Monitoring the Hybrid Cloud Ecosystem
Hybrid clouds give you the best of the public and private worlds, but they pose challenges for monitoring that need to solved from several perspectives.
Join the DZone community and get the full member experience.Join For Free
Monitoring plays a key role in cloud management, not only for service providers but also service consumers. All the cloud offerings (IAAS, PAAS, and SAAS) by any service provider are nothing but “metered services” that allow customers to pay proportionately to measured parameters. The entire revenue and monetary aspects of cloud computing like workload assessment, pricing, and billing are completely reliant on resource monitoring. Also, features like performance, security, SLA, and data center administration is entirely governed by metrics monitoring and measurement.
Monitoring cloud environments can be broadly classified into the following viewpoints:
- Service provider viewpoint
- Service consumer viewpoint
It is a typical subscriber provider model where the provider offers the consumer ready services and the customer subscribes and pays for what they consume.
A cloud service provider (private/public) would be more focused on the following:
- Usability: Ease of usability of the service by the consumer.
- Reliability: Capability of the service offered to perform over time.
- Safety: The ability of the offered service to operate in a secure and reliable manner.
- Performance: The timing characteristics of the service offered.
- Capacity: Physical characteristics (weight, size, RAM, hard disk, etc.) of the system, majorly related to the hardware requirements and other physical network conformations.
- Maintainability and extensibility: The operational efficiency to maintain service for correcting defects and new enhancements.
- Availability: The operational time of the service offered.
- Data center administration and governance.
- Regulatory, service level agreements, and compliance.
- Metering and billing.
A cloud service consumer would be more focused on the following:
- Quality of service.
- Services offered and the subscription plans.
- Allotted and utilized quota.
- Usage and consumption.
- Price and chargeback.
However, monitoring hybrid ecosystems encompassing various cloud interoperable models and federations has to be highly dynamic, agile and adapt to the intermittently changing environments. Collections of metrics from a single service provider will no longer be relevant in a hybrid cloud ecosystem. Accurate, seamless and well-established monitoring activities are essential to efficiently operate hybrid cloud platforms and administer their increasingly complex requirements. For an accurate, reliable monitoring capability, there needs to be a clear-cut consideration of how the monitoring will perform when the information is exchanged across federated cloud services. When the workloads are drifted to another cloud platform, the mechanism to distribute key statistics, data, and indicators will need to cross cloud boundaries. It is important that the frontlines and formats between clouds need to be normalized for a continuous and seamless operation of federated monitoring in heterogeneous environments.
Federated cloud monitoring also brings up another interesting issue. Migration of workloads from one site to another will make the originating platform lose the direct control over the execution environment, but nevertheless, the monitoring statistics have to be collected by the service assurance team. This augments the need of a coordination agent on each remote site that would act on the behalf of execution environments.
Some of the notable viewpoints of different stakeholders in a typical enterprise scenario using a hybrid cloud ecosystem are depicted below:
In this scenario, IT, legal, regulatory, and policy administration need to be exercised in a collective mode, and the respective vendor (internal/external) operations and support need to function as per the predefined agreements among all the parties. Also, the metering, billing, and chargeback methodologies need to be well-formulated. Information on metering and billing procedures would assist the service administrators to determine accurate returns from the service consumers. The monetary statistics would assist the portfolio owners or investment officers to validate disbursements, forecast budgets, and finalize the internal chargebacks within the enterprise service lines and departments. End users would be interested from a consumption and pricing point of view and are least concerned with the internal movements, execution environments, and platforms. The performance of information exchange gateways and monitoring process in rapid changing execution environments would become the key to delivering the anticipated results for the end consumers.
A much more composite and futuristic model might involve the union of various service brokers as well as providers and enterprise clouds, bringing in more dimensions to the same viewpoints from the perspectives of infrastructure, subscription, and policy administration.
Whatever the approach and the type of cloud federation, it is the combined effort of all the stakeholders or parties in the hybrid cloud ecosystem to ensure quality services to the customer. Subsequently, the unified platform to manage, govern, and monitor various clouds is gaining momentum.
Opinions expressed by DZone contributors are their own.