It has been widely acknowledged that the hybrid cloud is becoming the most important and popular variant of cloud computing. But in order to utilise it effectively it is important to understand that the hybrid cloud does not mean the end of dedicated hardware. If anything, the increased usage of I/O-heavy applications such as databases and Big Data platforms will necessitate a greater use of dedicated hardware in the future.
Major cloud players have recently been making moves which indicate the importance of dedicated computing for hybrid cloud services. For example, Amazon Web Services have a significantly reduced the price of their “EC2 dedicated instances”, while also reducing by a lesser proportion the cost of the dedicated instances themselves. This provides a clear validation of the fact that dedicated computing will continue to be an important component of IT infrastructure even in a cloud-based future.
Analysts at Wells Fargo have noted the fact that Amazon’s move indicates that it’s public pronouncements that all workloads should be migrated to the public cloud does not hold water. The reality is that hybrid setups offer more flexibility.
EC2 dedicated instances can be defined as EC2 cloud instances which are run using single-tenant hardware, and that are ultimately dedicated to a single customer account. While standard instances will remain popular, EC2 cloud instances after numerous advantages, most notably that compliance is improved via this system which benefits customers who would prefer not to share servers with other individuals or businesses. However, the downside of this setup is that it does not offer the sort of isolation which customers demand, and are delivered, on dedicated, bare metal servers.
When considering EC2 dedicated instances with bare metal servers, there are several key differences. Here are a few of the divergences between the two setups.
Dedicated servers operate in complete isolation from public clouds. With this particular setup you will operate on your own hardware within your own network, and be dependent on your own storage system. This can provide security advantages over EC2 dedicated instances.
When utilising EC2 dedicated instances, you are still using your own hardware, but are instead connected to a provider’s public cloud. You are effectively set aside a single tranche of that cloud. If the provider who is offering you the public cloud suffers any downtime, you will certainly be affected. Additionally, acquiring additional block storage can be problematical.
Dedicated servers after massive advantages in terms of customisation. The nature of dedicated service mean that they are highly customisable, and uses are able to choose a variety of elements within them. These include the operating system, storage, CPU, memory, network speed and and other factions of the cloud setup. A dedicated instance provided to you externally will be predetermined at source, offering you absolutely no options for customisation.
This is, of course, the key area for many businesses. After all, the reason most companies signed up for the cloud in the first place is because they help it will improve the performance of their business.
Thus, it is important to note that with dedicated servers there are no issues related to virtualisation. Your will not suffer any form of virtualisation penalty. When utilising a dedicated cloud instance, you’re still using a hypervisor, and this can have a significant impact on performance, particularly in relation to I/O and CPU.
Though no-one wants to see a reduction in performance, in this particular case it can prove particularly critical when attempting to run database programs. As you are still operating as part of the public cloud, it is a necessity to over-provision and over-engineer dedicated instances in order to allow for the fact that they are simply more likely to fail. Typically, failure occurs in these systems due to issues related to multi-tenancy, the hypervisor or the host service.
Eliminating the standard public cloud is often seen as a big plus point, however having your own workloads competing among themselves for the capacity of your dedicated instances can also be a negative influence on performance.
Dedicated servers tend to be more reliable. They are often backed by network uptime guarantees, and simply dedicating a form of hard work to a particular task naturally tends to produce better results.
It’s always difficult to truly assess this element without dealing with individual providers and services, but it is always important to consider the total cost which a business pays for a required level of performance and support.
Finally, hybrid technology offers a much greater flexibility and agility, and it is this above all else which has attracted businesses to it. When making the tricky switch to the cloud, hybrid offers an attractive halfway house that tends to SMEs in particular, with the possibility of moving applications around to run on and off-cloud as required appealing to many companies.