Network monitoring authority Paessler PRTG has seen the good, bad and ugly of network designs that prevent monitoring a system, which is why their product line solutions cover everything from network management to server performance.

To get an idea of what they’ve seen, PR rep Michael Krems got the company talking about the top five most common mistakes network administrators and IT systems managers make that cause their virtualization deployment to fail.

Here’s what Paessler says will mess up your enterprise’s virtualization:

1. Virtualizing systems without knowing their usual CPU/memory load, disk usage and network usage: You must monitor a system prior to virtualization in order to know how much load it will put on your VM host servers. System with high load may also be not suitable for virtualization at all.
2. Running too many VMs on a host: causing overloads: All virtualized systems suffer in performance.
3. Running too few VMs on a host: spending too much money buying too many host server[s]
4. Compare mid/long-term monitoring results before and after virtualization to ensure quality of service doesn’t suffer.
5. The performance of all virtual systems on a host usually suffers from one virtual system going amok or running into a performance/load peak. Without monitoring, such events often happen undiscovered.

It seems like applying common sense would fix a lot of the issues surrounding virtualization. I wouldn’t place a lot of blame on the network manager though. Much of what goes wrong with a virtual deployment just has to do with the capabilities of the technology. Take these issues for instance:

• Server virtualization impacts network latency.
• Server virtualization creates a network configuration burden.
• Virtual machines present dynamic environment issues for network pros.

As with any new technology, most of the struggle too, is not having the right information. In the end, do you feel like enough conversation surrounds the impact virtualization has on your network?

Who needs a DeLorean and its one point twenty-one jiggawatts when you can have a GigaStor SAS, which stores 288 TERABYTES of network data?

According to a press release from Network Instruments, GigaStor “is the largest retrospective network analysis (RNA) platform available for storing and capturing network packets and transactions for later analysis and investigation.” Network Instruments said that the vast capacity was needed as more networks tap into 10 Gigabit Ethernet, particularly for those companies with larger data centers. The appliance was originally designed for a military customer, according to the company, who needed to store a month’s worth of networking data.

[kml_flashembed movie="http://www.youtube.com/v/efF02kLZ5JY" width="425" height="350" wmode="transparent" /]

[kml_flashembed movie="http://www.youtube.com/v/B50jmPjJlrc" width="425" height="350" wmode="transparent" /]

They said they initially set out to create a very technical community, but they’ve since broadened that mandate to include high-level discussions about the place of DPI at both the network and service provider level. That topic tends to bring out ideologues on both sides, particularly as it’s linked to net neutrality, but hopefully the savvy searcher can avoid the worst of the flame wars and find some useful information, such as vendor profiles and a piece on using DPI to check and improve Skype traffic.

Rosenthal said one of dPacket’s big challenges would be to prove to readers that the site isn’t an industry mouthpiece, which might be tough with financial sponsors like Sandvine and Ellacoya. Still, the more resources there are for complex subjects like DPI the better, particularly if they are willing to tackle both the tough technical aspects of network deployment alongside the creation of ethical guidelines to address how far into the network should DPI peer, and how that information should be handled.

In order to download (or upload) files on any peer-to-peer network whatsoever, all on-campus users have to pass an online quiz on copyright infringement.

But not just once. Passing the test — with a perfect score — enables peer-to-peer access for six hours on the user’s on-campus registered machines, presumably enough time to download that (legal) song, TV show or e-book. The next time, the student, staff or faculty member has to go to the intranet Web page and take the randomized test again, for a maximum of eight uses per month (which, kind of like vacation days, can accrue to at most 20).

While it’s certainly innovative, what’s the point? Nobody is better off, except maybe the RIAA when they use the quiz to show illegal downloading was “willful infringement,” which can bump damages up to $150,000 per song. Those intent on downloading tend to find creative workarounds , while those with legitimate needs are unduly hassled.

In the enterprise, we often hear that education of Net do’s and don’ts is critical, but randomized quizzes and P2P privilege accrual seems a pretty complicated way of doing it.

On the other hand, the statistics are impressive: The year before the program was implemented, the university got 800 copyright complaints; since, they’ve only received eight. Imagine the work productivity boost if you inserted to a picture of the boss every time an employee accessed a flash game or MySpace profile.

On second thought, no.

[kml_flashembed movie="http://www.youtube.com/v/40Airs2yjmU" width="425" height="350" wmode="transparent" /]