Because of the mass shift to remote work the demand for terminal servers has increased. Meanwhile not all companies have an existing IT infrastructure and a system administrator to modify it for these tasks. It is not easy to make the right choice between a hardware server, virtual machine or hybrid solution, single server or cluster.
We'll help you to choose a hardware (physical) server, still you have a room for choice:
1. Place it in your own server room?
2. Choose a data center collocation?
3. Rent a dedicated server from a hosting solution provider?
And yes, the option you choose will greatly affect the capital investment and the final cost of infrastructure.
Whatever you choice is, this article will help all beginners understand the terminology and avoid mistakes when choosing a hardware terminal server. However, to deploy it you will need the system administrator help.
A user who works at a DEC terminal and VAX 11/780 mainframe.
Initially, the IT industry developed on the principle of centralized information processing. Private and public institutions bought large, expensive and very powerful servers (mainframes), and many specialists used it through terminals. Modern terminal servers are based on this concept.
IBM System/360 Model 20 mainframe
DEC VT terminal (1978)
A terminal server is a hardware and software complex with a large computing resources capacity (CPU, RAM, GPU, ROM), where software runs and data is stored. It is connected via a local or wide-area network to special devices (terminals), sends a certain image to them (screen sharing), and also tracks all user actions through input / output devices. All users are isolated from each other, as they work through a personal session. None of the ordinary users has administrator privileges, so cannot change any server settings. Some expensive configurations can simultaneously serve dozens or even hundreds of users, significantly reducing cost of office computers installation base.
Put simply, an abstract user has a computer, monitor, keyboard, mouse, etc. in his/her office. Server transmits an image over the network to the computer and tracks how the cursor moves and which keys are pressed. At the same time, this abstract user may not realize that all his actions are processed on a server located in another room / office / city / country. Everything looks routinely to him.
A terminal is any device for interaction between a user and a terminal server within the same network (local or wide area). Often, in order to save money, companies use outdated PCs as terminals or buy thin clients. However, when necessary any computer that meets the minimum system requirements can be used as a terminal, be it an expensive workstation, gaming PC, old netbook, tablet, or smartphone. So, if employees need to work from home, they do not have to take their computer from the office. They will have access to the remote desktop, software and files in a few clicks on a home laptop. Even open browser tabs will remain the same.
A thin client is a compact computer where unnecessary and expensive components and interfaces are simplified or removed. It contains only those parts that are necessary for elementary data processing and comfortable input or output. If an ordinary PC can be both a working tool, a multimedia combiner, and a terminal, a thin client is deisgned to serve exclusively as a terminal.
In the past, terminals occupied most of the desk, but today they are much slimmer.
But the price and size doesn't mean that thin clients are dummies. For example, Dell WYSE Thin 3040 boasts a quad-core Intel Atom x5-Z8350, integrated HD Graphics 400, 2 GB DDR3L RAM, 16 GB SSD. And this is an entry-level product. 20 years ago not every PC had such capacity. But nowadays this configuration is enough to launch OS on a thin client (Windows 7 Embedded, SUSE Linux, etc) and display pictures transmitted from a terminal server.
Perfect Pixar cartoon character: baby DELL WYSE Thin 3040 for about 20.000 rubles.
By 1980, personal computers had become much more affordable and powerful. This allowed companies to get rid of the expensive and clumsy mainframes, which had almost vanished by 1993. At that time, many predicted that the mainframes would completely disappear.
However, in 2006, the cloud computing era started, giving the new life for centralized computing. Classic cabinet-sized mainframes are still sold today, but many data centers base their infrastructures on small servers. They are more flexible and easier to modify and maintain. So, they perfectly meet the needs of small, medium and large businesses.
Racks with Dell servers in the data center
Anywhere access to a workspace. You hired an employee from another city or country? You can provide them with a fully-featured workplace in a few minutes. And if someone quits, administrator immediately blocks the account along with access to all enterprise data.
Remote desktops running on an ordinary Windows PC
Secure centralized infrastructure. All network traffic can be filtered with Network Firewall and custom filters; remote users can be easily controlled and restricted in their rights. For example, incoming non-VPN traffic will be immediately cut off; the accounting department will have access to 1C, but will not be able to enter the CMS, and vice versa for the marketing department. Moreover, if one of the employees leaves their laptop in a bar or engages in industrial espionage, the company data will remain safe on the server.
Alternative to expensive and complex install base. Server and thin clients are easier to manage, and this configuration is much cheaper, especially, in case of large offices. Instead of 100 expensive PCs you can buy a powerful server (or cluster) and 100 thin clients. As a pleasant bonus you can significantly lower energy costs. The most useful feature is that work can be instantly resumed from where the user stopped. They just need to log in.
PC fleet for a large office can cost a fortune.
Resource-intensive applications support. Since a terminal server is essentially more powerful and scalable than a standard PC, it can run greedy applications that will be available to all users via thin clients. And you don't need to duplicate data because one application instance will run for all sessions.
Mitigate risks. On the one hand, a failed server will block the work of all employees, but on the other hand, it will provide failover for hundreds of PCs that cost much more than a server. Server is a single point of failure for many users and workflows, so server infrastructure is backed up with redundancy. For example: additional power supplies, uninterruptible power supplies, diesel power generators, failover clusters, multiple backbone Internet providers, geo-distribution and much more.
Trying to choose the cheapest option is a big mistake. Some companies use one node as a terminal, database server and, for example, 1C server. Sometimes they don’t install a UPS, don’t buy a second power supply, don’t arrange disks into a RAID array, and don’t even backup data. And, for good measure, they also purchase cheaper hardware.
If you deploy everything on one node, and it fails in the midst of the working day (without clusters, redundancy and backups), be ready to face real chaos. And maybe a heart attack in case of a hard disk failure.
If you calculated the planned load wrong, then system will lag either from the first day or in the near future, when it is overloaded by users.
I hope you are not a samurai who has no goal, only path. This article will help you to choose an adequate terminal server with enough performance and reliability. And no stress and sound sleep as a bonus.
So consider the following aspects:
1. A set of services and applications you will be using
Depending on the applications used, you may need more than one node. Technically, you can use one server for all tasks, but it would be like putting a compact car engine into a tank. I do not recommend doing this. For a terminal server and database, it is better to use at least two nodes or virtualization.
2. Number of current users and future growth
If company staff is constantly growing, existing infrastructure will not last long. We recommend to choose a server base where you can put more than one processor, and add RAM-cards and/or disks.
3. Hardware
Disks with high IOPS capacity (ideally NVMe, but for cold data SAS HDD is OK), as well as a good disk controller with cache. This will not only increase communication speed, but also significantly enhance the RAID array resiliency.
So that servers do not work at their limits, provide a 20-30% spare capacity for the future. For example, now you need 128 GB RAM. Add extra 32 GB. This way your system will get more flexibility, so at peak loads your work will not turn into a slide show.
Vertical scalability is one of the advantages that servers have over PCs. You can install 2, 4 and even 8 processors (4×2 in Dell FX2S), and several terabytes of RAM per socket. Therefore, if you choose a basic system taking into account your future needs, it can be scaled several times. The second option is horizontal scalability via clustering and increasing the number of nodes. If you ignore this aspect, you will have to choose the third option (the most expensive), that is a complete hardware replacement.
4. Resiliency
Servers are essentially more reliable than PCs; besides, you can increase resiliency at a node level. But even 4 power supplies, 20 disks in RAID-10 and connection to multiple ISPs will not help if the server's motherboard burns out. So, a failover cluster with a load balancer is a good way to enhance reliability. Although this solution is quite expensive, it is justifiable for many companies. Compare possible downtime losses, reputational damage, and foregone earnings with the cost of an additional server. Thus, you will understand whether you need a cluster or not.
A terminal server is an optional component of IT infrastructure. If your office has 5-10 employees, then you can use the most simple solution, which is several PCs and backups of important data. But sooner or later switching to a modern option will give you more advantages.
A terminal server is an excellent choice for medium, large and branch offices. It will save time and reduce capital investments in large IT infrastructure. If your goal is to protect sensitive data from leaks and theft, a terminal server will also perform this task well.
At Servermall, we have been using terminal servers for several years. If you have any questions, we will be happy to help, give advice, and select the best server for your needs.