How to Do Capacity Management in the Cloud (Webinar Recap)
Cloud adoption is accelerating. Soon, 80 percent of all IT budgets will be committed to cloud solutions.
But as your organization moves into the cloud, how will that change your approach to capacity management?
In our recent webinar, How to Do Capacity Management in the Cloud, we answered that very question. Check out that webinar on-demand—or read the recap below.
The Cloud Defined
The public cloud is what everyone typically thinks of when they consider the cloud. Public cloud options include Amazon Web Services, Microsoft Azure, and Google Cloud Platform.
The private cloud is something else. It’s wrapping your own infrastructure in a cloud framework.
Hybrid cloud means a combination of public and private cloud. This allows you to potentially move things back and forth across cloud offerings.
No matter which way you slice it, the cloud is different from traditional IT. (Traditional IT involves everything from legacy workloads to highly virtualized systems.)
While moving to the cloud is probably in your future, you’re not going to switch everything overnight. That’s why most organizations are opting for a gradual adoption with hybrid IT. This gives you peace of mind—but it can also add some complexity to managing your infrastructure.
What Runs in the Cloud?
There are three types of offerings in the cloud:
- Infrastructure as a Service (IaaS)
- Platform as a Service (PaaS)
- Software as a Service (SaaS)
IaaS and PaaS require capacity management, while SaaS is more turn-key.
Cloud native applications are specifically designed to run in the cloud. These are built on micro-services, architected to scale out, and designed for automation. Typically, any application developed over the last five years will have gone through a cloud adoption.
Capacity Management is Still Relevant in the Cloud
Traditionally, there are seven driving factors for capacity management. And nearly all of them still apply to capacity management in the cloud.
- Business critical services (that you can’t afford to fail)
- Seasonality/peaks in business activity (shifts in demand that require planning)
- Business growth (meeting new demands takes careful planning)
- Regulatory requirements (having a continuity plan for your services—including capacity requirements)
- Provisioning lead times (which is less relevant in the cloud where you can spin up new instances pretty quickly)
- Cost optimization (being careful about how you spend your resources)
- Agility (having a clear understanding of your current position allows you to find and exploit new opportunities)
Best Practices for Capacity Management in the Cloud
1. Accept the Cloud
Resistance is futile. The cloud is here to stay.
That’s because the cloud offers simplified resource provisioning. Like it or not, there’s a self-service component to the cloud. Users can order their own infrastructure or resources.
That can make life harder for capacity planners. But instead of implementing policies to control their behavior, look for other ways to manage capacity. And try to avoid anything that negatively impacts provisioning lead time. That will just cause shadow IT.
2. Know Your Capacity Unit
To cope with growing demand, you need to make sure you have enough instances. And that means knowing the optimal capacity unit for each application.
The capacity unit needs to be determined by you, because your circumstances are unique. To find your capacity unit, you need to consider the following.
First, look at the cloud platform. The cloud can have different instance sizes, so you need to find one that fits your specific workload.
Next, consider your software—what is it doing?
Finally, take demand patterns into consideration. Is there seasonality involved?
As you weigh these together, you’ll find your optimal capacity unit. Then you’ll be able to answer questions like...
If you run out of resources on the existing set-up, how many should you add?
3. Prepare for Multi-Tenancy
When it comes to the cloud, you’re not alone.
The cloud has multi-tenancy. You’re sharing resources with other tenants, but you don’t know who you’re sharing with. You don’t have visibility. But concurrency of events across tenants can impact your performance.
Cloud providers are typically good at isolating this. Impacts from other tenants should be minimal, but you do need to plan for multi-tenancy when you size your applications.
4. Understand Cloud Distribution
When you scale out your applications—by adding instances to cope with demand—those may not end up in the same data center as your original application.
You do have the option to specify location—but that comes at an extra fee.
From an availability perspective, it can be good to have your application scaled out across multiple locations. If one goes down, you can still provide a service.
But it will always be a trade-off. You’ll need to consider things like:
- How sensitive your applications are to latency
- What the data transfer fees are
- What the availability of the application is
- Which regions you need to support with the application
5. Focus on Micro-Services
A traditional approach to capacity management isn’t going to work in the cloud. You’ll no longer see components. The smallest entities you can work with are micro-services.
Micro-services are aggregated into applications and then into business services. The aggregation becomes more difficult than before. Micro-services are smaller building blocks—and there’s plenty to combine to build an application.
6. Adapt to Speed of Change
Things change fast in the cloud.
The cloud is typically associated with DevOps and agile frameworks (methods of developing software in shorter release cycles). When you combine that with the component focus of the cloud, the number of changes that can potentially impact performance will increase.
You’ll need to have real-time access to the data to monitor that.
7. Switch from Reactive to Proactive
Proactive capacity management requires you to verify and do a proper analysis of provisioning requests before something gets provisioned.
In a self-service environment, that’s hard to do. You can’t be there all the time to make sure they’re asking for the right stuff. The best you can do is provide them with a different set of configurations.
To manage a self-service environment, you do need to do a mix of proactive and reactive. The reactive part will be focusing on reclaiming resources that aren’t in use to right-size your environment.
But you need to make sure you’re focused on the proactive side of capacity management.
8. Prioritize Efficiency
Ideally, cloud is cheaper than traditional IT. But that assumes you’re using the cloud in the right way.
Wasteful use defeats the that purpose.
The role of capacity management shifts to deliver capacity to the business at the lowest possible cost.
It’s up to you to make informed decisions based on the cost model of the cloud. Answer questions like… Should I use dynamic or reserved resources? What are the downstream transfer charges?
All of this needs to be included in the cost of operations.
What to Look for in a Capacity Management Solution
Finding the right capacity management solution means taking your needs into account. Are you running traditional IT? Are you moving to the cloud? Are you already in the cloud?
At minimum, a capacity management solution for the cloud needs to:
- Assist in the migration and deployment of workloads
- Monitor ongoing resource usage and charge
- Provide end-to-end visibility for hybrid IT
- Optimize and predict financial impact
- Do long-term planning to meet business demands
Capacity Management in the Cloud Doesn’t Need to Be Complicated
Some things are going to change. But others will remain the same.
The value proposition for doing capacity management is still there. Long-term planning and an understanding of demand will still require capacity management.
Capacity is infinite in the cloud—but the cost for that capacity is not. You need to understand what the cost will be and how to get a predictable cost for that operation.
Most organizations will have a hybrid IT approach for the foreseeable future. So you’ll need a capacity management solution that can deal with both sides. At some point, business services will cross those boundaries and go from traditional IT to cloud (and vice versa). And you’ll need to be able to see everything in one view.