Benson Schliesser - Latest Comments

Re: POTD 26.Aug.2010 – Hummingbird Moment

bensons — Fri, 27 Aug 2010 00:02:27 -0000

Just don't spill your drink. ;)

Believe it or not, this is my backyard. Em has been filling a bunch of Hummingbird feeders all year, and we've collected a few regular visitors. Check out some better images on my Flickr account; I posted 6 photos earlier starting with http://www.flickr.com/photos/bensons/4930294525/. They're not perfect, but they're pretty.

Re: POTD 26.Aug.2010 – Hummingbird Moment

facebook-1100760425 — Thu, 26 Aug 2010 22:42:28 -0000

whoops, comments appear below. I will blame ... um ... <glances around=""> ... vodka ... for that double post..</glances>

Re: POTD 26.Aug.2010 – Hummingbird Moment

facebook-1100760425 — Thu, 26 Aug 2010 22:41:19 -0000

Hummingbirds are my favorite. Where did you take this?

Re: POTD 26.Aug.2010 – Hummingbird Moment

facebook-1100760425 — Thu, 26 Aug 2010 22:40:55 -0000

hummingbirds are my favorite. Where did you take this one?

Re: Broadband vs. Internet Speed: Not So Fast

Dave — Wed, 25 Aug 2010 18:54:40 -0000

You can ignore George, he's a known troll whose organization lobbys on behalf of the likes of AT&T and Comcast.

Re: Pic >1k words: U.S. Health Care Ineffectiveness

narconon07 — Tue, 24 Aug 2010 06:32:59 -0000

Emerging Issues in Health Care Regulation: Protecting Patients or Punishing Providers? It's obvious that politicians getting into healthcare and the greater control of medicine (by the government) are signs of something not working right. When are we going to do something about this ?

Best Drug Rehabilitation

Re: Broadband vs. Internet Speed: Not So Fast

georgeou — Sun, 22 Aug 2010 19:32:54 -0000

"Oversubscription may not even need to be documented, per se, since it will be misunderstood and can be easily gamed."

If you understand that, I'm not sure why you keep hammering me about it when I go out of my way to talk about all ISP issues. As I said in my previous response, core contention can't be solved on a technical basis simply because the amount of aggregate traffic coming from the end points can always exceed the fastest fiber connection on the planet. The long-haul transit links will likely run the quickest fiber transceivers possible anyways since it's the cable plant that's the most expensive part of the game. Furthermore, caching is the fundamental solution for video on demand.

Caching can't be applied to real-time communications e.g., video conferencing. But that's a lot less traffic that the Tube sites because it's just a lot less popular. So the long haul connections should be used for things that can't be cached and that's pretty much what's happening already. The TCP speed penalty turns out to be a good limitation as it offers a great incentive to build things near the end users.

The type of contention that really needs to be solved is the last few miles. Once you have a network of 10,000 plus broadband users, it becomes economical to cache things for them. The cable industry faces a deeper problem of contention on the last mile as well as the backhauls near the end users. The Telcos don't really have to worry about the last mile, but they do have to worry about the backhauls that connect those miniature DSLAMs.

Re: Broadband vs. Internet Speed: Not So Fast

georgeou — Sun, 22 Aug 2010 19:23:34 -0000

I'm not sure why you keep saying that I don't want to talk about oversubscription when I specifically address the issue in my transparency article. It feels like you're almost looking for something to criticize me about by only reading 1/10th of what I wrote. The sync rate issue was only one of many issues I mentioned. Let me cut paste a section that specifically talks about oversubscription.

"Where does the most sharing occur? Does it happen on the “last mile(s)” which connects the home to the broadband provider’s equipment or does it happen further upstream? How oversubscribed is the system on the most oversubscribed portion of the network? In other words, how many customers are sharing each Mbps of bandwidth."

The latest article is specifically talking about the problem with the FCC report that completely misrepresents the issue and I'm under no obligation to repeat every point I've ever made in the past when I'm trying to focus on a specific issue in a specific blog post.

My transparency article does in fact talk about all the ISP issues that slow the end-user down including oversubscription and dynamic video bandwidth sharing e.g., U-verse TV. What I tried to do is come up with a simple way to measure and report oversubscription. Read the section on "Computing the average achievable performance of broadband". As far as I'm concerned, the ISP is responsible for any contention that touches their own network. However, the transit bandwidth issue can only be solved through more in-premise or same-IXP (Internet Exchange Point) caching over more peering connections paid or unpaid. There's simply no feasible way or even technical way to make the core of the Internet support every gigabit end point on the planet concurrently.

My problem with the OTI proposal is that they're calling for no more than 2:1 oversubscription which would be pure fantasy. Topolski's demand that ISPs have to write a check to the end user for not ensuring 50% minimum performance is silly because that would simply mean that the ISP would have to charge higher rates to begin with. As you even pointed in your article above, the low margin characteristic of broadband makes higher oversubscription ratios a requirement. This is especially true when consumers are so price sensitive.

Re: Broadband vs. Internet Speed: Not So Fast

bensons — Sun, 22 Aug 2010 19:08:14 -0000

Thanks for the link to your post "The need for a broadband transparency standard" (http://www.digitalsociety.org/2009/09/the-need-for-a-broadband-transparency-standard/). It does a better job outlining the issues, including a reference to oversubscription which is missing in your more recent article.

If you confine your view of the service provider's performance to sync rates and frame overhead, then it's a logical conclusion that the server is to blame. That's the direction in which the fax machine analogy misleads the reader. To your credit the article was updated to talk about transmission protocol (i.e. TCP) performance, which is another important factor. But it's a factor that can't fairly be assigned to anybody (user, server, OS vendor, broadband provider, etc) which I suspect is why you're happy to include it.

My frank suggestion is that you also consider more of the "middle" of the network path, specifically focusing on the part that is managed by the broadband service provider. Since oversubscription rates (on aggregation and core routers, as well as transit and peering links) within a broadband provider's network are critical to performance, I don't understand why you're hesitant to address them. Your comments on the New America OTI proposal (http://www.digitalsociety.org/2009/09/problems-with-new-america-foundations-transparency-standard/) also express a desire to avoid looking at oversubscription.

If you think oversubscription shouldn't be regulated, I agree with you. Oversubscription may not even need to be documented, per se, since it will be misunderstood and can be easily gamed. But it should be examined in the conversation. Perhaps the result of such a discussion is to agree on a proxy or set of proxies for network performance (such as packet loss, latency, and jitter) as it might be impacted by oversubscription. Commercial ISPs do this already in Service Level Agreements; might it be a good method for broadband providers to represent their network quality?

Re: Broadband vs. Internet Speed: Not So Fast

georgeou — Thu, 19 Aug 2010 20:35:24 -0000

“The fax analogy is not very strong. In fact, it misleads the reader to conclude that broadband access speeds on each end are all that matter.”

The fax analogy was admittedly an oversimplification, but it made a very good point about the problem with the recent FCC report and sensational media headlines. Your criticism would carry some weight if the fax analogy was all I offered, but I went into great detail as to what can cause single-download Internet performance to drop and most of those factors were not due to the ISP in any way.

I discussed the following factors.

1. Broadband service peaks substantially lower than advertised e.g, sync rates don’t even meet advertised speeds. This rarely occurs in the US (though it could be improved) but it is common in the UK. This should be blamed on the broadband provider.
2. Actual payload performance is generally 85% to 90% of sync rate performance on DSL services advertised at sync rates. While the issue is more minor, it deserves to be fixed. This issue doesn’t affect newer services like AT&T U-verse, Verizon FiOS, and cable broadband services because they advertise actual data rates.
3. Server performance is generally the most common issue. This could be due to server congestion, network congestion, or the TCP speed limit and performance penalties of latency even when the server/client/network can support much higher multi-flow TCP or UDP speeds.

I offered a much more thorough debate to broadband transparency here.
http://www.digitalsociety.org/2009/09/the-need-for-a-broadband-transparency-standard/

Re: Broadband vs. Internet Speed: Not So Fast

bensons — Thu, 19 Aug 2010 16:06:06 -0000

Since I published this yesterday, the original article “Conflating broadband speed with Internet speed is misleading“ was updated with some more information about the effect of latency on TCP. It also now includes a recommendation to use multiple streams to measure effective throughput. This is a good addition to the article, and gets closer to the truth. Still no discussion of provider oversubscription, but a step in the right direction.

Re: Cloud: Private vs Public, Internal vs External, Oh My!

Dust_Collector_Remote — Fri, 11 Jun 2010 03:16:05 -0000

Yep! I was agreed, I'll keep in touch to your blog. This blog is so usefully, Thanks for the posted ;)

Re: Network Virtual Appliances Are Silly

etherealmind — Thu, 10 Jun 2010 03:41:33 -0000

The Nexus 1000V isn't an appliance so isn't a valid comparison.. It's more like an attempt to upgrade the networking capabilities of the vSwitch and give networking some operational capability.

Re: Network Virtual Appliances Are Silly

twitter-16987758 — Wed, 09 Jun 2010 21:33:07 -0000

BTW, I guess this means Nexus 1000V is #fail? ;-)

Re: Network Virtual Appliances Are Silly

twitter-16987758 — Wed, 09 Jun 2010 20:03:27 -0000

I think these software appliances have their uses, but certainly the use case you discuss is not appropriate. At least that's true with the current generation of hypervisors and underlying hardware. I think both are maturing in ways that will eventually give us far more scalable and flexible virtualization.

Essentially the typical multi-tenant network appliance that is "hardware-based" is just running a specialized form of virtualization on specialized hardware. It may be that the specialized model will continue to be at the leading edge in terms of scalability, but I think we will start to see the gap close. Eventually the advantages of a general-purpose (or at least multi-function) platform may begin to outweigh the disadvantage posed by the remaining gap.

Re: Cloud: Private vs Public, Internal vs External, Oh My!

CiscoTech Wizard — Sat, 20 Mar 2010 18:02:05 -0000

Good arguments overall - widening the characteristics in the definiition would lad to an even more serious anarchy than muddying the waters - Internal and external clouds definition are best defined with ownership - likewise private and public clouds are also simply defined by control - the only problem is - how do we then expand the characteristics to define virtual private and virtual public clouds

Re: Knee at 6 Weeks

ForceOfNothern — Tue, 16 Mar 2010 00:53:57 -0000

I'm in town thru to friday morning. anything I can bring you from work or from anywhere else?

On Mar 15, 2010, Disqus <> wrote:

Benson Schliesser wrote, in response to ForceOfNothern:

I didn't realize I needed physical security around my workspace... I'm surprised that my posters, toys, and mounds of paper weren't enough to scare people away. :)

Regardless, thanks for your protection. Hopefully I'll be back at work soon, and I can chase people off with my crutches.

Link to comment: http://www.queuefull.net/~bensons/2010/03/05/knee-at-6-weeks/#comment-39967558

-----
Options: Respond in the body to post a reply comment.

To turn off notifications, go to: http://disqus.com/account/notifications/

</div>

Re: Knee at 6 Weeks

bensons — Mon, 15 Mar 2010 22:22:06 -0000

I didn't realize I needed physical security around my workspace... I'm surprised that my posters, toys, and mounds of paper weren't enough to scare people away. :)

Regardless, thanks for your protection. Hopefully I'll be back at work soon, and I can chase people off with my crutches.

Re: Knee at 6 Weeks

ForceOfNothern — Mon, 15 Mar 2010 14:19:18 -0000

Ouch! I hope you get better soon. we've had to evict numerous ppl trying to take over your office!

Re: Sleep Disorder, Creativity, and Productivity

derekapascualy — Wed, 17 Feb 2010 19:14:16 -0000

I think you have a lot of ideas going into your mind that is why you are not able to sleep. I used to have the same problem but I just try to empty my thoughts before sleeping and learn to relax.

Re: Ortho Follow-up, Stitches Removed

territunes — Thu, 11 Feb 2010 09:31:58 -0000

They ought to put this in a tat book. Fool 'em.

Re: Pain Killers Are Killing My Brain, Oh My!

boblozano — Tue, 09 Feb 2010 16:31:00 -0000

Ouch ... very sorry to hear this. I was only sort-of aware that you'd been hurt and caught up a bit via your blog posts.

Well, post-op pain is one thing that was not all that present after my tumor surgery. Had several (progressively stronger) meds ready to go, never took anything stronger than a migraine-grade tylenol.

An unexpected blessing, to be sure.

Hope you recover soon!

Re: Ortho Follow-up, Stitches Removed

territunes — Fri, 05 Feb 2010 09:38:47 -0000

OH, BENNY! This makes me so sad!
You are so brave. I'm thankful surgery involves less cutting
these days but, OWWW!

Re: Ortho Follow-up, Stitches Removed

bensons — Thu, 04 Feb 2010 18:08:15 -0000

No worries. i'm going to incorporate my scars into the work... not sure how, yet, but definitely planning it.

Re: Ortho Follow-up, Stitches Removed

Andrew — Thu, 04 Feb 2010 16:27:36 -0000

Thank gawd they didn't FU your "work". You would have had to start all over......