"SLA in the cloud": what to look for
Cloud service providers provide various guarantees, among which there is the availability of services and resources that are specified in the service level agreement. Most often, this agreement is denoted by the abbreviation SLA, and in this article we will talk about the important nuances that should be spelled out in the SLA IaaS provider ( example SLA 1cloud ).
/ Flickr / Dennis Skley / CC
An increasing number of applications and the amount of computational tasks are moving to IaaS. Therefore, IaaS-providers try to guarantee customers a certain level of security and comfort and adapt to changing market requirements, as well as the laws of the countries in which they provide services. All this leads to changes in the SLA.
')
“Some aspects of the SLA of cloud providers today are outdated or, moreover, cause confusion. For example, there are suppliers who prohibit the dissemination of information about infrastructure failures in their SLAs, says cloud services specialist David Linthicum. “That does not help either the provider itself or its customers, since it is difficult to call such conditions honest.”
In an attempt to standardize and provide recommendations on drafting a service level agreement, the European Commission began promoting a document that aims to provide more transparency for businesses when working with cloud services and studying SLAs. The initiative was named SLALOM and, as CIF notes, it will allow “reducing uncertainties” when migrating infrastructure to the cloud.
The new document contains suggestions and practices from influential cloud market players. “The team has compiled a list of cloud computing terms that cover all aspects of the relationship between the provider and the client,” says Oliver Barreto of Atos.
The SLALOM document itself can be found at the link , and then we will examine a few priority points that the provider should pay attention to when drafting their service level agreement.
Availability level
In the SLA of the cloud provider, key metrics should be defined that describe the performance of the cloud service. We are talking about the availability of the service, the number of users able to access the service at the same time, as well as the time needed to process user transactions. For example, an IaaS provider can guarantee the availability of applications 99.5% of the time.
You should also designate a time frame within which this level of accessibility is ensured so that customers do not have questions. If we continue the example, the provider can ensure the availability of applications up to 99.5% of the time from 8 am to 8 pm on weekdays. This parameter has its own name - the agreed service uptime, or SVR.
However, we note that when choosing a provider, it is not necessary to pay attention only to “interest”, you should also take into account how ready the provider is to solve the problems and difficulties that arise. For example, we at 1cloud offer a backup that runs once a day to protect user data from loss in case of emergency situations. Copying is performed on a remote disk array, which makes data available for recovery even if the server file system is unavailable (and in the case of irreversible actions by the user).
Exceptions
In SLA, you must prescribe acceptable conditions under which the provider does not guarantee the specified availability and performance. Since this section exempts the supplier from performing its duties, the list of restrictions should be strict and short.
For example, limited availability during scheduled technical work and software updates, as well as in case of force majeure, in particular, failure of third-party equipment. Another option is software failure due to customer’s violation of the recommendations specified in the product documentation.
Also in the SLA should be noted guarantees the availability of the management interface of the virtual infrastructure. This will help to avoid confusion and other incidents when the client needed to urgently increase power consumption, and the self-service console was unavailable.
Monitoring
Write in the agreement all that concerns the security and reliability of data storage. It is about reliability and confidentiality of information, as well as data integrity. Determine the rights of access to information, and the circle of persons responsible in case of its damage.
Provide information about disaster recovery tools, and discuss the procedure for reporting failures and equipment failure: format and timeframe. Indicate what measures will be taken to restore the functioning of the systems.
Compensation
The provider must determine the range of compensations that he pays in case of violation of accessibility indicators and other quality parameters. Responsibility of providers should be set at a rate not less than 100% of the payment for the reporting period with a long idle. For example, 1cloud guarantees a 100% refund of the cost of a service, with a network availability of less than 76.98%.
Calculations
Many customers are faced with the question of correct execution of calculations. In order for the client to be calm and understand “what awaits him”, his calculations should not diverge from the values ​​suggested by the supplier. A good option would be to use detailed formulas. We also recommend giving illustrative calculation examples. This will help the IT department of the client company to establish the amount of necessary hardware and capacity, as well as to determine the software for the organization.
It is important to remember that service level agreements for cloud services are different from other SLAs. Unscrupulous suppliers sometimes prefer to build a one-to-many relationship, implying the use of a single SLA and offering it to all customers. This approach is good for suppliers, but it is completely unsuitable for customers with specific tasks.
PS Our digest with 25 materials about security, the work of programmers and the experience of creating an IaaS provider can be found here .
Additional materials from the blog 1cloud :
/ Flickr / Dennis Skley / CC
An increasing number of applications and the amount of computational tasks are moving to IaaS. Therefore, IaaS-providers try to guarantee customers a certain level of security and comfort and adapt to changing market requirements, as well as the laws of the countries in which they provide services. All this leads to changes in the SLA.
')
“Some aspects of the SLA of cloud providers today are outdated or, moreover, cause confusion. For example, there are suppliers who prohibit the dissemination of information about infrastructure failures in their SLAs, says cloud services specialist David Linthicum. “That does not help either the provider itself or its customers, since it is difficult to call such conditions honest.”
In an attempt to standardize and provide recommendations on drafting a service level agreement, the European Commission began promoting a document that aims to provide more transparency for businesses when working with cloud services and studying SLAs. The initiative was named SLALOM and, as CIF notes, it will allow “reducing uncertainties” when migrating infrastructure to the cloud.
The new document contains suggestions and practices from influential cloud market players. “The team has compiled a list of cloud computing terms that cover all aspects of the relationship between the provider and the client,” says Oliver Barreto of Atos.
The SLALOM document itself can be found at the link , and then we will examine a few priority points that the provider should pay attention to when drafting their service level agreement.
Availability level
In the SLA of the cloud provider, key metrics should be defined that describe the performance of the cloud service. We are talking about the availability of the service, the number of users able to access the service at the same time, as well as the time needed to process user transactions. For example, an IaaS provider can guarantee the availability of applications 99.5% of the time.
You should also designate a time frame within which this level of accessibility is ensured so that customers do not have questions. If we continue the example, the provider can ensure the availability of applications up to 99.5% of the time from 8 am to 8 pm on weekdays. This parameter has its own name - the agreed service uptime, or SVR.
However, we note that when choosing a provider, it is not necessary to pay attention only to “interest”, you should also take into account how ready the provider is to solve the problems and difficulties that arise. For example, we at 1cloud offer a backup that runs once a day to protect user data from loss in case of emergency situations. Copying is performed on a remote disk array, which makes data available for recovery even if the server file system is unavailable (and in the case of irreversible actions by the user).
Exceptions
In SLA, you must prescribe acceptable conditions under which the provider does not guarantee the specified availability and performance. Since this section exempts the supplier from performing its duties, the list of restrictions should be strict and short.
For example, limited availability during scheduled technical work and software updates, as well as in case of force majeure, in particular, failure of third-party equipment. Another option is software failure due to customer’s violation of the recommendations specified in the product documentation.
Also in the SLA should be noted guarantees the availability of the management interface of the virtual infrastructure. This will help to avoid confusion and other incidents when the client needed to urgently increase power consumption, and the self-service console was unavailable.
Monitoring
Write in the agreement all that concerns the security and reliability of data storage. It is about reliability and confidentiality of information, as well as data integrity. Determine the rights of access to information, and the circle of persons responsible in case of its damage.
Provide information about disaster recovery tools, and discuss the procedure for reporting failures and equipment failure: format and timeframe. Indicate what measures will be taken to restore the functioning of the systems.
Compensation
The provider must determine the range of compensations that he pays in case of violation of accessibility indicators and other quality parameters. Responsibility of providers should be set at a rate not less than 100% of the payment for the reporting period with a long idle. For example, 1cloud guarantees a 100% refund of the cost of a service, with a network availability of less than 76.98%.
Calculations
Many customers are faced with the question of correct execution of calculations. In order for the client to be calm and understand “what awaits him”, his calculations should not diverge from the values ​​suggested by the supplier. A good option would be to use detailed formulas. We also recommend giving illustrative calculation examples. This will help the IT department of the client company to establish the amount of necessary hardware and capacity, as well as to determine the software for the organization.
It is important to remember that service level agreements for cloud services are different from other SLAs. Unscrupulous suppliers sometimes prefer to build a one-to-many relationship, implying the use of a single SLA and offering it to all customers. This approach is good for suppliers, but it is completely unsuitable for customers with specific tasks.
PS Our digest with 25 materials about security, the work of programmers and the experience of creating an IaaS provider can be found here .
Additional materials from the blog 1cloud :
- How we automated work with DNS records in the hosting panel
- Cloud Technology Myths. Part 1: technical support and management
- Cloud Technology Myths. Part 2: Russian hosters and data protection
- Cloud Technology Myths. Part 3: Talking About Iron
- A bit about VPN: A brief overview of software implementations
Source: https://habr.com/ru/post/318432/
All Articles