customers won’t be knocking at your door about long waits either!
The other bonus is also that if your server is sat idle during the night, let’s say you won’t be paying ridiculous fees. For a service that you really don’t need – which can often be ££££’s a month! If you only need that power half of the time, surely it’s better to only pay for what you need? The energy used is much less, so it’s more environmentally friendly too.
It’s scalable more than anything- it reacts at the relevant times and to what’s necessary.
But how does it work then?
Imagine you’ve got a little server bot that provides your website hosting. If that becomes overwhelmed, a new one will spin up to tackle that with you. If those two get overwhelmed, your team will have a third one join, and so on. However, each one of these will cost you money by the hour! But it will also scale back if you tell it to. (If it’s below a certain percentage of use by X minutes, it scales back).
You only pay for what you use, so it’s definitely beneficial. But there is a fear of writing that blank cheque to the hosting company. It’s not the most common thing. However, it definitely seems like the popularity of it will be big/growing in the future!
Some hosting IS terrible, some are good. Look for server location- you want a host that is as close to your client base as possible and makes it geo-located, which means shorter distances to speed things up naturally.
You then want to look at the difference in terminology between hostings. Like shared hosting, dedicated hosting, VPS, and understand what it is that you want.
That’s probably too much for THIS conversation, so I might have to go into that later! After that, it’s making sure you don’t go for the cheapest first- the company should have a good reputation too.