Table of contents of the article:
One of the most absurd questions that we usually read on the net is: "Does anyone recommend me a performing and cheap WordPress hosting?", following the festival of responses from Hosting companies, each brought up according to direct or indirect personal experiences.
The problem is that we are talking about absolutely different cases and situations that are impossible to evaluate without a proper analysis of the requirements and needs.
What would you answer to those who asked you:
was Michael Jordan or Diego Armando Maradona stronger?
From a hosting point of view, the question sounds more or less the same. He has no answer, but he definitely deserves one.
The reason is easy to say, computer science is a science almost exact and therefore it is necessary to measure and then decide. This step is vital as a bad valuation causes the customer to lose money.
It makes them lose essentially for two reasons, the first because the site would risk being slow or even crash under the weight of a lot of load, the other reason is that it could be pushed to buy a highly oversized hosting plan for real needs and therefore maybe throwing tens of thousands of euros a year when it would have been enough to simply evaluate the situation calmly and competently and then direct it towards the best hardware and software solution in order to meet the technical needs without going to oversize the hardware and related costs that in some cases with some of our customers could have even reached 12 thousand euros per month.
So here's what good hosting is a good system engineer should ask before referring a customer or prospect to a hosting solution, be it a shared hosting, a VPS, Cloud or a Dedicated Server :
What is the site?
Which CMS do you use?
What hardware technology do you currently use?
What are the traffic peaks?
Do users just read or must / can log in?
What kind of business continuity do you need?
What budget do you have?
Let's try to understand the importance of these 8 questions and why every good system engineer, especially if a Linux system engineer should ask you.
1 - What is the site?
By simply knowing the name of the site, an expert systems engineer is able to obtain important data to size the necessary hardware and choose the best software.
Through online tools such as SimiliarWeb, Weppalizer and BuildWith in fact we can know in a matter of seconds the number of monthly visits that it makes, and the server-side technologies that currently, as well as through SEOZoom understand the growth or decrease trend and therefore weigh the choice not only in the immediate present but also considering the near future.
Put simply: it allows us to have a rough view of our future client.
2 - Which CMS do you use?
Answering this question immediately makes us aware of the software-side configuration that we could go to use, and any problems to be solved. Do you use Joomla? Do you use WordPress? Do you use Magento, WooCommerce or Prestashop? For example, we could evaluate the margins for improvement and optimization, aware that normally WooCommerce allows decidedly significant improvements only by intervening on the software configuration while Prestashop usually already works at its own speed and therefore the margins for improvement that can be obtained from a good tuning are minimum and you must necessarily evaluate the hardware and size it appropriately if the current performance does not satisfy us.
But above all do you use a CMS or a custom site?
Becoming aware of the use of a customized system perhaps written ad-hoc that does not use any CMS immediately warns us of an unknown terrain, in which we do not know what the weak points may be, the possibility of using a static cache and dynamic without upsetting the code, possible problems of inefficiency and excessive slowness on the database, slow queries, use of indexes, related optimizations and so on.
3 - What hardware technology do you currently use?
Answering this question means knowing the hardware architecture that currently hosts the site. If we talk about shared hosting, it is enough for us to know that it is in shared hosting with all the problems of overselling and inadequate resources that normally all large hosting providers tend to "pass on" to their customers.
If they answer us instead, VPS, Cloud or Dedicated Server, at that point we must necessarily know the number of cores, how much RAM has available, the speed of the network card, the speed of the disks and possibly also the supplier.
This will help us to immediately evaluate any problems deriving from the current supplier who maybe even "giving away" CPU and RAM there is an extremely slow disk I / O, or vice versa even if with SSD or nVME performing disks the system is underpowered in terms of of CPU and RAM to properly run the WordPress hosting you need.
Furthermore, it will allow us to size the new hardware and evaluate whether to increase the number of cores, RAM, or even reduce them if oversized.
In the usual habit of measuring to decide, knowing the current hardware allows us to make the right choices with extreme serenity.
4 - What are the traffic peaks?
Although looking at the first point of this list through the aforementioned analysis tools we can know quite precisely the monthly traffic, what we cannot know are the traffic peaks that a site can receive.
For example, if a newspaper publishes the news of an earthquake, rather than that of a bomb attack against a political office, how many visitors can it do?
We had the case of a press report shared on the Facebook and Twitter wall of the current Prime Minister Salvini, which led the site in question to score well 5 million visits in a single day, of which the first two million within two hours. Obviously such a case can be considered an exceptional case, unlikely but not impossible.
An answer from memory
5 - Do users just read or must / can log in?
This question is of vital importance especially when dealing with interactive systems such as forums, ecommerce or restricted areas. While it is true that pitiful native performances can be cleverly masked through the use of static caches such as Varnish SuperCache or similar solutions, it is also true that a logged in user as such cannot have this benefit as the cache is bypassed for obvious reasons.
In this case it is good to know the nature of the application and predict what the number of logged-in and non-logged-in users may be and measure the load in order to subsequently size the hardware to be chosen.
6 - What kind of business continuity do you need?
This question may seem obvious as everyone would answer 100% of course, but in fact we know that a real 100% actually means having such a geographical redundancy that it imposes double DNS, double geographically redundant instance perhaps on Amazon AWS which is not really cheap. The difference between 99,9% and 100% while being only 0,1% in terms of hardware and software infrastructure uptime means spending about 10 times as much.
To be clear, are you spending 300 euros a month to have 99,9% uptime? Be prepared to spend AT LEAST 3000 if you want to have 100% uptime.
Does this make sense ? It depends.
With the help of Uptime calculator in fact, we can determine how much 99,9% uptime is monthly, or better still a how much is downtime.
As we can see from the site, 99,9% corresponds to 43 minutes of monthly downtime. It must be said that usually with a 99.9% guarantee, double the time for disaster recovery procedures is always calculated if something really goes wrong.
Therefore, let's imagine that your business stands still for 2 hours twice a year, for a total of 2 hours a year. Can these 4 hours of down time justify a hypothetical expenditure of 4 thousand euros more per year to reach a hypothetical 30%, compared to an expenditure of 100 euros per year to have 3600%?
It depends on the business you do and how much your site makes either in terms of image or in terms of revenue or earnings. The reality is also quite different, in the sense that downs are still rare (a couple a year) lasting between 5 - 15 minutes and in fact even in redundant solutions with 100% uptime they can have problems of nature. software that still lead to problems. This also happened to big players such as Ebay, Amazon, Spotify, Netflix who stayed down for over 4 hours ( The original Amazon S3 failure report here ) and therefore it is undisputed to say that a down of a few minutes a year can be there.
This aspect must therefore be evaluated very well, actually taking into account whether you prefer to risk having 1 hour of down time but a considerable saving in infrastructure costs (an indicative average saving of 500% up to over 1000%) or if it is of vital importance. reach 100% and therefore spare no expense.
The choice in these cases often depends on business logic that is not too intelligent but still legitimate, especially on the part of large publishing groups and large companies that are used to stipulating service contracts between managers without knowing in the least what they are talking about, demanding the best on the market. with the sole purpose of protecting their working position in the face of any problems, justifying their position as that of the one who made the choice of the market.
7 - What budget do you have?
An awkward question, almost rude apparently because in Italy talking about money (unlike other countries) seems vulgar, uncomfortable, inappropriate and even incorrect. Although it is the last of the points on the list it should probably be the first question.
It is often useless to think about hardware and software if we do not know the budget available, as it is useless to "save" with hardware sized at the limit of capacity where the available budget is much higher and you could easily have more powerful hardware leaving behind. useless calculations and related doubts concerning the sufficiency of the hardware on which the site will run.
In conclusion we can say that if you don't know how to choose a hosting or how choose a dedicated server you should know at least this data above and be able to answer these seven questions. If you have any doubts, you would like to try to understand which is the most suitable solution for your needs and know all the pros and cons, feel free to contact us.
We have a truly vast experience with many case studies ranging from the hobbyist, to portals and newspapers and sectorial sites with over 50 million monthly visitors per month.