How many servers do I need for my business? This may seem like a straightforward question. For more users, you need more servers. For fewer users, you need fewer. While this is true, when you actually get down to answering the question, it often reveals itself to be more complicated than it first seemed. It can be especially challenging if you’re unsure of factors such as how many users you’ll have.
In fact, many different considerations go into determining the right amount of servers for your business’ needs. This guide will walk you through the factors you need to keep in mind when making this critical business decision.
Defining “Server” and Types of Servers
Before you can decide how many servers you need, you should understand at least the basics of what a server is and the functions it performs.
A server is a computer that sends data to other computers connected to it by a network and manages network resources. Servers connect various disparate devices and allow them to share information and access to software. You might use a server to provide processing power, data storage, data backup services or to create a central location for data management. Servers tend to have large amounts of memory and powerful processors. They’re designed for 24/7 operation and exceptional reliability because they are essential to keeping IT services available.
You can use servers to enable various IT functions or groups of functions. In smaller organizations, a single server might run all of the business’ desktop computers. Other companies might have servers dedicated to single tasks such as managing email or running a compute-intensive software program.
Different types of servers work best for different kinds of tasks. You might choose different hardware for hosting a website, for example, than for providing database services and yet another kind for managing shared equipment such as printers.
You also have the option of choosing between physical servers and cloud servers. Physical servers require you to have space to store your hardware onsite and to have the IT resources and capabilities to manage it. With cloud-based servers, a third party hosts the hardware for you and provides you with access to the services through an Internet connection. This has the advantage of freeing you up to focus on non-IT tasks and doesn’t require you to have space to store equipment yourself.
One reason that answering questions such as, “How many servers do you need for one million users?” is more complicated than it seems is that servers vary widely in terms of their capabilities. A small number of very powerful servers can take the place of many less impressive ones. Three main aspects of server hardware determine the capabilities of any given server.
You can think of the processors as the brains of the server. They are responsible for responding to the instructions that enable the operation of a computer. The faster your processors are, the more programs you can run and the faster you can run them. Today, most servers use multi-core processors, which essentially means that they have multiple processors that are plugged into the same socket instead of being separate. This enables them to communicate with each other faster, resulting in improved speed and efficiency.
2. Random Access Memory
Random Access Memory, commonly referred to as RAM, is like the short-term memory of a server. It keeps information needed for the task at hand readily available but doesn’t store data permanently. Accessing data in RAM is much faster than using other types of storage such as hard disk drives. If your RAM fills up, the computer must go to the hard disk to retrieve the necessary data, which results in slower operation. Because of this, the more RAM you have, the faster your server can operate.
3. Hard Drive Space
The hard drive is the longterm memory of your server. The amount of hard drive space you have determines the total amount of data your server can store. When choosing a server, you need to ensure that it has enough hard drive space to store all of the data needed for the functions you need it to perform. Longterm data storage is a growing concern among businesses since global storage requirements are increasing by around 40 percent every year as the amount of data we collect grows.
These three features make up a full server package. Once you determine the overall capacity you need, you will consider these three components and make sure each of them meets your requirements.
Factors to Consider
So, how do you know how much capacity you need? Keep the following elements in mind when making your decision.
Functions of the Servers
Understanding the kinds of tasks your server will handle is essential for determining the capabilities you need. Some uses will rely more on RAM, while others depend heavily on hard drive space. Requirements for other applications will focus mostly on computing power, in which case processors will be the primary component you consider.
For example, if you are using a server for data backup or as a file server, you will need more hard drive space and may look for a server with multiple bays for hard drives. A server used to host a website or database that gets a lot of requests from users should have large amounts of RAM. A server that will run a video game, power financial modeling software or handle data mining will need lots of processing power. A server that powers equipment such a printer typically doesn’t need much power.
While different applications need more from either processors, RAM or hard drive space, you shouldn’t focus on one component to the exclusion of all others. Every server use requires some of each functionality.
Number of Users
The number of users that rely on a server is another critical factor, especially for servers that may get high volumes of users at particular times, such as those used for website hosting. To determine the number of users that each server can support, take the amount of processing power and memory that each request from a user requires and multiply that by the number of users.
There are numerous ways to measure the number of users. Users per minute describes the average number of users over a given time frame. Concurrent users describes the amount that use the server at one time. You should make sure that your server can handle the number of concurrent users it would have during times of peak traffic, when the highest volume of people use the server. If you consider a business’ website, for instance, this might occur after the company makes an important announcement or a piece of marketing content takes off. If you don’t account for peak traffic, your site may work slowly or crash completely at the height of its popularity.
Sometimes, especially for servers used internally by a company, it’s apparent how many users you will have. For instance, if you’re a small business with 10 employees and are purchasing a server to handle your company’s email, you know that 10 people will be utilizing that server. Sometimes, though, it’s less obvious how many users you have. If you’re launching a new product, website or software, for instance, you will have to predict the future demand to determine how many servers you require for your website or software.
You can do this by testing the server privately by using it internally or utilizing virtual users, which simulate actual users, to perform a load test.
If this isn’t possible in your situation, you can use a similar server as a reference. Measure the number of requests made to the server during a given period and then multiply that by that amount of power needed for the server’s tasks. If you don’t have a server of your own to reference, you can use your market research or study similar servers used by other businesses.
Redundancy and Backup
It’s critical that servers operate reliably and often continuously so that the services they enable are available. Dependable server operation is often crucial to a business’ success. Because of this, servers typically have multiple redundancies built into them so that if one component malfunctions, another one kicks in to keep the system up and running. They may have redundant hard drives, processors and power supplies. Dividing the workload between multiple servers can also increase reliability, a strategy known as server load balancing. These extra components and servers increase the costs of your server system and the power needed to run it, but it’s typically worth the extra expense since it greatly reduces the risk of a complete shutdown if an error occurs.
How to Calculate System Capacity
Once you’ve figured out all of the factors that impact the number of servers you need, you can do some calculations to get a solid answer to the question. If you’re setting up an entirely new server and have little background information on how much capacity you’ll need, you can use the estimations you make based on the complexity of the server functions, the expected volume or users and the redundancy required for adequate reliability.
Take the amount of capacity that one user would require when performing the associated function, such as making a database query or storing the contents of an email account. Then, multiply that by the expected number of users. Next, add in the extra capacity you need to ensure reliability, and you have an estimate of the amount of capacity you need. You can then find a server or servers that meet those requirements.
Once you’ve run the server for a while, you should periodically test whether the capacity you have is sufficient or whether you need to expand it. You can do that by observing how the server operates over a given period. You can use various mathematical formulas to calculate your capacity needs.
The three basic variables you’ll use in your calculations are:
- Observation time (T): the amount of time during which you monitor a server for activity
- Busy time (B): the amount of time during which the server is active during the observation period
- Completions (C): the number of transactions executed during the observation time
Using these three values, you can calculate various useful figures including:
- Processor utilization (B/T): the portion of the processor power used during a given period
- Transaction throughput (C/T): the average number of transactions during a given period
- Average service time (B/C): the average amount of time needed to perform a transaction
Risks of Over- and Under-Preparation
Estimates are, by their nature, not exact. You’re not going to guess the precise amount of capacity you’ll need. So, is it better to err on the side of having too much capacity or too little?
If you end up with too many servers, you won’t have to worry about running out of capacity, but you will have IT costs that are higher than necessary. The money, time and other resources you put into the extra servers will essentially be going to waste and could have been used more productively to grow your business.
If you don’t have enough servers, your software, website or other server-backed resources may run slowly or even crash, rendering them unusable. You might find that your website can’t perform during times of peak load, potentially robbing you of some of your best opportunities for expanding your business.
In both scenarios, you end up losing out on potential opportunities and growth. This can sometimes be highly destructive to a business.
Observing the operation of your server is essential for ensuring that you don’t have excess capacity or too little capacity. Given that you are adequately monitoring your server usage, overpreparing may be the better option since you can scale down your server capacity without incurring too much damage, besides unnecessary expenses. You may even be able to recoup some of your extra costs by selling the equipment you don’t need. The ideal outcome, though, is striking a balance and finding the number of servers per user that will perform best for your needs.
Servers and IT Services From Worldwide Supply
Worldwide Supply is your go-to source for new and pre-owned servers, as well as both physical and cloud-based solutions, from a range of leading OEMs. We test, configure and certify all of our products and can often ship them same-day. Purchasing pre-owned equipment from us could save you as much as 90 percent compared to purchasing from an OEM. We even offer a lifetime warranty on all of our pre-owned equipment.
You can also let us take the work out of IT for you. We provide full installation on all of our products and offer managed IT services including infrastructure management, maintenance programs, technical support and network security services. When you partner with Worldwide Supply, you get expert IT services and reduced IT management costs.
To learn more, feel free to explore our website and blog. Fill out this short online form to request a quote for our IT services and solutions today. We look forward to hearing from you and providing you with the answers to your IT questions.