No matter which company you work for, you are probably about to implement a new virtualization infrastructure or...
ready to upgrade your existing virtual infrastructure. Since virtualization is growing rapidly more sophisticated and server hardware changes every day, how can you select the best server hardware for your data center infrastructure or project?
To save you the time and headache of doing research, working with vendors, and testing out hardware, here's my short list of the best hardware for virtualization and some of the pros and cons of these hardware technologies for virtualization environments.
The top server hardware for virtualization
There is a lot of great hardware available in the market. If I were designing a new virtual infrastructure, there are three companies that I would consider first. These three companies were winners in SearchServerVirtualization.com's Best of VMworld 2009 product competition:
- Cisco Systems Unified Computing System (UCS). Cisco's entry into the server market is like no other traditional rackmount or blade server. Virtualization and 10 GB Ethernet are standard in this high-performance and highly scalable chassis. If you are implementing a new virtual infrastructure, you should seriously consider Cisco's UCS.
- Xsigo Systems VP780 I/O Director 2.0. Pronounced "z-I-go," the VP780 is the flagship solution for I/O virtualization. You might not have heard of I/O virtualization but, believe me, once you understand how it can tremendously reduce the cabling and make expansion easy, you will be sold on virtual I/O, just like I am.
- Afore Solutions ASE3300. Is it possible to move a running virtual machine from one physical location to another across a wide area network (WAN), with zero downtime? It may sound impossible, but Afore's ASE3300 enables vSphere to perform long-distance VMotion.
Virtualization hardware comparison
For purposes of this article, I won't do a full-blown virtualization hardware comparison. Instead, here is a brief, general description of hardware options for typical virtual server and cabling implementations.
- Traditional vs. blade servers.While the blade chassis may cost more up front, over the long term blades save a company money over traditional servers. Still, you should verify that the blade chassis you consider will scale to the level of performance your data center infrastructure needs. In addition to saving your company money, blades simplify server management and cabling in your data center. Leading blade server vendors are Hewlett-Packard Co., IBM Corp., and Dell Inc. On the other hand, you may want to use traditional physical servers as your virtualization hosts. While your rack density will be lower, the hardware can scale much higher - allowing for huge amounts of CPU and RAM that could run many more virtual machines than can a blade system. Also, for smaller implementations, traditional physical servers may actually cost less than blades because you won't have to purchase the blade chassis. (And see this article for more on blades vs. rack servers.)
- Traditional I/O cabling vs. virtual I/O. With traditional I/O cabling, every server has at least 2 or more GB Ethernet cables, and storage area network (SAN) connections such as Fibre Channel (FC). In many caes, the FC connections are dual-ported (i.e., two cables) and there could be four or more Ethernet cables. As the demand on the server grows, you may have to add additional adapters and connections. As you add additional FC and Ethernet connections, you must add ports on your SAN and GB Ethernet switch. These ports are quite costly and, like all the cabling, these switches take up space. For those considering a new virtual infrastructure or upgrade to their existing VI, solutions such as Xsigo's virtual I/O solve these problems - simplifying cabling and reducing the cost of cabling, connections, and switches. But for smaller virtual infrastructure implementations, traditional physical cabling may cost less because you don't have to purchase the virtual I/O Director and special adapters.
About the author
David Davis is the director of infrastructure at TrainSignal.com – the global leader in video training for IT pros. He has several certifications including vExpert, VMware Certified Profession (or VCP), CISSP, and CCIE #9369. Additionally, Davis has authored hundreds of articles and six video training courses at Train Signal, where one of the most popular course is the VMware vSphere 4 video training course. His website is VMwareVideos.com. You can follow Davis on Twitter or connect with him on LinkedIn.
Dig Deeper on Introduction to virtualization and how-tos