Storagepipe Is Now Thrive

GridWay Is Now Thrive


Selecting a Compute Platform for Virtualization Hosts

The digital revolution started off with vacuum tubes and punch cards, and machines the size of a building’s entire floor. For its time, Univac was a modern marvel but nowadays the phones we all have are significantly more powerful than the computers that flew man to the moon.  In all fairness, it is unrealistic to compare the two given the decades of electronics developed from the space program’s initial work but the fact that the Apollo computers couldn’t crash meant all the difference.

In the more complex electronics world, we now find ourselves in we have to design IT systems for redundancy. Selecting the hardware is a critical starting point. Do you go with individual servers, hyper-converged systems (see our previous post on what HCI is), or blades?

Each type has their role to play. Blade servers were quite a bit more popular in the past but are due for a comeback. Virtualization has led to dense installations in data centers and a blade chassis is a great way to minimize the number of network and power cables required compared to the traditional “pizza box” style deployments of standalone servers. The downside is the added complexity and management that comes with this middle layer of hardware.

When selecting a hypervisor host a blade can be a great fit. Generally, hypervisor hosts need significant CPU and Memory resources and a few networking uplinks but little in the way of local storage. Blades offer a solution of deployments with efficient use of power and cooling while minimizing the floor space in use. Blade chassis systems usually hold 16 half-height blades in the same space as 10 single U height servers. That’s a 60% improvement!

The up-front costs associated with the