Also, improving ROI, Cisco’s Network Admission Control, more
Intel’s Wi-Fi focus
Regarding “Intel-Cisco deal may be big for Wi-Fi”: Interesting that Intel is hedging its bets and keeping its eye on Wi-Fi 802.11 products, which are getting better with each release. It will also do Intel good to focus on the initial 802.16-2004 release chips for the base stations because the 802.16e (mobile) WiMAX systems certification will be a long time coming.
The 802.16-2004 products will eventually provide a cost-effective broadband wireless gateway link (10-40M bit/sec) to these open standards-based wireless mesh nodes being deployed in major metro markets. This WiMAX device will extend the life and value of these 802.11 standards-based, unlicensed, low-cost systems for 3-4 more years. Plus, the ubiquitous Wi-Fi customer access devices will finally be able to use their broadband capacity (5-10M bit/sec) because of these wireless mesh networks.
Jim Aimone
Myrtle Beach, S.C.
Improving ROI paramount
Regarding Kevin Tolly’s column, “The fairy tale of ‘net neutrality’”: In a so-called competitive environment, why can’t I charge both application providers and consumers? If I am a service provider in a non-monopolistic situation, then I am going to look for all ways to improve my ROI — that is just normal business practice. The “not my” VoIP providers (Vonage, Lingo, Skype) are competitive and parasitic as far as my investors are concerned, so I have no obligation to provide them anything other than basic access. I certainly should not be forced to give them guaranteed transport or enhanced QoS.
As for the consumer, “my” VoIP will still have to be competitive with “not my” VoIP, as well as with the PSTN, mobile and other service provider voice services. I am under no obligation to make it easier for my competitors to beat me, surely?
Joseph Neil
Director of marketing
Caspian Networks
San Jose, Calif.
Creating problems
Regarding “Net access security plans sow confusion”: Cisco’s Network Admission Control (NAC) seems to have value for managed devices. But how about unmanaged devices? How does one re-mediate devices that may or may not have appropriate software subscriptions (for example, AV signature update)? For LAN access, blocking makes sense. For remote access, especially SSL VPNs, it may create more problems than it solves.
Mark Elliott
Austin, Texas
IPv6: Right on time
Johna Till Johnson’s column, “IPv6: Time’s still not right” is full of misinformation. First she claims that IPv6 requires 250% more bandwidth than IPv4 for 64-byte messages. But if you do the math, you’ll see that an IPv6 packet’s 20 additional bytes amounts to only a 24% larger packet with a 64-byte payload (104 vs. 84 bytes). She admits that little traffic on the Internet consists of such small packets, so this argument is a red herring. When you look at the actual packet size distribution on the Internet, IPv6’s overhead is negligible.
Then Johnson claims that IPv6’s only other advantage is a larger address space, and even dismisses that, claiming that we don’t need more addresses because network address translation (NAT) solved that problem. But NAT is nothing more than an ugly hack, and an expensive one at that. I can’t count the number of hours I’ve wasted debugging problems eventually traced to NAT compatibility issues. IPv6 completely eliminates NAT. And dynamic IP address assignment. And packet fragmentation. And network renumbering. The value of IPv6 is not its directly addressable endpoints, it’s that NAT isn’t munging IPv6 packets from one endpoint to the other.
The death of NAT also vastly simplifies router and firewall configuration. NAT misconfiguration is a leading cause of configuration errors and inadvertent security holes in these devices. Johnson believes IPv6 security is no better than IPv4’s. But IPv6 provides end-to-end encryption and address authentication, where IPv4 is at best end-to-gateway encryption with no authentication. And IPv6’s hierarchical address allocation scheme makes it impossible to spoof IP addresses, eliminating a host of denial-of-service attacks.
Johnson’s claim of IPv6 routing troubles is overblown as well. IPv6 was designed to route just fine on the current network infrastructure, bringing advantages like QoS and Internet-wide multicast to the table at no extra charge.
Mel Beckman
Ventura, Calif.
Winrot and what-not
Regarding Mark Gibbs’ Gearhead column, “Days of Winrot and OS what-not”: Thanks for bringing attention to Winrot, which has to be the single biggest pain in the butt for Windows users. In the good old days of Win98, I used to back up my data files, then boot up from a floppy. I’d simply do a deltree command on the Windows & Program Files directories, throw in the Win98 CD, install Windows, install apps and restore data. Since the rest of the disk was left intact, and I was in the habit then of not installing most of my apps in the Program Files directory, when I re-installed my apps to the existing folders, most of my settings would remain, and after tweaking Windows back to my liking, I was off and running.
Today, I don’t seem to need to re-install XP as often as I did with Win98 or even Win2K. However, when I do, it kills a day. I have to back up my data, wipe the drive clean, then re-install everything. The other problem with data backup today is that it’s not as simple as it used to be. Now, I have to remember to get my bookmarks and passwords from FireFox, my remotes from PC Anywhere, my Outlook .pst files, my OneNote files, and more. None of these things are easily located in My Documents and the Favorites folder. And what about all the users who have no idea how to do something like this re-load and just live with the Winrot? It’s no wonder they just trash the whole computer and buy a new one when it becomes unbearable.
Ron Stockwell
Sales representative
Data Futures
Overland Park, Kansas
