Removing a major barrier to enterprise agility – virtual connectivity

It often takes a start-up VC company to spot and exploit a fundamental weakness when an IT infrastructure trend emerges. Suddenly everyone “got” the idea of enterprise virtualisation and data centre clouds, but a basic weakness was physical connectivity that could block the way to enterprise agility.

Now that virtual server and virtual storage environments are becoming commonplace, the goal of enterprise agility – to quickly and easily change the IT infrastructure to keep up with rapidly changing business needs – has become closer. But the ‘spaghetti-cabling’ between physical servers across the network to physical storage of different types – often with missing links that prohibit some connection paths – is a major barrier to making virtual changes.

Yet Xsigo came up with a solution back in 2007, a couple of years before the market realised it even had a problem; it has since taken this on. Its solution, loosely a “virtualised server fabric”, is conceptually very simple: connect all the servers with very fast (40 Gb/s) links to its new VP780 I/O Director box, and connect all the network devices including all the storage to the other side of the box – and put some clever software inside to make all the needed virtual connections.

Then make it easy for the server manager to instruct the software to make or break these virtual connections using a visual drag-and-drop GUI, so changes are achieved within a few seconds without involving a specialist network manager. (The OS will recognise any change in 1-2 seconds and Xsigo’s software manages all worldwide name and MAC addresses.)

Xsigo points out that, typically, only 5% of the bandwidth of the physical connections is used. So, for example, many virtual servers can, by this means, share one physical line to a storage module before a bottleneck occurs. As a result, connection consolidation is facilitated – to greatly reduce spaghetti-cabling complexity and costs. Currently the software handles up to 15 types of I/O, including 4 or 8Gb Fibre Channel (FC) and 1 or 10Gb Ethernet.

This “north-south” network connectivity simplification gives high granularity for better QoS. This is matched by inherent “east-west” server-to-server connectivity; if two virtual servers residing on different physical servers need to directly communicate, they can now do so via the fast 40Gb/s links (touching without passing through the I/O Director box).

This is such a very obvious solution (albeit much more complex than I have probably made it sound), that one might wonder why the likes of Cisco, HP, Brocade did not think of it. They and others are now busy working on their own functionality to better compete with Xsigo but, no surprise, an independent start-up tackled this basic problem first – and is reaping a revenue reward right now.

Meanwhile, enterprises can move a major step closer to the goal of full IT agility to match their business needs.