How do you accelerate VDI performance? VDI offers great opportunities for efficiencies, but can also present deployment challenges. How do you ensure a great user experience and keep costs within budget? All while dealing with the space, power and networking challenges?
Here’s what we’ve learned in deployments large and small (one of which was the largest Citrix deployment so far). There are three issues we’ve seen our customers grapple with in VDI deployments. Here they are, along with the ways that virtual I/O was able to help.
1) VDI Performance Bottlenecks: Virtualization Review wrote, “Ultimately the most important in the success of VDI is the I/O required to provide a better-than-PC or equal-to-PC end-user experience.”
VDI is I/O intensive, and it operates in an environment where I/O latencies can add up. Latencies include server-to-server communications (when booting, or when accessing a thin app server), server-to-storage links, and the obvious server-to-end user link. To accelerate VDI performance, you want to minimize latencies and maximize bandwidth to avoid bottlenecks.
Xsigo eliminates bottlenecks with up to 80Gb bandwidth to each server. And Xsigo’s low I/O latency – typically 3X faster than 1G Ethernet – accelerates network, storage, and server-to-server communications to deliver responsiveness that users will appreciate.
2) Network Complexity: When specific workgroups require physically isolated networks, complexity may grow. You could host those groups on separate servers, or provide multiple physical connections to each server. But those options add cost and complexity. With virtual I/O, you can configure isolated environments within a single group of servers, free of hardware limitations. Establish multiple connections to each server, in software, without even entering the data center.
3) Cost: Data centers are expensive, yet VDI costs inevitably get compared with the price of a lowly desktop. Between the capital costs of servers, storage, and networking, and the cost of space, power, and cooling, it can be tough to keep VDI expenses at the level that managers expect.
Virtual I/O helps cost in three ways:
Reduced server size: With virtual I/O, all connectivity types are consolidated to one wire. You need just two I/O links per server (for redundancy), giving you the flexibility to choose the smallest, most efficient server that meets your needs – even 1U servers or blades. PCI slot count is no longer a concern since you can configure all needed I/O with just one or two host adapter cards — or possibly even with no cards at all if you deploy Xsigo’s 10G Ethernet fabric. All of these options flexibly provide redundant, isolated Ethernet and Fibre Channel connections from your servers to the networks and storage.
Simplest possible infrastructure: I/O consolidation reduces switch port and cabling requirements by 70%, saving space and cost. Furthermore, as you grow you can add servers with just two I/O cables each. Minimal space requirements plus the simplest possible cabling accelerate new roll outs.
Re-use that 1G infrastructure you have: Virtual I/O is not a rip-and-replace technology. It installs along with servers you’re adding and connects to all of the Ethernet and FC switches you already have. Because it connects to existing gear, it makes upgrades easier. If you want 10G Ethernet to each server but also want to leverage the 1G infrastructure you have, here’s the answer. Just connect the 1G to the I/O Director and those networks are accessible to every server.
