Virtualization technology provides IT organizations with many tangible ways to reduce costs in the datacenter. By consolidating physical servers through virtualization, companies can reclaim expensive data center space, reduce capital expenditures and save on energy and cooling costs. However, when you begin to expand the virtual environment and you have a number of virtual machines competing for resources – effective visibility and management becomes extremely complex.
Quite often IT pros are focused on commissioning new virtual machines to support new projects, but determining the optimal settings can be difficult due to the dynamically changing characteristics of workloads which share common server and storage resources. Virtualization specialist often need to set VM capacity configurations by collaborating with application owners who wish to adopt best practices promoted by software vendors, which were established to support the applications within physical environments. The result is an abundance of oversized virtual machines that are wasting resources—otherwise known as VM sprawl.
In order to realize the full cost-saving benefits of virtualization, IT pros must have the right policies and tools to effectively minimize operational complexity and optimize workload performance to ensure server and storage resources are used most efficiently. Virtualization specialists need to be able to visualize across multiple hypervisors, VMs, servers, and storage to understand how deploying new VMs or configuration changes will impact the entire environment. Most importantly, virtualization specialists need the right information to collaborate with application owners and effectively demonstrate capacity configurations that will provide optimal application performance.
So how do you get there? Here are 9 steps to better Capacity Management strategies that will help you achieve the high performance you want, without sacrificing the cost savings you need.
Step 1: Three Mechanisms to Leverage in Addressing Capacity Management Challenges:
1. Virtualization of Resources
a. Group resources into pools that are assigned to different VMs by a hypervisor
2. Dynamic Resource Scheduling
a. Automatically assign and direct resources to VMs based on current demand
3. Seamless Migration
a. Enable implementation of server clusters through resource scheduling
Step 2: Understand the Six Steps of the Capacity Management Workflow
- Determine total capacity
- Calculate usable capacity
- Find peak utilization
- Determine leftover capacity
- Calculate average VM size
- Compute new VM capacity
Step 3: Know Your Memory Metrics
Memory is the resource that is normally the most constrained. It’s useful to consider two memory metrics: Average Memory Active and Average Memory Consumed.
Step 4: Cache Your Database
Consider caching in a database application that has been allocated 8 gigabytes of memory, but that actually needs only 2 or 4 GB. Because the application can cache, it will, which means it will routinely be using its full allocation of 8GB.
Step 5: CPU Ready – or Not
In a VMware environment the best metric to use is CPU Ready. Consider a simple cluster with two hosts, each with dual quad-core processors.
Step 6: Right-size Your Virtual Machine
Over-provisioning can actually degrade overall performance. When VMs are correctly sized, it does not matter if they are consuming all of the memory they have been allocated. In fact, a right-sized VM should consume all of its allocated resources and never be in need of any more.
Step 7: Remember That Over-allocation Also Increases Overhead
It is important to consider overhead. The reduction in overhead can be significant, especially when no additional memory can be added because the hosts are already configured with the maximum supported.
Step 8: Sizing a new VM
This process involves the following steps:
- Al locate the minimum
- Monitor resources intensely
- Need more resources?
Add small increments
Add VM to regular monitoring schedule
Step 9: Resizing existing VMs
- Determine analysis period
- Find peak utilization
- Compare peak to allocation
- Forecast and add buffer and change VM allocation
The right performance and operations management software can go a long way to help simplify capacity planning and management. Go beyond one-dimensional performance monitoring to accelerate IT and cut operational costs by reducing infrastructure complexity with insight into virtual environment performance. Foglight for Virtualization provides the insight to know now, with end-to-end performance monitoring, and the flexibility to do now, with powerful operations management.
For more information on how Foglight for Virtualization can help you visualize, analyze and optimize your virtual environment visit: https://www.quest.com/products/foglight-for-virtualization-enterprise-edition/.