May 8, 2015 at 14:48Choosing Cloud Hosting: Where to Start?
Dear Habrazhiteli, after experimenting with Samara coworking and other projects, we finally decided to start writing something for Habr on the main type of our activity, that is, hosting, servers and related technologies. To begin with, I would like to write about cloud resources that are now popular among developers and compare the main players in this market.
The word “cloud” is well known to all site owners. Cloud technologists are provided as a service - there is access to resources (RAM, processor, storage devices, etc.), without specifying where exactly at what point the resources are located.
“Cloudiness” implies the possibility of providing resources with a high guarantee of their availability. In other words, if the hardware resources on a part of the physical computers of the cloud fail or are disabled, the cloud as a whole can continue to function (in some cases without significant loss of efficiency), or even restore a working state automatically ("self-healing", self-healing).
In case there is a need to choose a cloud provider, below is a list from which you can start dating. All of them are suitable including for hosting sites.
The most budgetary of all cloud providers for the mass user, Digital Ocean (hereinafter DO) offers a minimal but sufficient toolkit for fully automatic resource management. The typical time for creating and starting a new virtual machine is no more than one minute.
DO provides the ability to create “snapshots”, backup copies of your servers (currently the service is free). The snapshot captures the state of the virtual machine at the time of its creation - you can later return the machine to this state by recovery (unlike starting a new machine, recovery can take much more time). The snapshot can be transferred to another user - this way you can configure the turnkey virtual machines.
Extras: DO users have access to a large knowledge base with typical recipes for configuring virtual machines for specific needs, specific software.
In American data centers you can get the internal IP address for the virtual machine. The exchange rate on the internal grid reaches 1Gbps, which allows, for example, to form cluster structures from virtual machines.
Since August 2014, DO began to support the IPv6 family of protocols.
The API allows you to fully automate all actions with virtual machines (everything that is accessible from the interface in the browser is accessible through the API).
Vultr largely duplicates the services provided by Digital Ocean, while at the same time it provides an opportunity to check the service for free.
It also makes sense to pay attention to Vultr if you need IPv6 support in all available data centers.
Before you start using the Vultr services, it is recommended that you read their user agreement and other documents governing the relationship between the user and the provider, in order to avoid possible misunderstandings in the future.
Vultr and DO offer only Unix-type operating systems (from a limited list of versions and architectures).
DO and Vultr offer a fixed list of machine types (by the number of allocated resources). In case you need the ability to assign resources yourself, you can contact CloudSigma.
Creating a virtual machine in conditions of such a “constructor” requires more time. It also makes sense to use CloudSigma services if you want to install the OS from your own boot image (you can download the required ISO file and install from it).
CloudSigma provides an interface and technical support including in Russian.
Provider data centers are located in Zurich, San Jose, Las Vegas and Washington.
The service also has a developed API - you can manage all resources and enter all virtual park services automatically.
CloudSigma is optimal if you need to store and configure servers on exotic OS types, or in an “atypical” configuration.
The Italian provider Prometeus provides a cloud service (iwStack) where you are provided with an API and a set of several types of virtual machines. Technical support in English and Italian only.
Like all the providers already listed, Prometeus allows you to dynamically redefine the type of virtual machine (you can transfer your server to one of the older models of the virtual machine without data loss; the reverse procedure is also possible).
Data centers are located in Milan (Italy), Dallas (USA) and Pune (India).
Amazon Web Services (AWS), the cloud patriarch, provides almost all the services (like platforms) needed to create a variety of network resources (including virtual machines, data warehousing, database services, resource monitoring, etc.).
AWS is aimed at people well versed in system administration. All APIs and services are well-documented - you can understand them without the participation of (paid) technical support of this provider.
For virtual machines, AWS does not provide arbitrary configuration capabilities.
The review does not mention such well-known cloud services as Windows Azure, RackSpace, GigeNet Cloud, SoftLayer and dozens of others, first of all - so as not to oversaturate information during the first acquaintance with the reader. The above providers are suitable both for the first acquaintance with cloud technologies, and for using them on an ongoing basis.
The main advantage of all of them is the ability to train in server settings and work with different OSs with the ability at any time to restore the initial (or saved by your efforts) state of a virtual machine in a few minutes.
Introduction
The word “cloud” is well known to all site owners. Cloud technologists are provided as a service - there is access to resources (RAM, processor, storage devices, etc.), without specifying where exactly at what point the resources are located.
“Cloudiness” implies the possibility of providing resources with a high guarantee of their availability. In other words, if the hardware resources on a part of the physical computers of the cloud fail or are disabled, the cloud as a whole can continue to function (in some cases without significant loss of efficiency), or even restore a working state automatically ("self-healing", self-healing).
In case there is a need to choose a cloud provider, below is a list from which you can start dating. All of them are suitable including for hosting sites.
Digital ocean
The most budgetary of all cloud providers for the mass user, Digital Ocean (hereinafter DO) offers a minimal but sufficient toolkit for fully automatic resource management. The typical time for creating and starting a new virtual machine is no more than one minute.
DO provides the ability to create “snapshots”, backup copies of your servers (currently the service is free). The snapshot captures the state of the virtual machine at the time of its creation - you can later return the machine to this state by recovery (unlike starting a new machine, recovery can take much more time). The snapshot can be transferred to another user - this way you can configure the turnkey virtual machines.
Extras: DO users have access to a large knowledge base with typical recipes for configuring virtual machines for specific needs, specific software.
In American data centers you can get the internal IP address for the virtual machine. The exchange rate on the internal grid reaches 1Gbps, which allows, for example, to form cluster structures from virtual machines.
Since August 2014, DO began to support the IPv6 family of protocols.
The API allows you to fully automate all actions with virtual machines (everything that is accessible from the interface in the browser is accessible through the API).
Vultr
Vultr largely duplicates the services provided by Digital Ocean, while at the same time it provides an opportunity to check the service for free.
It also makes sense to pay attention to Vultr if you need IPv6 support in all available data centers.
Before you start using the Vultr services, it is recommended that you read their user agreement and other documents governing the relationship between the user and the provider, in order to avoid possible misunderstandings in the future.
Vultr and DO offer only Unix-type operating systems (from a limited list of versions and architectures).
CloudSigma
DO and Vultr offer a fixed list of machine types (by the number of allocated resources). In case you need the ability to assign resources yourself, you can contact CloudSigma.
Creating a virtual machine in conditions of such a “constructor” requires more time. It also makes sense to use CloudSigma services if you want to install the OS from your own boot image (you can download the required ISO file and install from it).
CloudSigma provides an interface and technical support including in Russian.
Provider data centers are located in Zurich, San Jose, Las Vegas and Washington.
The service also has a developed API - you can manage all resources and enter all virtual park services automatically.
CloudSigma is optimal if you need to store and configure servers on exotic OS types, or in an “atypical” configuration.
Prometeus
The Italian provider Prometeus provides a cloud service (iwStack) where you are provided with an API and a set of several types of virtual machines. Technical support in English and Italian only.
Like all the providers already listed, Prometeus allows you to dynamically redefine the type of virtual machine (you can transfer your server to one of the older models of the virtual machine without data loss; the reverse procedure is also possible).
Data centers are located in Milan (Italy), Dallas (USA) and Pune (India).
Amazon web services
Amazon Web Services (AWS), the cloud patriarch, provides almost all the services (like platforms) needed to create a variety of network resources (including virtual machines, data warehousing, database services, resource monitoring, etc.).
AWS is aimed at people well versed in system administration. All APIs and services are well-documented - you can understand them without the participation of (paid) technical support of this provider.
For virtual machines, AWS does not provide arbitrary configuration capabilities.
Conclusion
The review does not mention such well-known cloud services as Windows Azure, RackSpace, GigeNet Cloud, SoftLayer and dozens of others, first of all - so as not to oversaturate information during the first acquaintance with the reader. The above providers are suitable both for the first acquaintance with cloud technologies, and for using them on an ongoing basis.
The main advantage of all of them is the ability to train in server settings and work with different OSs with the ability at any time to restore the initial (or saved by your efforts) state of a virtual machine in a few minutes.