Data centers and server virtualizaion
Globalization, security and disaster recovery considerations are driving business to diversify locations across multiple regions. In addition, organizations are looking to distribute workloads between computers, share network resources effectively and increase the availability of applications. With the ultimate goal of eliminating downtime and sharing data across regions, enterprises are deploying geographically dispersed data centers to minimize planned or unplanned downtime.
The size of these data centers vary depending on the organization’s business needs. Some enterprises are building mega data centers that have upward of 450, 000 square feet of floor space that house thousands of servers, storage and networking equipment including switches, routers, firewalls, load balancers, tape backup libraries and disk arrays; banks of uninterrupted power supply batteries that ensure continuous running of a data center in case of power failure. These data centers run standard and mission critical applications and continuously process and store data. To give you a perspective, the square footage is equivalent to 8 football fields put together. Facilities of this magnitude demand lot of power, not just to run the equipment but also for cooling. In an effort to save power, enterprises are deploying data centers in countries where the temperatures are mild to cold throughout the year. Some even build data centers near rivers so that hydropower can be used to reduce the carbon footprint.
Previously, organizations deployed different servers running multiple operating systems and software applications. Maintaining servers of various hardware types and updating diverse OS’s and software applications became very difficult. The solution was to consolidate servers, which allowed sharing of resources between applications, thus lowering cost of ownership and maintenance, and increasing efficiency.
The same problem existed with storage. Direct attached storage became inefficient and difficult to manage. Enterprises consolidated storage and deployed SAN and/or NAS for different applications, depending on the required performance.
As data centers have grown in size and complexity, enterprises have started adopting server virtualization technologies, including cluster servers, NIC teaming and WMWare to achieve increased efficiency. These virtualization technologies require layer 2 connectivity and thus pose several challenges when deployed in fault tolerant mode across remote locations.
Next week, I will discuss this requirement in detail and review the issues that arise when extending layer 2 VLANs across a WAN. In a future post, I also will discuss solutions for overcoming the problems that this requirement poses.
