This content is part of the Essential Guide: AWS re:Invent 2015: A guide to Amazon's sold-out event

Cloud performance monitoring maximizes AWS efficiency

Whether a startup or an established enterprise, cloud costs must be closely monitored and managed in the cloud. This includes scaling unused computing power.

Cloud adoption isn't a marital commitment; it's a strategic business decision. Many new businesses lack the staffing and capital to deploy a production-quality data center, and moving directly to the public cloud may seem like an appealing choice. Amazon cloud users can manage AWS resources with cloud performance monitoring to get the most value for their IT investment.

Public cloud helps startups jump-start operations where workloads are sporadic or where computing demands change rapidly. But companies also need to understand the practical benefits and limitations of public cloud providers like Amazon Web Services (AWS).

Know when to scale down

Cloud is billed monthly, and there are computing situations in which a public cloud can be more expensive than hosting your own servers, storage and other data center infrastructure. Startups need to watch costs with cloud performance monitoring, weigh the benefits on an ongoing basis and look for stabilization in workload demands. It's important to understand when it makes good business sense to bring some workloads in house where dedicated IT staff can exercise better control and cost-effectiveness.

Public cloud is not an all-or-nothing decision. Savvy businesses mix and match public cloud resources with private cloud infrastructures to achieve a flexible hybrid environment that offers a continuous spectrum of cost/benefit tradeoffs for the business.

For example, one workload might run well in house until a spike in demand -- perhaps several days each month -- causes resource shortages. These applications could be moved to the public cloud where additional AWS resources are rented for the short term and then returned to the local data center once the demand abates; this tactic is sometimes called cloud bursting. As another example, the public cloud might be a perfect choice to run occasional big data projects where vast computing potential can crunch enormous data volumes -- then release the computing resources to save money.

Know how to manage AWS performance

The public cloud costs money for each gigabyte of storage and every computing instance invoked. That's fine when someone is actively practicing cloud performance monitoring, but the very nature of public cloud computing means that anyone with access to a company cloud account can spin up more AWS resources in a matter of seconds. If those resources aren't released or throttled back again, the company winds up paying for cloud computing that might not be doing any useful work, and the monthly costs for public cloud computing can grow unwieldy.

Consider a simple software development project. A developer might establish an AWS Elastic Compute Cloud (EC2) instance to test a new build. After identifying some bugs and performance issues, he deploys the new build on a different EC2 cluster. If developers continue to launch new workloads on new compute instances -- and old instances aren't killed or deleted -- the test and dev department can find itself facing a bill for far more computing resources than might otherwise be necessary to complete the development cycle.

CIOs and business leaders can't just assume business units will use the public cloud efficiently and cost-effectively. Startups seeking to adopt the public cloud should implement policies, processes and tools to manage, contain and report public cloud computing costs. Public cloud providers like AWS supply comprehensive cost tracking and reporting.

Track usage, spending to optimize cost

AWS provides frequent detailed billing reports. Reports can distinguish costs by time (hours, days, months), by account (such as test and dev), by AWS product (such as costs in Amazon S3) or even by custom tags. Reports are configured and accessed through the AWS Management Console. Assigning a business or IT manager to review these reports regularly can be an important first step in public cloud cost management. Sharing those reports with group leaders or budget owners can also help to drive cost-reduction discussions.

There are also varied third-party tools such as Cloudyn, Cloudability, CloudCheckr and others to help with cloud performance monitoring, such as tracking service usage, optimizing costs and promoting accountability within the organization. Managed service providers like Connectria, mindSHIFT Technologies Inc. and Datapipe Inc. (Cloud Analytics for AWS) and others can help independently manage AWS resources and architecture to assist with service configurations, management and cost monitoring.

Next Steps

Elevate to the cloud with AWS startup program

Effectively scaling AWS applications

Buy or DIY? Get the right performance monitoring service

Discover four AWS-friendly app performance management tools

Dig Deeper on AWS case studies and startups