How performance tracking and billing can improve your OpenStack cloud performance
Posted by: John Meadows, Talligent
[Editor's Note: Below is the guest post of John Meadows from Talligent, the developer of the OpenBook billing solution for the OpenStack platform. Want to post a guest post? Contact Nick Chase .]
This post describes best practices for improving the use of the OpenStack cloud. The performance of an OpenStack deployment can be limited for the following reasons: 1) lack of control and automation, leading to inefficient use of resources; and 2) poor visual accessibility of key indicators, leading to planning difficulties. Self-service automation, reporting and billing are three basic steps to increase productivity and manage a competitive OpenStack cloud.
')
Providing on-demand services based on OpenStack is a clever trade-off. On the one hand, customers expect instant and automatic provision of IT services (given their experience with AWS, etc.). On the other hand, IT managers must maintain high performance to compete with public cloud offerings. Increasing the staff of OpenStack specialists who would respond to user requests manually increases the total cost of the cloud, while at the same time reducing competitiveness and responsiveness to requests. High performance and required flexibility are inverse correlated. In an effective cloud solution, a balance must be struck between resource intensive use and customer needs for prompt response and ensuring the necessary performance.
As they say, if you cannot measure something, you cannot control it. OpenStack has its own project for making measurements called OpenStack Telemetry (Ceilometer), but it has limited built-in capabilities for reporting on cloud usage, billing / real cost accounting by chargeback (calculating the cost of IT services provided by the department to determine their cost and accounting consumption units of the organization), the preservation of the history of records and financial planning. IT administrators are otherwise forced to rely on a disaggregated set of data points, queries, spreadsheets, and resource usage reports at a specific point in time.
So, what processes and controls are needed to run an effective cloud on demand? Our clients have been successful with the following common steps:
1. Automate customer self-service.
2. Planning reporting on key performance indicators.
3. Implementation of the corresponding billing (or, in the case of organizations, charge-based costs or showback-cost tracking).
Automation of self-service eliminates the need for recruiting staff to manually process user requests. The standard implementation of the self-service system includes, at a minimum, a client portal where you can see accruals, levels, quotas and resources provided; integration with Keystone for user administration at the project level; as well as a workflow for routing requests for quotas for approval by the manager (if necessary).
With self-service client support and automatic provisioning of resources, the OpenStack cloud operation is similar to the work of the public cloud. Clients' expectations are justified, responsiveness increases, SLA indicators improve, fewer staff are required. Efficiency will also increase, because users will better understand what they are charged for, and what leverage they need to be strengthened or weakened in accordance with their project requirements and available budget. Moreover, quotas can be limited if users can more easily adapt to changing project needs.
Who are your biggest tenants? What types of workload are used? Is your cloud SLA compliant? KPI reporting plays a key role for managers at all levels to make informed decisions about the development of the cloud, for example, how and when to add resources, create new services, change prices, and move to a new level of workload. The OpenStack Ceilometer project provides short-term collection of detailed measurement data, but is not a system for long-term storage of record history and does not provide visual accessibility of reports for making such types of decisions.
Detailed reporting on tenant workload can help managers better balance this workload and carry out development planning. Certain applications or workloads can make more economic sense when deployed to a third-party cloud. The acquisition of additional capacity for rare seasonal workload can be costly and lead to an overall decrease in the efficiency of use. Computing nodes, which are added at peak load and can be used only 5% of the time, are your most expensive cloud resource, because they are not depreciated when working at peak load.
A public cloud requires a reliable billing solution that ensures a return on your cloud investment. Chargeback is a mechanism for distributing private cloud resources. The automatic provision of resources without billing or chargeback can result in a demand that quickly exceeds available resources and leads to the tragedy of shared resources . Free, easily provided resources can quickly be exhausted to such an extent that IT managers will either have to purchase additional hardware, the cost of which will go far beyond the allocated budget, or accept the poor quality of service.
The billing solution must be flexible, allowing service providers to quickly adapt to competitors' changing proposals and customer requirements. Customer needs are changing at a tremendous rate; You need a solution that does not tie you to a specific configuration and does not limit the types of chargeable items. Key questions to answer: How do you want to promote new cloud offerings? Will discounts, special prices and the possibility of savings due to prepayment be provided? Will you offer cloud solutions in different currencies, or will you have to keep track of local tax rates?
For a public cloud, using the chargeback method can be strategically difficult, because obtaining the consent of the management of the company, as well as accurate tracking of indicators of individual projects, structural units or geographical locations. A softer approach is to start using a showback report to determine the level of resource consumption by various tenants and projects. IT customers will begin to understand their impact on the environment, and management will have time to prepare budgets for real cost accounting in the future using the chargeback method.
Customers should be able to proactively manage their resources so as not to get overspending at the end of the month. To do this, they need, among other things, a mechanism that would allow to view the accrued expenses for an incomplete billing cycle, notify when thresholds have been reached, establish clear quotas and restrictions, and also make it possible to adjust resource requirements accordingly.
All this requires software that provides visual accessibility and control capabilities necessary to maintain high operational efficiency, as well as billing flexibility to meet the ever-changing customer requirements and market conditions.
John Meadows is an employee of the Business Development Division at Talligent , the developer of the OpenBook billing solution for the OpenStack platform.
OpenBook makes it easy to create detailed pricing plans for various types of tenants, usage models and levels of cloud solutions resellers. Customers can examine in detail their consumption figures and make the appropriate changes to IT resources (within the framework of their current conditions). The OpenBook solution is integrated with the Horizon and Keystone components of the OpenStack platform to support user self-service when managing cloud resources.
Original article in English .
[Editor's Note: Below is the guest post of John Meadows from Talligent, the developer of the OpenBook billing solution for the OpenStack platform. Want to post a guest post? Contact Nick Chase .]
This post describes best practices for improving the use of the OpenStack cloud. The performance of an OpenStack deployment can be limited for the following reasons: 1) lack of control and automation, leading to inefficient use of resources; and 2) poor visual accessibility of key indicators, leading to planning difficulties. Self-service automation, reporting and billing are three basic steps to increase productivity and manage a competitive OpenStack cloud.
')
If you can't measure something, you can't control it.
Providing on-demand services based on OpenStack is a clever trade-off. On the one hand, customers expect instant and automatic provision of IT services (given their experience with AWS, etc.). On the other hand, IT managers must maintain high performance to compete with public cloud offerings. Increasing the staff of OpenStack specialists who would respond to user requests manually increases the total cost of the cloud, while at the same time reducing competitiveness and responsiveness to requests. High performance and required flexibility are inverse correlated. In an effective cloud solution, a balance must be struck between resource intensive use and customer needs for prompt response and ensuring the necessary performance.
As they say, if you cannot measure something, you cannot control it. OpenStack has its own project for making measurements called OpenStack Telemetry (Ceilometer), but it has limited built-in capabilities for reporting on cloud usage, billing / real cost accounting by chargeback (calculating the cost of IT services provided by the department to determine their cost and accounting consumption units of the organization), the preservation of the history of records and financial planning. IT administrators are otherwise forced to rely on a disaggregated set of data points, queries, spreadsheets, and resource usage reports at a specific point in time.
Path to the cloud ready for work
So, what processes and controls are needed to run an effective cloud on demand? Our clients have been successful with the following common steps:
1. Automate customer self-service.
2. Planning reporting on key performance indicators.
3. Implementation of the corresponding billing (or, in the case of organizations, charge-based costs or showback-cost tracking).
Automate customer self service
Automation of self-service eliminates the need for recruiting staff to manually process user requests. The standard implementation of the self-service system includes, at a minimum, a client portal where you can see accruals, levels, quotas and resources provided; integration with Keystone for user administration at the project level; as well as a workflow for routing requests for quotas for approval by the manager (if necessary).
With self-service client support and automatic provisioning of resources, the OpenStack cloud operation is similar to the work of the public cloud. Clients' expectations are justified, responsiveness increases, SLA indicators improve, fewer staff are required. Efficiency will also increase, because users will better understand what they are charged for, and what leverage they need to be strengthened or weakened in accordance with their project requirements and available budget. Moreover, quotas can be limited if users can more easily adapt to changing project needs.
Reporting Key Performance Indicators
Who are your biggest tenants? What types of workload are used? Is your cloud SLA compliant? KPI reporting plays a key role for managers at all levels to make informed decisions about the development of the cloud, for example, how and when to add resources, create new services, change prices, and move to a new level of workload. The OpenStack Ceilometer project provides short-term collection of detailed measurement data, but is not a system for long-term storage of record history and does not provide visual accessibility of reports for making such types of decisions.
Detailed reporting on tenant workload can help managers better balance this workload and carry out development planning. Certain applications or workloads can make more economic sense when deployed to a third-party cloud. The acquisition of additional capacity for rare seasonal workload can be costly and lead to an overall decrease in the efficiency of use. Computing nodes, which are added at peak load and can be used only 5% of the time, are your most expensive cloud resource, because they are not depreciated when working at peak load.
Relevant billing
A public cloud requires a reliable billing solution that ensures a return on your cloud investment. Chargeback is a mechanism for distributing private cloud resources. The automatic provision of resources without billing or chargeback can result in a demand that quickly exceeds available resources and leads to the tragedy of shared resources . Free, easily provided resources can quickly be exhausted to such an extent that IT managers will either have to purchase additional hardware, the cost of which will go far beyond the allocated budget, or accept the poor quality of service.
The billing solution must be flexible, allowing service providers to quickly adapt to competitors' changing proposals and customer requirements. Customer needs are changing at a tremendous rate; You need a solution that does not tie you to a specific configuration and does not limit the types of chargeable items. Key questions to answer: How do you want to promote new cloud offerings? Will discounts, special prices and the possibility of savings due to prepayment be provided? Will you offer cloud solutions in different currencies, or will you have to keep track of local tax rates?
For a public cloud, using the chargeback method can be strategically difficult, because obtaining the consent of the management of the company, as well as accurate tracking of indicators of individual projects, structural units or geographical locations. A softer approach is to start using a showback report to determine the level of resource consumption by various tenants and projects. IT customers will begin to understand their impact on the environment, and management will have time to prepare budgets for real cost accounting in the future using the chargeback method.
Final goal
Customers should be able to proactively manage their resources so as not to get overspending at the end of the month. To do this, they need, among other things, a mechanism that would allow to view the accrued expenses for an incomplete billing cycle, notify when thresholds have been reached, establish clear quotas and restrictions, and also make it possible to adjust resource requirements accordingly.
All this requires software that provides visual accessibility and control capabilities necessary to maintain high operational efficiency, as well as billing flexibility to meet the ever-changing customer requirements and market conditions.
John Meadows is an employee of the Business Development Division at Talligent , the developer of the OpenBook billing solution for the OpenStack platform.
OpenBook makes it easy to create detailed pricing plans for various types of tenants, usage models and levels of cloud solutions resellers. Customers can examine in detail their consumption figures and make the appropriate changes to IT resources (within the framework of their current conditions). The OpenBook solution is integrated with the Horizon and Keystone components of the OpenStack platform to support user self-service when managing cloud resources.
Original article in English .
Source: https://habr.com/ru/post/246015/
All Articles