Network Virtualization: A Primer

What is it?

PDF DownloadSun pitches it this way: The computer is the network. That is, to the user, all the resources on the network appear to be part of the computer sitting on the user’s desk. This “network virtualization” effect is achieved by installing software and services to manage the sharing of storage, computing cycles and applications.

Is it just spin?

No. Network virtualization (NV) can reduce the cost of running data centers and corporate networks by making both hardware and software more efficient. Servers can share their processing power, for example, and applications their components. Data center managers will be able to manage computing and storage capacity more effectively.

How does it work?

Server software and network hardware sits between clients (desktop computers, Web application servers, wireless devices, etc.) and the data center resources (storage boxes, enterprise software and extra processing power) being shared. Major NV initiatives, like Sun’s N1 and Hewlett-Packard Labs’ “planetary computing,” build on three current technologies: storage virtualization, grid computing and distributed application services.

What’s storage virtualization?

A technology from the world of storage area networks (SANs; see primer, Baseline, January 2002). Other servers on the network can access a SAN regardless of where the storage area is or what operating systems it’s running. In most storage virtualization systems, a virtualization “router” intercepts commands from a server, then redirects those requests to the appropriate storage device.

What’s grid computing?

A sort of peer-to-peer computing, in which a program can access spare computing cycles on other, networked computers based on demand, taking advantage of underutilized processing time to maximize efficiency. Software agents run on each “gridded” system, taking commands from a master system that assigns processing tasks and tracks utilization levels.

What are distributed application services?

Software tools that essentially allow apps to run across the entire network. This includes Web services (see primer, Baseline, October 2001), Web service technologies like The Mind Electric’s Glue (a combination of Web services standards, Java interfaces and the eXtensible Markup Language) and Microsoft’s .NET architecture, which can be used to build virtualized application services.

What’s the downside?

Current storage virtualization systems may be incompatible. Most commercial software isn’t written in a way that can take advantage of grids, so it may be some time before they’re of any value to business. There are still security and compatibility concerns about Web services standards.

Who’s using NV technology now?

Internet imaging provider Global Explorer uses Sun’s grid computing software to turn its Web server farm’s spare computing cycles into a virtual image processing supercomputer, and several software companies use grids to handle the heavy lifting of compiling software code. Maimonides Medical Center, the third-largest independent teaching hospital in the U.S., uses storage virtualization technology from DataCore to reduce the cost of adding storage.