Most every company will face a time where they need to increase their company’s infrastructure and capacity but do not have the room or money to bring in more hardware to satisfy what is needed. In this 2 part series we will describe the obstacles we faced and how we over came them to get our virtualization lab up and running.
A virtualization solution is one that is capable of holding multiple servers virtually on one physical server. This is done so that expensive hardware does not need to be added every time your organization needs to expand. This allows an organization to maximize the capabilities of all the hardware they own. A virtualization lab is used as it provides a significant degree of efficiency, flexibility and cost savings. When Sath set out to create their virtualized lab we did it so we could create a whole identity management lab and production within it. In the process of creating our lab we tried many different systems to get to where we are today.
We needed to build and expand on our current virtualization infrastructure as we were at capacity with our current system. We needed to add fifty virtual machines and provision twenty more terabytes of space. In evaluating our options we faced many challenges. First, we needed to explore a VM based installation that could host a hundred machines and increase storage up to sixty terabytes of data. Secondly, we had to expand our storage capacity without changing our SAN infrastructure, if we did this with our current infrastructure a planned outage would have to happen. With most of our customers being round the clock businesses we cannot bring down the system for a planned amount of time just to preform system updates and upgrades. We had to build a solution that is able to increase our overall lab capacity without experiencing any downtime and that every resource is utilized to their maximum. In addition, we had to be able to create isolated networks for different environments and customers. While doing all this we had to eliminate any single point of failure while maintaining online scalability. In the end our desire was a system that is able to handle and scale up to five-hundred virtualization machines with five-hundred terabytes of data.
We had to move out of a Linux based DRBD storage using iSCSI and VMware with vSphere solutions for virtualization. Keeping the same solution would require us to purchase additional licenses of VMware and move to a traditional SAN and RAID based solution. We looked in many areas to solve these problems. In the end we needed new architecture for our virtualization lab.
Our current SAN was built off a Linux DRBD with iSCSI storage within sixty terabytes of capacity. This has a load factor rate of twenty-five percent with actual usable capacity of fifteen terabytes. The only VMware compatible options available off the shelf were using iSCSI or NFS, or Fiber Channel. All of which were very expensive, came with a high overhead and do not scale up efficiently. After evaluating all the options it was decided that the best option for a scalable and cost effective SAN solution was to utilize a CEPH. CEPH is a unified, distributed storage system designed for excellent performance, reliability and scalability.