Gone are the days of businesses owning their own software and hardware and keeping it all on premises in data centers.
And even though cloud computing is not a new way of doing business, technology leaders are finally embracing and harnessing its potential by beginning to move non-business critical applications to the cloud.
This trend has picked up significantly in the last five years, and now, more than ever, the cloud is gaining momentum. At tech conferences, in customer meetings and in boardroom discussions, cloud storage and compute is no longer a conversation for tomorrow. It’s an imperative for today.
But this migration begs the question, “How do you monitor something you no longer own?” Moving resources to the cloud can create significant visibility gaps over your infrastructure performance.
In this whitepaper, we’ll examine use cases for monitoring hybrid cloud infrastructures and best practices for ensuring delivery of applications and services in the cloud.
Best Practice 1
Monitor Cloud and On-Premises Infrastructure from a Single Platform
Businesses have expectations for the performance of their on-premises infrastructure. But now they must utilize that existing infrastructure and the new cloud infrastructure as one. This often presents problems.
However, with a sophisticated monitoring platform, organizations can pull in third-party data – from Amazon Web Services (AWS), Microsoft Azure, IBM Bluemix and others – from the cloud. Just make sure you get a monitoring platform that can marry the metrics collected via the cloud platform’s API with your existing infrastructure metrics collected via SNMP, IP SLA, or other standard protocols.
You’ll want to monitor the following KPIs:
- Network KPIs will inform you about the bytes sent and received on the instance network interface.
- CPU KPIs will monitor estimated CPU utilization for the current billing period at the time of last collection.
- Billing KPIs will tell you the estimated credit balance, credit usages, and charges accrued for the current billing period at the time of last collection.
- System Integrity KPIs will let you know if the system instance, customer instance or system has failed.
- Storage KPIs will give you data on the length and number for Write Operations per collection cycle on an EBS Volume.
It’s important to get a monitoring platform that can take this cloud data and “normalize” it alongside the traditional metric and flow, data the business is accustomed to working with. The right monitoring platform will treat cloud metrics the exact way it treats data from other sources. That means you can monitor, baseline, alert, and report on the data, regardless of the source. It also makes problem correlation much easier.
When your data is uniform across whatever objects and devices you’re monitoring – whether they’re in the datacenter or in the cloud – you have complete visibility into your network and applications.
Use Case: Monitoring AWS
A leading mobile carrier company, until recently, was relying on AWS CloudWatch – a monitoring service for AWS cloud resources and applications that runs on AWS.
Their leaders made the decision to test the cloud by offloading some of their less-critical, internal IT applications to it. In the meantime, the company is maintaining an application in the data center. However, if the cloud proves beneficial, they will rely on the cloud alone for some of these applications.
The company decided to test the cloud for two reasons – cost and ease of implementation. Every time a version of an application changes, the IT department spends a lot of time and energy working on fixes and upgrades that are not core to the business.
Recently, however, the company’s IT department said they have zero visibility into how the applications they are moving to the cloud are behaving. The CloudWatch API works somewhat well to provide insight into what’s happening on the cloud, but the mobile carrier already monitors other devices and doesn’t want to have to check two monitoring platforms to know the health of its applications.
By adding the monitoring platform’s AWS adapter, the mobile carrier company can continue to receive routine monitoring data and metrics for datacenter-hosted applications, along with AWS metrics on the same dashboard.
Best Practice 2
Monitor, Trend and Alert on Cloud Resource Consumption
While leveraging the cloud provides unparalleled access to scalability and agility, it also results in IT losing direct control and visibility. Simply measuring cloud service availability isn’t enough.
Organizations must measure consumption levels (especially KPIs used for billing), as well as the impact of cloud service access on the rest of the infrastructure. They also need to ensure their cloud-based resources are keeping up with what the business needs to evolve.
Billing charges are another unique metric that can be obtained by monitoring AWS. With the AWS adapter, businesses can keep track of the estimated EC2 charges, data transfers and total monthly charges they expect during any given month.
AWS adapter KPIs can inform users of the estimated credit balance for the current billing period at the time of last collection, estimated credit usage for the current billing period at the last time of collection, and estimated charges accrued for the current billing period at the last time of collection.
This can be critical if they require alerting on such metrics to keep costs under control. Also, when a business has planned capacity and has estimated their charges – if their estimates differ from reality – these metrics would provide an early warning.
And since all metric and flow data is treated the same in a sophisticated monitoring platform, businesses can baseline and set alerts to be prepared for any kind of spike or drop in services needed.
In addition to monitoring AWS compute and storage, you should monitor, trend, and alert on your cloud resource consumption to ensure monthly usage remains in line with your budget.
Best Practice 3
Integrate Metrics & Flows Data for a Complete View
When your network lives in a multi-vendor environment, it’s essential that your monitoring platform treat data from different vendors and places equally.
You’re not just getting the cloud metrics, or just the flow data. You need all of this data to work together. The value of having a sophisticated monitoring platform is its ability to turn all of this different data into uniform metrics, which you can baseline, alert and trend on. These uniform metrics can provide a data point about an event, that you can compare with other data points.
Transitioning to the cloud is a big step, and your business needs to have the full data picture – with integrated metrics and flows – to tell the story of your journey from the datacenter to the cloud.
Get a Holistic View of Your Cloud Deployment
Consider this scenario: your organization hosts a document repository application and wants to deploy it to the hybrid cloud. Keeping in mind that an employee may access the application from office headquarters or from a remote location, what components should you monitor to ensure consistent application delivery?
Application deployment: You are hosting the application as an internal wiki solution, but also expose some portions to your customers. You host a local version at your Virginia datacenter, which is accessed by internal employees who work at the local offices.
Datacenter deployment: The Virginia location is your primary datacenter, but you have a full disaster recovery site located in San Francisco. This allows you to provide business continuity in the event of a disaster at the primary site. There is a constant replication of the application occurring from Virginia to San Francisco.
Compute: On the bottom left of our status map, you’ll see the internal Virginia servers. On the bottom right – indicated with clouds – you’ll see the AWS cloud-deployed servers. Status maps allow the user to upload custom images to logically or physically depict the environment to provide a holistic view.
Network: You’ll see the infrastructure that provides users and applications accessibility to one another. Wireless utilizes laptops and other wireless-enabled devices; the physical network infrastructure provides wireless infrastructure connectivity for those using desktops and hardwired solutions.
Internet / Gateway connections: You’ll see the gatekeeper or Internet firewall, and from here, you’ll monitor additional nodes, which are tied to synthetic tests and the first hop of each provider. The links between the hops represent individual interfaces on the firewall that provide the physical connectivity. The last indicator provides data from synthetic tests to the border of the internal application server, as well as the external instances of the application. This allows for troubleshooting issues, and pinpointing where they may be occurring within the environment.
Two alert summaries: Alert summaries are broken up between on-premises equipment and cloud deployments. A sophisticated monitoring platform will aggregate on device instead of showing detail in order to force a user to utilize a monitoring report as part of a troubleshooting workflow. Your monitoring platform should not just be an alerting engine, but a powerful reporting platform.
When monitoring cloud deployments, you need to be able to depend on the data to ensure your business’ application and service delivery. An advanced monitoring platform will treat data from both the datacenter and the cloud in a normalized fashion, supplying your IT department with a single integrated view.
When your monitoring platform can hook into cloud platform performance metrics and end user experience data, you have a complete, end-to-end view of your network. And that’s what you need to successfully run your business as you transition from your on-premises datacenter to the hybrid cloud.