Most modern data centers have become quagmires of chaos – mulitiered applications, hybrid clouds, multiple sites, 24x7 expectations, latency-sensitive applications, slashed budgets and a constant rate of change. Many data center managers and admins struggle to simply keep from falling further into the abyss.
The truth is that the future of your data center doesn’t have to be bleak. Most enterprises have adopted virtualization for some of their data and applications, which is a step in the right direction. However, like an airplane aloft without fuel, most enterprise data centers won’t be able to glide along much further as things stand.
The fuel that many data centers are missing is two-fold:
The more servers you have virtualized, the greater management and cost efficiencies you’ll gain. And because virtualization is about maximizing your data center investment, it requires a smart performance management solution to ensure that you don’t exceed your hardware’s capacity or mis-size your virtual machine (VM) containers. With the right tools, your virtual infrastructure will perform and operate efficiently on a daily basis. Read on to understand more about this virtualization management vision.
Once upon a time, the speed of hardware outpaced the speed of software. The CPU, memory and storage capacities of data center servers left the inefficient operating systems and applications behind. Like a jumbo jet carrying just one passenger, servers in the data center spent most of their time sitting idle while the company was paying the price in dollars and IT admins were paying the price in management overhead. Technologists around the world realized that there had to be a better way – modern server virtualization.
As with most data center changes, server virtualization came on the scene as a low priority. At most companies, server virtualization first proved its worth in the development and testing environment. Not only did server virtualization provide greater return on investment for data center hardware, but it also made the life of administrators and developers easier.
If server virtualization would provide such vast improvements for “dev and test,” then why not implement it in production? Because first server virtualization had to mature.
As usual, the “bleeding edge” companies tested the waters and proved that server virtualization could work. After major advancements in performance, reliability and availability, virtualization now is used for production at most companies in the world and for just about every type of application.
On average, companies that started with server virtualization a few years back have now virtualized roughly half of their servers. Virtualization is mature, yet the pace of consolidation has slowed. Why?
Many enterprises have realized that while server virtualization is ready, the management tools they have in place aren’t. Some companies are using aging physical server management tools that have sort of morphed into virtualization management tools but can’t quite do the job. Other companies have implemented a “shiny and new” virtualization management tool but are realizing that, while it has pretty charts and graphs, it really doesn’t help them as promised.
The journey from virtualizing your first test server to virtualizing your last production server has several defined stages along the way. It’s important to understand where you are on the journey so you can figure out how to get to the finish line.
The waypoints along the virtualization journey:
Companies that complete the consolidation of their servers with virtualization can begin the journey to cloud computing. However, the vast majority of enterprises remain stalled out at roughly the 50 percent virtualization mark.
Good virtualization management software can make your life easier in a number of ways.
Gain confidence that your virtual infrastructure is in harmony through analysis, reporting and visibility. Virtualization management software allows you to see what’s happening with your servers, virtual machines, resource utilization and trends over time with one glance. You also can determine whether or not there is a problem.
To help you understand the performance of your virtual machines as well as overall capacity utilization – historically and in real time – virtualization management tools must show deep insight into how virtual infrastructures work. Virtual machines are dynamically sized containers that move from one server to the next and from one storage container to the next. Their resource demands fluctuate on a daily, weekly, monthly or even yearly basis. Virtualization management tools must take all of this into account, know what is normal and be able to use that knowledge to help you prevent performance problems and capacity shortages before they happen.
Service-level agreements (SLAs) between IT departments and higher-level business management define what the acceptable level of service is for the organization’s most critical applications.
Service levels cannot be accurately monitored or proven without a tool capable of those measurements. Combining service-level monitoring and reporting capabilities in a performance management tool is, by far, the most efficient design.
Undeniably, one goal of server virtualization is to drive up utilization of CPU, memory, disk and network resources. As you add more and more virtual machines to the virtual infrastructure, you inevitably experience new bottlenecks. Can you predict your next bottleneck and understand how it will affect your applications?
Good virtualization management tools alert IT admins to how many days until their CPU, memory, disk (I/O and capacity) and network resources are depleted so they can prevent performance problems before they happen.
In most organizations, when a new virtual machine container is created, it is assigned a fixed amount of resources. However, resource utilization changes over time, so ideally resource allocation versus utilization should be analyzed on a regular basis to ensure that your virtual machines are the right size.
Underprovisioned virtual machines can create performance problems, including application slowness or even a total outage.
When a virtual machine is underprovisioned, many IT managers immediately add more resources to it, which can actually lead to overprovisioning and wasted resources. In other cases, the waste occurs when virtual machines are first provisioned because it can be difficult to predict resource utilization prior to server virtualization.
Virtualization management tools can help analyze virtual machine resource utilization, report under- and overprovisioning on a regular basis and even take action to help, when authorized.
As you work toward complete virtualization, what additional hardware resources will you need and when will you need them? In conjunction with bottleneck prevention, your virtualization management tool should model resource utilization to allow enough time for you to size new hardware, budget for it, await approval, deploy it and bring it online in the cluster to share the load – all prior to resource depletion.
Along with forecasting and resource provisioning, your virtualization tool should allow you to run what-if scenarios. For example, you may want to explore – and run actual numbers for – questions, such as “If I doubled the memory on all my servers, how many more virtual machines could I add? And when would I hit another memory bottleneck?” or “If I added five more servers at X capacity, how many more virtual machines could I add?”
Even with the best capacity and performance analysis tools, eventually trouble will occur in the virtual infrastructure. You may encounter a hardware failure, a network switch failure, a SAN controller reset or other issue. When these events happen, your virtualization management solution should be able to identify the source of the problem quickly so that you don’t waste time investigating dead ends until you stumble upon the source.
In many enterprises, even a short outage can cost millions of dollars. Virtualization management tools must provide root-cause analysis data quickly.
Virtual infrastructures require that you employ a monitoring and alerting tool with higher intelligence. When a failure occurs in the virtual infrastructure, rarely does just a single virtual machine fail. For example, if a SAN LUN fails, you need virtualization management tools that will not only alert you that the LUN failed but also that will give you a list of the virtual machines running on the LUN – not 100 separate alerts for every virtual machine that goes down. Or, if a physical server fails, you want to be alerted that the physical server failed and that all virtual machines on that server were successfully restarted on other servers (thanks to vSphere High Availability, for example) – not 100 separate alerts when the virtual machines go down and 100 more alerts when they came back up.
With the rising popularity of cloud computing, many enterprises are rightly considering moving some portion of their data center to a public cloud. However, public clouds are sold in specific computing units and you pay for what you use. A smart tool will be able to analyze your current virtual infrastructure utilization and help you estimate the cost savings you would see if you moved one or more servers to the public cloud.
Your virtual infrastructure supports your company’s applications. When it comes time to justify new infrastructure purchases or replacements, you need to know what company departments, divisions or business units are using what resources. You can’t divide the cost evenly nor should you ask the departments to throw out estimates. You need real analytics to make sound business decisions about upgrades. A smart tool can help you review virtual infrastructure performance utilization over time combined with common infrastructure costs (or customizable values) to provide IT chargeback or showback reporting. You can use these reports to divide the costs of an upgrade, for example, or provide a monthly report to the business units detailing the associated cost of their utilization.
Many outages are caused by unplanned or mistaken changes to the infrastructure. That’s why change monitoring and management is a critical component of any virtualization management tool. In addition to monitoring changes in the virtual infrastructure, many virtualization tools can remediate those changes to “fix” what was changed.
For true visibility, your virtualization management tools need to provide application insight. For example, service-level reporting shouldn’t just notify you that the “SQL Server VM was available 99.9 percent of the time.” You must actually know if SQL Server, as an application, was available 99.9 percent of the time. In another example, the “ping response to SQL Server has been < 5 ms” is not adequate information. You need to know that the “SQL server test query response takes consistently < 5 ms.”