Red Hat ブログ
When it comes to delivering cloud services, enterprise architects have a common request to create a public cloud-type rate plan for showback, chargeback, or billing. Public cloud packaging is fairly standardized across the big vendors as innovations are quickly copied by others and basic virtual machines are assessed mainly on price. (I touched on the concept of the ongoing price changes and commoditization of public clouds in an earlier post.) Because of this standardization and relative pervasiveness, public cloud rate plans are well understood by cloud consumers. This makes them a good model for introducing enterprise users to new cloud services built on OpenStack.Enterprise architects are also highly interested in on-demand, self-service functionality from their Openstack clouds in order to imitate the immediate response of public clouds. We will cover how to deliver on-demand cloud services in a future post.
Pricing and Packaging Cloud Services
Public cloud rate plans are very popular, seeing adoption within enterprises, private hosted clouds, and newer public cloud providers alike. Most public cloud providers use the typical public cloud rate plan as a foundation for layering on services, software, security, and intangibles like reputation to build up differentiated offerings.Enterprise cloud architects use similar rate plans to demonstrate to internal customers that they can provide on-demand, self-service cloud services at a competitive price. To manage internal expectations and encourage good behavior, enterprises usually introduce cloud pricing via a showback model which does not directly impact budgets or require exchange of money. Users learn cloud cost structures and the impact of their resource usage. Later, full chargeback can be applied where internal users are expected to pay for services provided.
As evidenced by easily accessible published rate plans, on-demand compute instances can represent a wide range of sizes, locations, operating systems, and optimizations (memory, storage, compute). Reserved instances provide the user the opportunity to make a one-time upfront payment in exchange for a discount on the hourly charge for the instance over the course of one or three years. Spot instances provide a discount in exchange for flexibility of running the instance at a time when the cloud has unused compute capacity.
Red Hat Enterprise Linux OpenStack Platform and Talligent OpenBook
Red Hat Enterprise Linux OpenStack Platform provides an integrated set of OpenStack modules and APIs that Talligent OpenBook connects to. OpenBook leverages the Keystone and Horizon modules for customer authentication and self-service. OpenBook also directly connects to other OpenStack components such as Nova, Swift, and Cinder to create an initial list of cloud tenants and assigned resources.
The key attributes of the rate plan are available from OpenStack via the Ceilometer metering component. Ceilometer data is non-repudiable and therefore auditable for billing of cloud services. OpenBook captures a full set of configuration details and usage metrics from OpenStack via the Ceilometer module. This includes all the meters associated with Nova (Compute), Neutron (Network), Glance (Image), Cinder (Volume), Swift (Object Storage), and Kwapi (Energy). A detailed list of the meters is available here.
As new metrics are added to Ceilometer, those metrics are picked up by the resource manager in OpenBook and made available as billable elements included in rate plans. For example, the Juno release of OpenStack is expected to add key SDN metrics such as Load Balancer as a Service, Firewall as a Service, and VPN as a Service. We plan to make those SDN metrics available in OpenBook within a few weeks of the GA release of the Juno version of OpenStack.
Creating an Public Cloud Rate Plan in OpenStack
Using a solution like OpenBook from Talligent, cloud architects can create a rate plan for Infrastructure-as-a-Service similar to any of the large public cloud providers. In order to approximate typical public cloud pricing in OpenStack, here are the key elements of the rate plan:
- Server images in Openstack correspond to on-demand instances. Generally, they are categorized according to size (small, medium, large, extra large, etc.) and have vCPU, memory, and instance storage configured accordingly.
- While OpenBook can prorate charges for fractional billing periods, most public cloud providers rounds up all fractions to the next hourly interval when computing charges.
- OpenStack includes the concept of geographical regions. The region attribute enables the cloud architect to charge different rates for the different regions. This might be done because of variations in cost structures, data center capabilities, currencies, taxes, or other location specific variables.
- Once the image, billing period, and regions have been established, users can designate a unit charge for each server image. Based on a recent survey of public cloud prices, prices range from $0.013 per hour for a tiny instance to $0.28 per hour for an extra large instance.
- Block storage can be billed by volume type (SSD, SATA,..) and by GB hour for volumes and GB month for images and snapshots. GB hour and GB month calculations can be based on either peak or average hourly size.
- Object storage can be billed by storage object size, container size, outgoing bytes transferred, and number of API requests.
That covers the basics. Advanced users can develop services that take into account dedicated connections between the customer and cloud, GPUs, DNS, VPN Connection, high i/o instances, high storage instances, high performance computing, etc. Public cloud rates change regularly as the market grows and competition increases. Instance types might not correspond directly to instances configured within OpenStack so please note that these rates will be an approximation only. Talligent.com for more information
A Business-Ready CloudTM requires business processes and controls. Once your OpenStack cloud is deployed, you need the visibility and control to maintain high operational efficiency, as well as the billing flexibility to keep up with ever-changing customer requirements and market conditions. Self-service capabilities from Talligent support the high response on-demand cloud functionality that customers expect. Public and private cloud offerings and prices are evolving rapidly – Talligent supplies the tools to evolve your cloud and compete.