Scale Server in the cloud
Undoubtedly, cloud computing is currently a trend in the hosting services market. Almost every hosting provider providing virtual server rental services says that it works in the “cloud”. Often, these statements are just a beautiful marketing cover that hides the traditional approach to providing VDS / VPS rental services.
Following the market trends, we set aside marketing candy wrappers and are pleased to announce our new service on Scale Server - Scale Server .
As part of the Scale Server service, we tried to implement all the principles of Cloud Computing, in our opinion, we did it perfectly. What is the difference between Scale Server and virtual dedicated server in the classic incarnation? I dare to assume that this question will be the first one you want to ask us. In this announcement, we will try to answer this question. To do this, we consider the defining principles of the technology of cloud computing and describe how these principles are implemented as part of the Scale Server service.
One of the main principles of cloud computing is the “On-demand self service” approach, which means that the necessary amount of resources will be available to the service consumer at any given time. At the same time obtaining the required resources should take a minimum amount of time.
')
Currently, many of the providers provide their customers with the opportunity to independently choose the amount of resources in their cloud infrastructure, as well as manage these resources through the control panel or using specialized APIs. Those. the service user is invited to independently monitor the loading of leased resources and take actions related to an increase or decrease in leased resources.
As part of the Scale Server service, our clients receive a virtual server with an automatically scalable RAM resource (up to 56GB) and no restrictions on the use of processor cores (14 cores). Your virtual server will receive the required amount of RAM without your participation. After the need for RAM disappears, the memory allocation will be automatically reduced to the level required by the operating system. The response of our system to the need to increase or decrease the size of RAM is from 1 to 20 seconds. All that our client has to do when creating a virtual server is to select the range of dynamic changes in RAM.
“Metered use” is a second, equally important financial aspect of cloud services. Almost all cloud service providers provide their services with hourly rates. Of course, this approach is very user friendly, but we decided to go even further.
By providing our customers with Scale Server, we have reduced the time range for calculating the cost of RAM resources to 1 minute, and we rate processor usage based on seconds of computer time. As a result, our client pays only for the actually consumed resources in the shortest possible time interval. Disk resource allocation is charged on an hourly basis, which spoils the picture somewhat, but this approach is justified from an economic point of view - by reducing our indirect costs, we can offer our customers favorable prices for the rental of a disk resource.
“Elasticity” - elasticity of the resource purchase volume. The translation turned out to be quite peculiar, but fully reflects the essence of the approach. Each of our clients can use both a small amount of resources over a long period of time, and the maximum possible during a short time interval. At the same time, the need to plan the amount of necessary resources fades into the background in the way that our system solves this problem.
All resources that are ordered by our customers become available within a few minutes after their payment. Subsequently, you can change the parameters of the order yourself, without contacting technical support. We practice this approach not only for the provision of the Scale Server service, but also for the classical virtual server rental scheme.
Any remote service loses meaning if it is not accessed. One more principle of cloud computing regulates exactly this question. “Ubiquitous network access” is the principle of network accessibility. The solution created by our client should be available at any time, from any computer or mobile device connected to the Internet.
We will not dissemble and say that we fully comply with this approach. To do this, we lack, at a minimum, the replication of our cloud with a similar infrastructure solution located in another data center. This would completely eliminate access problems that could be caused, for example, by an accident in the data center or malfunction of network equipment. At the moment we are working on it, and next year our cloud will be geographically distributed across various data centers both within Russia and in Europe. Today, we only replicate the backup system of user data between remote data centers.
The independence of application availability and the continuity of the virtual infrastructure lies in the “Resource pooling” approach. Our clients do not know, and, most likely, it doesn’t matter on which particular node its virtual server is currently running. When a single hardware node fails, users should not be inconvenienced by the inoperability of their services. When the hardware node on which our client’s virtual machine was “fallen down”, the system will automatically restart an unavailable resource on other nodes that are currently running. The disk resources of our clients are replicated four times, which increases the level of data safety by several times compared to traditional solutions based on RAID arrays.
Our cluster uses server hardware and Hewlett Packard storage systems. The solutions offered by this manufacturer are time tested. The reliability of Hewlett Packard equipment allows you to minimize downtime in the provision of services, thanks to its reliable operation.
So, what does our client receive as part of the Scale Server service:
In order to create a Scale Server and start using it, it is necessary and sufficient to have only 100 rubles on your personal account (for a virtual server with a 5GB disk and without a prepaid traffic package).
Our current customers can create Scale Server through the resource control panel . New customers can order Scale Server from the main page of our site .
PS Announcing in September the service of providing virtual servers for rent, we made a mistake, not bothering to provide our customers with a server management panel. We are correcting :)
To all customers, we can offer to install and configure our technical support through the ISP Manager control panel. Our customers can rent a panel or purchase a license to use it:
Rent ISP Manager
Customers who replenish the account at the same time more than 6 than 2,500 rubles receive an ISP Manager Lite license for one virtual server for free. If you replenish your account for more than 4000 rubles - ISP Manager Pro license for one virtual server is free. In this case, the license is not transferred to the client’s property in the same way as when a license is purchased, and is valid for the entire duration of the virtual server operation.
For purchase and licensing of ISP Manager, please contact max@clodo.ru .
Following the market trends, we set aside marketing candy wrappers and are pleased to announce our new service on Scale Server - Scale Server .
As part of the Scale Server service, we tried to implement all the principles of Cloud Computing, in our opinion, we did it perfectly. What is the difference between Scale Server and virtual dedicated server in the classic incarnation? I dare to assume that this question will be the first one you want to ask us. In this announcement, we will try to answer this question. To do this, we consider the defining principles of the technology of cloud computing and describe how these principles are implemented as part of the Scale Server service.
Resources on request
One of the main principles of cloud computing is the “On-demand self service” approach, which means that the necessary amount of resources will be available to the service consumer at any given time. At the same time obtaining the required resources should take a minimum amount of time.
')
Currently, many of the providers provide their customers with the opportunity to independently choose the amount of resources in their cloud infrastructure, as well as manage these resources through the control panel or using specialized APIs. Those. the service user is invited to independently monitor the loading of leased resources and take actions related to an increase or decrease in leased resources.
As part of the Scale Server service, our clients receive a virtual server with an automatically scalable RAM resource (up to 56GB) and no restrictions on the use of processor cores (14 cores). Your virtual server will receive the required amount of RAM without your participation. After the need for RAM disappears, the memory allocation will be automatically reduced to the level required by the operating system. The response of our system to the need to increase or decrease the size of RAM is from 1 to 20 seconds. All that our client has to do when creating a virtual server is to select the range of dynamic changes in RAM.
Payment on the fact
“Metered use” is a second, equally important financial aspect of cloud services. Almost all cloud service providers provide their services with hourly rates. Of course, this approach is very user friendly, but we decided to go even further.
By providing our customers with Scale Server, we have reduced the time range for calculating the cost of RAM resources to 1 minute, and we rate processor usage based on seconds of computer time. As a result, our client pays only for the actually consumed resources in the shortest possible time interval. Disk resource allocation is charged on an hourly basis, which spoils the picture somewhat, but this approach is justified from an economic point of view - by reducing our indirect costs, we can offer our customers favorable prices for the rental of a disk resource.
Elasticity
“Elasticity” - elasticity of the resource purchase volume. The translation turned out to be quite peculiar, but fully reflects the essence of the approach. Each of our clients can use both a small amount of resources over a long period of time, and the maximum possible during a short time interval. At the same time, the need to plan the amount of necessary resources fades into the background in the way that our system solves this problem.
All resources that are ordered by our customers become available within a few minutes after their payment. Subsequently, you can change the parameters of the order yourself, without contacting technical support. We practice this approach not only for the provision of the Scale Server service, but also for the classical virtual server rental scheme.
Availability
Any remote service loses meaning if it is not accessed. One more principle of cloud computing regulates exactly this question. “Ubiquitous network access” is the principle of network accessibility. The solution created by our client should be available at any time, from any computer or mobile device connected to the Internet.
We will not dissemble and say that we fully comply with this approach. To do this, we lack, at a minimum, the replication of our cloud with a similar infrastructure solution located in another data center. This would completely eliminate access problems that could be caused, for example, by an accident in the data center or malfunction of network equipment. At the moment we are working on it, and next year our cloud will be geographically distributed across various data centers both within Russia and in Europe. Today, we only replicate the backup system of user data between remote data centers.
fault tolerance
The independence of application availability and the continuity of the virtual infrastructure lies in the “Resource pooling” approach. Our clients do not know, and, most likely, it doesn’t matter on which particular node its virtual server is currently running. When a single hardware node fails, users should not be inconvenienced by the inoperability of their services. When the hardware node on which our client’s virtual machine was “fallen down”, the system will automatically restart an unavailable resource on other nodes that are currently running. The disk resources of our clients are replicated four times, which increases the level of data safety by several times compared to traditional solutions based on RAID arrays.
Our cluster uses server hardware and Hewlett Packard storage systems. The solutions offered by this manufacturer are time tested. The reliability of Hewlett Packard equipment allows you to minimize downtime in the provision of services, thanks to its reliable operation.
So, what does our client receive as part of the Scale Server service:
- the hosting project clodo.ru corresponding to all the principles of Cloud Computing, provided as part of the IaaS service - infrastructure as a service
- the ability to scale memory resources from 256MB to 56GB
- CentOS operating system (work on the kernels of other systems is actively being carried out by our specialists)
- reliable data storage built on four-replication SAS disks
- High access speed to disk resources (Disk storages are connected to computing nodes via the Infiniband QDR bus)
- daily, free backup of the virtual server state
- technical support 24/7/365
In order to create a Scale Server and start using it, it is necessary and sufficient to have only 100 rubles on your personal account (for a virtual server with a 5GB disk and without a prepaid traffic package).
Our current customers can create Scale Server through the resource control panel . New customers can order Scale Server from the main page of our site .
PS Announcing in September the service of providing virtual servers for rent, we made a mistake, not bothering to provide our customers with a server management panel. We are correcting :)
To all customers, we can offer to install and configure our technical support through the ISP Manager control panel. Our customers can rent a panel or purchase a license to use it:
Rent ISP Manager
- Lite version -
3.5 Euro per monthfor free - Pro version - 250 rubles per month
Customers who replenish the account at the same time more than 6 than 2,500 rubles receive an ISP Manager Lite license for one virtual server for free. If you replenish your account for more than 4000 rubles - ISP Manager Pro license for one virtual server is free. In this case, the license is not transferred to the client’s property in the same way as when a license is purchased, and is valid for the entire duration of the virtual server operation.
For purchase and licensing of ISP Manager, please contact max@clodo.ru .
Source: https://habr.com/ru/post/107445/
All Articles