Cloud Monitoring: What you need to know
This post is also available in: Spanish
Migrating their services to the cloud has brought great changes to companies and their IT infrastructure operations, and at a pace which has left decision-makers without time to define a clear post-migration strategy. Meanwhile, SLAs still have to be respected irrespective of if the operation is cloud-based, host-based or housed in the company’s own data center. The problems are the same; fails, performance degradation, security issues, only more complicated.
Many monitoring tools that have managed to keep the pace over the last decade find themselves overwhelmed when it comes to integrating with hybrid environments requiring flexibility, and rapid response times. The shift to the cloud has left many tools struggling to adapt to change.
The present article is an attempt to outline some basic but necessary concepts to keep in mind if your company is thinking of migrating services to the cloud; does your organization require cloud monitoring? what do you need to evaluate?; what is the cloud? And finally, monitoring, our favorite topic, but this time focused on the cloud and monitoring the services you have there. Many clients leave the cloud monitoring part of the operation to the cloud host, not realizing that outsourcing such an essential part of your operation can occasion headaches down the line.
Cloud Monitoring. Before we begin…
- Take it for granted that everyone has different ideas and expectations about what the cloud is and what it can do. The cloud is an ever-changing concept
- Some people have some extreme opinions about the pros and cons of cloud-based tech. Take radical opinions with a pinch of salt
- Avoid simplifications: the cloud is a complicated thing to define, and even more so to implement
- First and foremost is your business, not technical questions. In the final analysis, the cloud is another tool, therefore, why shouldn’t you implement a cloud monitoring?
Ten myths about The Cloud
- The Cloud will always save you money.
- If you’re not in The Cloud, you’re a loser.
- The Cloud should be used for everything.
- Wise CEOs advise going to The Cloud.
- A single strategy and a single provider is all you need to be on The Cloud.
- The Cloud is less safe than on-site premises.
- The Cloud is not for critical services.
- The Cloud is just a big, virtual database.
- Migrating to The Cloud means a wholesale move, no half-measures.
- Virtualization is the same as having a private cloud service.
The 5 most common mistakes when migrating to the cloud
Forgetting how important the connectivity between the cloud and other infrastructures is. Not measuring the connectivity capacity or monitoring the SLA of those connections can lead to serious service shortfalls.
Thinking that the cloud works by itself. You’ll contract the infrastructure but not the parallel services, making it all too easy to lose oversight of your operation and limit your service.
Not having a crisis-plan in place. How will an unexpected service fail on the part of our provider affect our operation? Does being on The Cloud worsen the impact of outages? The easy promise of The Cloud lulls many into forgetting the need to consider crisis drills, and to have a plan in place. The Cloud carries connotations of omnipotence; it’s not. Even the almighty Amazon has suffered cloud computing failure and downtime. Are you more prepared than Amazon?
Assuming your provider is impenetrable. Don’t do it. If your provider has a security issue, it’s a problem for you, possibly even with legal ramifications if your own clients are affected. Were you this exposed before you were on The Cloud?
Thinking that virtuality and real-world tech give the same performance. They don’t. Not in terms of disc access or CPU capacity. In crisis situations it’s better to have hands-on access to hardware. Don’t believe otherwise.
Thinking that change is cheap. Your TCOs are still there, they’re just hidden in costs like training, and the experiential costs you pay for migrating to a complex service, made up of different elements, and which can end up converting your operation into something quite different.
Are you ware of the relevance of a good cloud monitoring after reading those common mistakes? Then, keep reading.
Delegating your monitoring to The Cloud
If you think moving your infrastructure to the cloud will mean you won’t have to maintain or monitor it, too bad. Your infrastructure might not be a worry anymore but your customers are going to expect the same level of service, or more, now that they know you’re in The Cloud. Monitoring your service, your customer experience, your process through flows, are all more important than ever, and your cloud monitoring tool should still be doing its customary tasks of checking all those things, and locating where any problems might be occurring even though your infrastructure is cloud-based. Since no-one knows your business as well as you do, no-one can offer you the service you want out of the box. You still have to configure the metrics for the elements and components you want to monitor.
Cloud-specific monitoring tools
Amazon, Microsoft, OpenStack, Google, CloudStack-all the big players-offer their own services, including monitoring. All very well for checking on their service provision, but will their tools be as adapted to your business as you would like?
Given the diversity of information inputs it would be nice to have a tool that can adapt to what’s already out there and what’s still to come. Many manufacturers offer plugins in order to integrate with distinct platforms, but that is short-term thinking, as each platform amplifies its APIs and imposes restrictions and/or conditions as and when they please. To take an example, the majority of manufacturers offer metrics which don’t stray too far from the infrastructure itself, EC2, VPC, and S3 in AWS, instances, drives, Azure sites, machine status, images, storage (OpenStack, VMware). These infrastructure metrics are the same as the ones you get from installing agents on your machines, meaning they don’t need integrating, they can just be left to get on with their usual tasks, as always.
Most organizations have ever more fragmented environments, both in technologies and distribution. If the plan is to integrate monitoring among your private infrastructure, private internal cloud and private outsourced cloud, private cloud hosting, public cloud and SaaS you’re going to need a plethora of connectors, hours of your technician’s time and it’s never going to stop needing retuning due to constant reconfiguration of parameters. Keeping it simple, with your systems under the oversight of a single tool is imperative to avoid oversegmenting your infrastructure. In the case of oversegmentation it’s advisable to take the following into consideration.
Infrastructure monitoring vs service monitoring
Wouldn’t it be more efficient to monitor the services which all these infrastructures employ? Independently of the technology deployed, and covering the most important area (the service), down to the specific nuts and bolts (the infrastructure components supporting the service).
Despite Pandora FMS being able to integrate with the main cloud and virtualization tech providers (Amazon, Vmware), we don’t recommend this approach to our clients. Our transaction monitoring tool, combined with the deployment of remote probes and classic agents allows us to offer more adaptive monitoring which is also less dependent on the specific technology you’re using to do business.
Pros of Pandora FMS in hybrid environments
- Doesn’t depend on cloud infrastructure or tech
- Oriented to measuring the final service
- Customizable service reports
- Speedy implementation
What do you think? Has this article been helpful? If you have an opinion about cloud monitoring we’re interested in hearing it