The State of Servers: From Physical to Virtual Machine
Are you prepared for the Big Data blast?
For most Fortune 500 companies, hardware availability is no longer a problem. Systems have become more powerful over time (multicore, lots of RAM, etc.) and are constantly available to detect breakdowns and alert you if something happens - automatically. This helps Administratorstake proactive actions early on, before issues become serious problems.
Two of biggest hardware issues are Cost Optimization (power, square meters, etc.) and Storage (which is still very expensive for big data).
For over 10 years, a majority of server landscapes have moved from boxes to virtual machines. This optimizes cost when mapping out the traditional setup of one critical application per server.
The performance of the application still relies heavily on the virtual machine and underlying physical server itself. CPU and RAM are critical factors that all Administrators need to carefully review and analyze in order to anticipate potential problems on the “Application Service Supply Chain”: Box – VM – OS – Application.
Applications and servers are completely linked (you don’t buy boxes just for fun...). Application Administrators need to understand how the critical application is using the server’s resources (by processes, interactions with other applications on other servers, user requests, etc.) in order to conduct capacity planning.
How do you know when you need to buy a new box to deploy new VMs that will support new users for your main applications (like Exchange and SharePoint, which are used by the entire company) if you don’t collect and analyze performance and usage statistics at the application server levels?
Whether you are a private cloud user or have a massive virtualized datacenter, you need to continuously adjust the needs of the business in relation to your IT resources.
In order to know when and how the company needs to extend/reduce server resources, you must collect data at the usage level of the application to detect any bottlenecks, forecast needs (on a historical basis or for major changes that will happen in the company), and plan the costs.
Analyzing the performance of the application and its usage requires correlating these statistics with the system performance (VM or physical box) if you want to grow based on an optimal cost path.
But, not everyone works in a Fortune 500 company. Many companies do not have a big datacenter organized in the private cloud that guarantees the system to be available 99% of the time.
Most mid-sized companies that have an IT department are still impacted by hardware outages. Therefore, it's essential to conduct constant checks on key performance indicators on their individual servers. Analyzing theimpacts of application processes on the performance and availability of the VM and box is mandatory to avoid any global breakdown of services provided by critical applications.
Servers, Virtual Machines, OS and applications are completely linked to the service provided to end users.
To optimize costs, avoid availability/performance problems and ensure strong management on your IT resources, monitoring and analyzing the global IT chain for specific services is mandatory.
At the end of the day, only two things really matter: the cost and performance of the service delivered to your business.