Why a good IaaS-provider does not build its data center: 5 main reasons
Although providers with their own data center have certain advantages , including greater control over infrastructure, the disadvantages of this approach often outweigh. This is the need for a large initial investment, and the difficulty of scaling. Therefore, in today's article we are talking about why a good IaaS provider should not build its own data center.
/ photo Ecole Polytechnique CC
As they say in the research agency SP Home Run, to build your data center you will have to allocate about 10 thousand dollars per square meter. And data centers take up a lot of space.
')
For example, IBM’s London-based Softlayer can host 15,000 servers, while the SingleHop data center covers an area of ​​more than 4,300 square meters. If we use the SP Home Run approximation, the price of the latter is approximately 43 million dollars. And data centers the size of Google or Facebook can cost hundreds of millions.
Consideration should be given to server and storage costs, license purchases, electricity and cooling costs. Data centers are very “heated”, and it is believed that the cooling of “iron” takes most of the power consumption. And they consume a lot. For example, in 2012, Facebook built in the county of Crook, USA, a 28-megawatt data center. The same energy is consumed by the whole district.
Do not discount the additional costs of cable channels, lighting, fire systems and the server hardware itself. In terms of pricing servers are like cars - after buying their value immediately drops. In addition, hardware is regularly becoming obsolete, and if you follow the classical scheme , you have to upgrade every three years.
It turns out that the IaaS-provider owning its own data center, faces another item of expenditure - regular updating of the server fleet. Yes, old equipment can be sold, and the proceeds can be spent on the purchase of a new one, but often resale does not bring much benefit.
It is also important to understand that after the introduction of the data center, it will be necessary to employ trained employees who will monitor its condition. And although the number of people in the engine rooms can be small - according to Facebook, only 35 people worked in the data center of the city of Prinville in 2010 - they are still “supported” by thousands of system administrators working remotely. They also need to manage and pay wages.
It is believed that if the provider owns its own data center, it will be able to protect user data from all the "troubles." However, this is a dangerous delusion. The presence of the data center in the property often, on the contrary, is a disadvantage.
IaaS-provider with its data center falls into the conditions of serious competition. It will have to provide high-quality infrastructure services and at the same time “compete” with other data centers, which in many aspects will obviously win.
Data centers constantly fill their machine rooms with modern equipment. IaaS-provider, whose main purpose - the provision of cloud services, sooner or later fall behind. In addition, not every supplier with its own data center can be certified to the Tier level, which is a significant indicator of the reliability of the data center.
Moreover, “professional” data centers are investing funds and expertise in enhancing perimeter security. For example, the building of the Xelent data center in which IT-GRAD equipment is located has a multi-level security system with several perimeters. Entry and access to the territory is carried out strictly by passes - if the employee has forgotten his house ID, he will not be allowed into the workplace.
The facility has an armed guard, and the whole territory is covered with video surveillance cameras and motion sensors. The metal fence is equipped with vibration sensors.
/ photo Xelent / IT-GRAD cloud platform based on Xelent data center
For example, it concerns the standard PCI DSS. It is intended to unify the requirements of international payment systems to ensure the security of cardholder data. And bringing the infrastructure in line with the requirements of PCI DSS is a time consuming task.
The standard spelled out the safety standards, including the mandatory presence of video surveillance systems and the introduction of employee identification processes. The cloud provider receives these services "automatically", placing the "hardware" in the finished data center. For example, all IT-GRAD equipment is located in lockable racks under the supervision of cameras 24/7.
This allows the IaaS provider to focus on developing its own services and expand the range of services provided.
By outsourcing the management of the data center, the cloud provider gains access to a large number of telecommunication providers. This allows his clients to choose the best connection option for business and protect the infrastructure from equipment failures (due to the use of spare capacity). Additionally, the possibility of choosing an Internet provider creates a competitive environment that gives customers the opportunity to choose the most profitable connection packages for their high-performance services.
For example, in the DataSpace data center, where the IT-GRAD facilities are also located, come the trunk lines of 39 major telecom operators - these are Rostelecom, Gars Telecom, AKADO Telecom, Beeline, Megafon, MTS other. The building also has independent telecommunication inputs from two different MGTS wells at a distance of more than 100 meters from each other and six 100-millimeter cable channels leading to the main and additional data entry points (MPoE and SPoE).
Maintenance of the data center requires the availability of qualified personnel and time. It is necessary to monitor the cooling systems, power sources, safety.
According to Phil Collerton (Phil Collerton), the former head of operations at Interxion, there are a lot of "mobile components" in the data center that have to be taken into account. Everything from security systems to redundancy should work to the maximum capacity so that the provider can provide its services. As Phil says, data center service is a full-time job.
One person responsible for the performance of uninterruptible power supplies most likely does not have the same level of expertise in maintaining backup systems for data stored in the cloud. The creation of "universal soldiers" can tell on the quality of the services of the provider. Therefore, it is good if the maintenance tasks for hardware and software fall on the shoulders of different organizations.
A scenario in which the data center expands the list of services and begins to provide IaaS services is possible. However, it will not become a full-fledged cloud provider, since its main activity will be focused on equipment placement (co-location), data center lifecycle management and the provision of telecommunications services.
In this case, cloud services remain only an “expander” of the data center product range and constitute approximately 5–10% of the total number of services. And most often this happens due to the involvement of partners in the face of IaaS-providers, who consider the infrastructure of the data center as a reliable platform for placement.
PS Another couple of materials from the First Corporate IaaS blog:
/ photo Ecole Polytechnique CC
It is a big expense.
As they say in the research agency SP Home Run, to build your data center you will have to allocate about 10 thousand dollars per square meter. And data centers take up a lot of space.
')
For example, IBM’s London-based Softlayer can host 15,000 servers, while the SingleHop data center covers an area of ​​more than 4,300 square meters. If we use the SP Home Run approximation, the price of the latter is approximately 43 million dollars. And data centers the size of Google or Facebook can cost hundreds of millions.
Consideration should be given to server and storage costs, license purchases, electricity and cooling costs. Data centers are very “heated”, and it is believed that the cooling of “iron” takes most of the power consumption. And they consume a lot. For example, in 2012, Facebook built in the county of Crook, USA, a 28-megawatt data center. The same energy is consumed by the whole district.
Do not discount the additional costs of cable channels, lighting, fire systems and the server hardware itself. In terms of pricing servers are like cars - after buying their value immediately drops. In addition, hardware is regularly becoming obsolete, and if you follow the classical scheme , you have to upgrade every three years.
It turns out that the IaaS-provider owning its own data center, faces another item of expenditure - regular updating of the server fleet. Yes, old equipment can be sold, and the proceeds can be spent on the purchase of a new one, but often resale does not bring much benefit.
It is also important to understand that after the introduction of the data center, it will be necessary to employ trained employees who will monitor its condition. And although the number of people in the engine rooms can be small - according to Facebook, only 35 people worked in the data center of the city of Prinville in 2010 - they are still “supported” by thousands of system administrators working remotely. They also need to manage and pay wages.
Increased reliability and security
It is believed that if the provider owns its own data center, it will be able to protect user data from all the "troubles." However, this is a dangerous delusion. The presence of the data center in the property often, on the contrary, is a disadvantage.
IaaS-provider with its data center falls into the conditions of serious competition. It will have to provide high-quality infrastructure services and at the same time “compete” with other data centers, which in many aspects will obviously win.
Data centers constantly fill their machine rooms with modern equipment. IaaS-provider, whose main purpose - the provision of cloud services, sooner or later fall behind. In addition, not every supplier with its own data center can be certified to the Tier level, which is a significant indicator of the reliability of the data center.
Moreover, “professional” data centers are investing funds and expertise in enhancing perimeter security. For example, the building of the Xelent data center in which IT-GRAD equipment is located has a multi-level security system with several perimeters. Entry and access to the territory is carried out strictly by passes - if the employee has forgotten his house ID, he will not be allowed into the workplace.
The facility has an armed guard, and the whole territory is covered with video surveillance cameras and motion sensors. The metal fence is equipped with vibration sensors.
/ photo Xelent / IT-GRAD cloud platform based on Xelent data center
Part of the requirements of the standards is carried out "automatically"
For example, it concerns the standard PCI DSS. It is intended to unify the requirements of international payment systems to ensure the security of cardholder data. And bringing the infrastructure in line with the requirements of PCI DSS is a time consuming task.
The standard spelled out the safety standards, including the mandatory presence of video surveillance systems and the introduction of employee identification processes. The cloud provider receives these services "automatically", placing the "hardware" in the finished data center. For example, all IT-GRAD equipment is located in lockable racks under the supervision of cameras 24/7.
This allows the IaaS provider to focus on developing its own services and expand the range of services provided.
More options for customers
By outsourcing the management of the data center, the cloud provider gains access to a large number of telecommunication providers. This allows his clients to choose the best connection option for business and protect the infrastructure from equipment failures (due to the use of spare capacity). Additionally, the possibility of choosing an Internet provider creates a competitive environment that gives customers the opportunity to choose the most profitable connection packages for their high-performance services.
For example, in the DataSpace data center, where the IT-GRAD facilities are also located, come the trunk lines of 39 major telecom operators - these are Rostelecom, Gars Telecom, AKADO Telecom, Beeline, Megafon, MTS other. The building also has independent telecommunication inputs from two different MGTS wells at a distance of more than 100 meters from each other and six 100-millimeter cable channels leading to the main and additional data entry points (MPoE and SPoE).
You can concentrate on services.
Maintenance of the data center requires the availability of qualified personnel and time. It is necessary to monitor the cooling systems, power sources, safety.
According to Phil Collerton (Phil Collerton), the former head of operations at Interxion, there are a lot of "mobile components" in the data center that have to be taken into account. Everything from security systems to redundancy should work to the maximum capacity so that the provider can provide its services. As Phil says, data center service is a full-time job.
One person responsible for the performance of uninterruptible power supplies most likely does not have the same level of expertise in maintaining backup systems for data stored in the cloud. The creation of "universal soldiers" can tell on the quality of the services of the provider. Therefore, it is good if the maintenance tasks for hardware and software fall on the shoulders of different organizations.
A scenario in which the data center expands the list of services and begins to provide IaaS services is possible. However, it will not become a full-fledged cloud provider, since its main activity will be focused on equipment placement (co-location), data center lifecycle management and the provision of telecommunications services.
In this case, cloud services remain only an “expander” of the data center product range and constitute approximately 5–10% of the total number of services. And most often this happens due to the involvement of partners in the face of IaaS-providers, who consider the infrastructure of the data center as a reliable platform for placement.
PS Another couple of materials from the First Corporate IaaS blog:
Source: https://habr.com/ru/post/341570/
All Articles