This NVTC guest blog post is written by Marc Burkels, manager of dedicated servers at LeaseWeb. LeaseWeb, an NVTC member company, is an Infrastructure-as-a-Service (IaaS) provider offering dedicated servers, CDN and cloud hosting on a global network. LeaseWeb recently exhibited at the Capital Cybersecurity Summit on Nov. 2-3, 2016.
Let’s say you want to become the new Facebook. Believe it or not, I regularly run into people who have this ambition. The number one question these new Mark Zuckerbergs ask me is which server they need.
It is always a challenge to convince them to not rush into anything. Instead, I have them sit down and tell me what they really want. Since many companies switch servers within a few months after buying and this is always time consuming (not to mention the costs), it is certainly worth your while to think well before you decide. What is the service you want to deliver? What is your workload? Does it involve large databases?
I always discuss the following 8 things to help people decide on the right hosting provider and hardware configuration of a dedicated server:
1. Business impact of downtime
What is the business impact of potential failure of your hosting environment? One of the first things to consider when selecting a dedicated server is how to deal with potential downtime. In a cloud environment, the setup of the cloud protects you against hardware failures. With a dedicated server, you know you are not sharing resources with anyone else. But since there is always a single point of failure in one server, you need to decide whether you are able to accept potential downtime – if you do not have the option to scale to multiple dedicated servers.
2. Scalability of your application
Scalability is another important issue when choosing a dedicated server. How well does your application scale? Is it easy to add more servers and will that increase the amount of end users you can service?
If it is easy for you to scale, it doesn’t matter whether you use a dedicated server or a virtual solution. However, some applications are difficult to scale to multiple devices. Making sure a database is running on multiple servers is a challenge since it needs to be synchronized over all database servers. It might even be easier to move the database to a server that has more processing capacity, RAM and storage. Moving to a cloud environment – where you can clone a server, have a copy running in production and can add a load balancer to redirect traffic to multiple servers – could also be a good option for you.
3. Performance requirements of your server
What are your performance requirements? How many users do you expect and how many servers do you potentially need? Several hardware choices influence server performance:
Generally , you can choose the amount of processors and cores in a server. It depends on the application you are running whether you will benefit from more cores (but any multi-threaded application will benefit from more cores, for instance web servers or database servers). Consider also the performance of the core defined in clock speed (MHz): some processors have a better turn-around time with less cores and more GHz per core. The advice on which processors and how many cores to choose will ideally come from someone who is managing the application or the vendor of the software. Of course, they need to also take into account the expected amount of users.
The faster the CPU and the more cores it has, the more RAM options are available to you. If you are unsure about your RAM needs, choose a server that allows you to add RAM if needed since this is relatively easy. The ranges of RAM choices, especially with double processors, are enormous.
The size of your server is important when choosing RAM, as is the latest technology. Current generation servers use DDR4-technology, which could have a positive effect on database performance. DDR4 is priced interestingly nowadays, since it is the standard.
Choose a RAID set-up for your hard drives, so you are well protected against the failure of a single hard drive. Your system will still be up and running – with some performance loss – until the hard drive is replaced.
The larger the server, the more hard drive options you have. SATA drives stand for high volume but relatively low performance. SAS performs twice as well as SATA, but has a higher price and lower capacity. SAS has been succeeded by SSD, which is 50 to 100 times faster than SATA.
4. Load balancing across multiple dedicated servers
If your application can scale across multiple dedicated servers, a form of load balancing where end users are split across all available servers- is necessary. If you are running a website and traffic is rising, at some point you will need to use multiple web servers that serve a multitude of users for the same website. With a load balancing solution, every incoming request will be directed to a different server. Before doing this, the load balancer checks whether a server is up and running. If it is down, it redirects traffic to another server.
5. Predictability of bandwidth usage
The requirements in bandwidth naturally relate to the predictability of data traffic. If you are going to consume a lot of bandwidth but predictability is low, you could choose a package with your dedicated server that has a lot of data traffic included, or even unmetered billing. This is an easy way of knowing exactly how much you will be spending on the hosting of your dedicated server.
6. Network quality
As a customer, you can choose where a dedicated server is placed physically. It is important to consider the location of your end user. For instance, if your customers are in the APAC region, hosting in Europe might not be a sensible choice since data delivery will be slow. Data delivery also depends on the quality of the network of the hosting provider. To find out more about network quality, check a provider’s NOC (Network Operation Center) pages and test the network. Most hosting providers will allow you to do this.
7. Self-service and remote management
To which degree are you allowed to manage your server yourself? If you are running an application on a dedicated server, you probably have the technical skills and the knowledge to maintain the server. But do you have access to a remote management module? Most A-brand servers are equipped with remote management modules. Providers can allow you secure access to that module.
A remote management module can also help if you are in a transition from IT on premise to a hosted solution (perhaps even a private cloud solution). It can be an in-between step that will leave existing work structures intact and ease the transition for IT personnel, since they will still be able to manage their own software deployments and the customized installation of an operating system.
8. Knowledge partner
And last but definitely not least: make sure your hosting provider involves his engineers and specialists when trying to find a solution tailored to your needs. A true knowledge partner advises on best practices and different solutions. This may involve combining different products into a hybrid solution.
The above will probably give you a good idea of what to consider before renting a dedicated server. If you are looking for specific advice or need assistance, please feel free to contact the LeaseWeb team. They can help you find the solution that is right for you.