Cloud hosting types according to their use
There are currently three types of cloud hosting in the marketplace, even if most people know about just two of them. IaaS – Infrastructure as a Service is the first popular form of the cloud. It entails providing an infrastructure (hardware) by means of a service. Hardware means servers, storage, active elements (such as switches, routers, controllers, firewalls) and a load balancer in this context. Providers will connect and adjust all of this and take care of the operation, so the infrastructure is in the form of a service. With this service, you will obtain both hardware and the know-how of the company (the above-mentioned connection, administration and servicing). It can be noted that the “IaaS” cloud is very similar to the VPS we discussed in the previous article. The main difference is that in the case of the IaaS, everything is handled redundantly, so that a failure of the service is prevented to the maximum possible extent. Moreover, more advanced virtualisation is used. This makes it different even from rentals of dedicated servers; the cloud solution is more accessible and can be adjusted in greater detail, as its settings have more options. Of course, even in the market, there are companies which sell cloud hosting very cheaply, almost at the level of VPS hosting. However, this type of cloud will probably not offer the necessary robust and complex level of implementation.
The second, less known, form of the cloud is PaaS – Platform as a Service, something between IaaS and SaaS. When you look in wikipedia, you can see examples such as Google Apps Engine and Salesforce. You will often find out that PaaS means IaaS, where an entire operation system (OS) is installed, sometimes even the entire LAMP package, and the cloud provider takes care of all of this. Here, I think, the border between what is IaaS and what is PaaS is very thin. PaaS can also be seen as a set of developer tools, by means of which the final product can be made (a software or web service) without having to start from the very beginning.
The third alternative, which we come across most often as end users, is SaaS – Software as a service – i.e. programs or applications as a service. In the past, you could buy a program on a CD and install it on your computer. Today, the tendency is to buy a program or service “in the cloud”. This means that the program does not need to be installed anywhere; you can use it immediately on all devices with a web browser (cellphone, PC, tablet, notebook or TV). Classic examples include MS Outlook (mail client) and MS Office (office applications package), which are still often installed on computers today. As regards the corporate environment, companies also need to install an MS Exchange server in the background in order for the mail client to work. To sum it all up, you need to have someone install the software on the computers, then on the servers and then adjust everything and take care of the servers. An alternative may be SaaS – an application as a service, for example when Google offers the package Google Apps for companies as a service, where the cost for one corporate user may be 40 € per year, for which the user receives a mail client and the Google Docs office package, including the calendar, file sharing and Google Hangouts (chat, calling and videoconference). All this works through a web browser, so – with the exception of the initial simple adjusting and employee training – you do not need to worry about anything. From this point of view, you will save money on licences for program packages installed on the computers and you will not have to buy servers and storage which your IT administrators would have to take care of.
Cloud hosting types according to their implementation
There are three main forms of cloud implementation. As regards cloud hosting offered by companies on the internet, the first type is the public cloud. This solution, including such things as active elements, servers or storage, can be shared by more users. However, it is necessary to remember that the particular used performance, such as RAM, CPU, is not shared. The disk space is not shared either, just IOPS – that is reading/writing per second. If anyone requires the guarantee of IOPS, the only solution is the second form of cloud: the private cloud. Here, particular physical servers and storage (hard disks) are dedicated to one customer. Or an entire solution is built for larger projects. The private cloud is used both if you want IOPS guaranteed or you are a big company which needs its data stored at its own place. It can also be implemented for reasons of various licence policies or certifications. Sometimes we discuss the hybrid cloud with our customers, which is probably the most frequently used corporate cloud. It is a combination of a private and public cloud. The hybrid cloud is used by companies which do not want to let their precious data leave the company or they just need to keep some corporate applications within the company, but at the same time, the company uses applications which must be in a public cloud, such as a mail server, web pages or CRM. Both environments – a private cloud (installed in the central office of the company) and a public cloud housed by the provider (installed in a datacentre on the backbone of the internet) – can be interconnected or they can work independently of each other.
How does a cloud work
The public cloud hosting provider has several racks in the datacentre which are connected to the internet. The rack includes servers, storage, firewalls and routers, all interconnected by a switch and SAN disk controller. The cloud is connected in such a manner that a failure of any hardware element can be handled. The technology of two vendors (hardware manufacturers) at most is often used to achieve the optimum synergy between the active elements. Another layer under the hardware is virtualisation. The two most famous virtualisation solutions are VMware and HyperV. You can find others in the list in wikipedia. So far, this concerns IaaS, so you will get performance, disk space and licences for the operation system (OS), database (DB) and the like. Then the environment for the operation of the application is installed and you can sell a web application as a service – SaaS.
Connecting the cloud
If the company considers transferring servers onto the cloud, this change is often accompanied by connecting the cloud to the corporate network. Each cloud is connected to the internet, but many customers require that the cloud be connected to their MPLS VPN (corporate network). Alternatively, an Ethernet service (L2 connectivity) is created – connecting the cloud and the company central office. Telecommunications companies have a great advantage in this respect, and this is the reason why they sell the cloud.
A very important chapter of the cloud is the backup. The backup of the hardware part is managed by the provider, so you do not need to worry about that. So why should you worry about backing up your operating environment in a public cloud? There are many reasons, but the main one is human error. Each such error is usually discovered within several minutes or hours at most. So you need to have your work environment backed up going back 14 to 30 days. This is usually not included in the solution, and you must pay an extra charge for it or handle it another way. Generally, you can back up the entire disk plus increments or you can back up a particular application, database, file or anything else by means of an agent. Weeam and CommVault are two most widely used backup tools in the cloud.
How do you calculate the cloud
As in the case of VPS, we need to clarify the units we pay for in a public cloud. The processor (CPU) especially is calculated differently by many providers. Then, we need to clarify the storage type (SATA, SAS or SSD) with regard to the IOPS speed. You can pay either in the classic way for cloud hosting – for reserved performance per month. An alternative option is to pay per hour according to what you really use. There are also combinations, where you pay for particular dedicated performance per month and then you have the option to manually add or reduce the performance and you pay for the added performance per hour. The payment per hour option enables you to save money, as you can activate the testing and development servers just once in a while, rather than permanently, so you do not need to pay as much for them (an exception is the disk space, which you must still pay for). Owing to this, you do not need to have dedicated servers somewhere for testing purposes.
Latest posts by Radek Kucera (see all)
- How Does the Internet Work? - 2016-08-19
- Tutorial: How to redirect domain.com to blogger.com - 2016-02-24
- How to create your own web pages - 2015-09-25