ie8 fix
madison

Networking

Steven J. Vaughan-Nichols

Changing DNS probably won’t help your Video Streaming

By | December 22, 2010, 1:12pm PST

Summary: Despite recent claims that universal DNS services like OpenDNS, DNS Advantage, and Google DNS can slow down Apple TV video-streaming, the truth is that they don’t for most users.

When I first read that some Apple TV users were seeing significant speed-ups when they start using a local ISP Domain Name System (DNS) server instead of continuing to use one of the universal DNS services, such as OpenDNS, DNS Advantage,or Google Public DNS, my first thought was, “That’s wrong.”

I understand their logic that “When millions of users all tap into the same DNS server addresses to resolve domain names, as Google DNS does by design, Akamai and other CDNs [Content Delivery Networks] route content to those users along the same path, preventing the network from working optimally.” The problem is that this isn’t really how the big DNS networks and CDNs work these days.

For starters, this proposed fix starts with the notion that your ISP has a local DNS, hence you’ll get a better, less-crowded route for your video. You probably don’t have a truly local DNS though. The national ISPs like Comcast. Verizon, or ATT, just like the universal DNS services, spread their DNS servers around. In this case, their DNS server isn’t going to be much ‘closer,’ in terms of network distance than Google’s.

Even if you do have a true local ISP, they may not have a local DNS. Many ISPs outsource their DNS services to DNS providers like Dynect, DNS Made Easy, or DynDNS. Still other ISPs are now using Google DNS. So, switching to your local ISP DNS may not make any difference, since its DNS server isn’t actually ‘local.’

It probably won’t matter anyway to most users though because as David Ulevitch, OpenDNS’ founder and CEO explained to me:

Many CDNs (including Akamai) have lots of tricks to do geo-targeting of users. Using DNS to target users is one of the most common ones:

When a user makes a DNS request, which happens before they make an HTTP request (for data) the path usually looks like this:

User –> ISP Recursive DNS Server –> CDN’s DNS Server

With OpenDNS, it’s:

User –> OpenDNS –> CDN’s DNS Server

At the DNS level, CDNs don’t see the User IP address, they only see the recursive DNS server asking it questions. CDNs often assume that the location of the DNS server is “near” the location of the User and so they give back a response that is near the DNS server as opposed to near the User. When the User is really far from their DNS server or on another network, this can provide suboptimal results.

But, that’s not all there is to it.

There are very good solutions for this, many of which are in place today with large CDNs, including this IETF [Internet Engineering Task Force] proposal: Client IP information in DNS requests.

We do something very similar to that IETF draft proposal with folks today, and it works beautifully, actually providing an even BETTER response than just using the location of the recursive DNS server. So if more people would support proposals like that, things wouldn’t just be the same, they would be even better from a performance standpoint.

The IETF and OpenDNS approach work better by giving the CDN or Web site enough information about the user’s actual location to pick out the best route for the video stream. Google, which is working on the IETF proposed standard, tries currently to get around the DNS location problem by hosted “Google Public DNS in data centers worldwide, and uses anycast routing to send users to the geographically closest data center.”

So, in the current Google system, “If a content provider hosts mirrored sites around the world, that provider’s name-servers will return the IP address in closest proximity to the DNS resolver. … however, that because name-servers geolocate according to the resolver’s IP address rather than the user’s, Google Public DNS has the same limitations as other open DNS services: that is, the server to which a user is referred might be farther away than one to which a local DNS provider would have referred. This could cause a slower browsing experience for certain sites.”

Page 2: [The Rest of the DNS & Streaming Story] »

Topics

Steven J. Vaughan-Nichols, aka sjvn, has been writing about technology and the business of technology since CP/M-80 was the cutting edge, PC operating system

Disclosure

Steven J. Vaughan-Nichols

Steven J. Vaughan-Nichols is a freelance writer. He does not own stocks or other investments in any technology company.

Biography

Steven J. Vaughan-Nichols

Steven J. Vaughan-Nichols, aka sjvn, has been writing about technology and the business of technology since CP/M-80 was the cutting edge, PC operating system; 300bps was a fast Internet connection; WordStar was the state of the art word processor; and we liked it.

His work has been published in everything from highly technical publications (IEEE Computer, ACM NetWorker, Byte) to business publications (eWEEK, InformationWeek, ZDNet) to popular technology (Computer Shopper, PC Magazine, PC World) to the mainstream press (Washington Post, San Francisco Chronicle, BusinessWeek).

15
Comments

Join the conversation!

Just In

RE: Changing DNS probably won't help your Video Streaming
JACOBSONR 14th Oct
Good day to confirm this comment I would appreciate T h e b e s t o f Z D N e t d e l i v e r e d your website very nice to everyone Yes, Oracle is the only one with shared-disk architecture, but that is there advantage. It means you can add or remove nodes and the database lives on. In a shared nothing architecture, if you lose a node, you lose the system. I'm sure Oracle appreciates EMC highlighting their advantage.I also desire to signal in your RSS feeds. Thank you as soon as once again and maintain up the great operate Awesome post! Thank you very much || thanks for nice content this is really benefit to me.
what is exactly the solution in short... its blabbering words.. with no shot right through my eyes
0 Votes
+ -
Contributr
@draugnavz That's a story for the next time. But, for staters, get the fastest connection you can get; optimize your network; check your real Internet speed:

http://www.speedtest.net/

and, check just how good that connection really is no matter how many Mbps:

http://www.pingtest.net/

See #s that don't jive with the bill of goods your ISP sold you? Yell.

Steven
0 Votes
+ -
Overused solution
jscott418 23rd Dec 2010
I too think the Apple forums overuse the DNS change for a solution. Its true that some issues occurred with some ISP DNS servers and Safari's prefetching which some DNS servers interpreted as denial of service requests. Or it just could not handle all the requests. But I have found little else to conclude that changing DNS will help speed. In fact in my own experience with Comcast. They actually had better ping times then Open DNS. I suspect shorter series of jumps then with Open DNS was the reason. If anything your ISP or just simply network traffic is too blame. Which may or may not be fixable.
0 Votes
+ -
Not choosing the ROUTE!
rjcarlson49 23rd Dec 2010
Please stop saying the DNS is choosing the route! It is choosing the IP address to return, or you could say it is choosing the server to resolve you domain name to. The route is then a result of IP routing between your node and the server. I know you know this. It's obviously just lazy (or maybe pandering to naive readers.)

BTW, thanks for the pointer to DNSBenchmark. Very nicely done tool.
@rjcarlson49

Correct, DNS is at the application layer of the OSI model and has nothing to do with ip routes. IP routes is the routing protocol, residing on a much lower layer.

DNS stack:
Computer will ask
is it me?
Then it will check the host file to see if it has a entry for said name.
If none of those are successful it forwards the packet on to its DNS.

The process will continue and repeat until it finds the address, or goes to the Root servers.

Root servers are asked what DNS registry is responsible for domain name and is either redirected to said registry or killed.

After name is resolved the routing protocol can choose the best route.

It is apparent that the author Steven J. Vaughan-Nichols has spent too much time writing about technology and not enough time refreshing his knowledge and catching up to todays tech.
0 Votes
+ -
I noticed my throughput when hitting Apple was molasses
betelgeuse68 Updated - 2nd Jan 2011
I noticed my throughput when hitting Apple for downloads was absolute molasses. I had been wondering for weeks why. I thought maybe Apple was experiencing high loads due to their crazy growth.

But it turns out I had setup my primary DNS server to be Google's and my secondary DNS server to be OpenDNS'.

Then yesterday I saw Slashdot mentioning that the use of Google & OpenDNS' servers was causing slow transfer speeds against Apple's services (which I had been experiencing). Now this writeup.

Anyway, there's something definitely going on. Being more than capable of devising a solution (which I haven't worked on yet), I did just that (in my head).

The Solution:

Sniff the hostnames iTunes is trying to resolve, look up those values against your provider's DNS server (Comcast in my case), then explicitly stick those values in your computer's "hosts" files:

c:\Windows\System32\drivers\etc\hosts (Windows)
/etc/hosts (*NIX)

To make it all complete run a script every so often and re-query your provider's DNS server and replace the previous reply with the latest good reply (in the hosts file).

Easy.

For those of you who have no idea what I'm talking about... you'll have to find a layman's solution. The stuff I'm talking about involves command line tools and scripting.

-M
Better DNS may make sufing static pages faster, due to faster lookups - but yeah, it's not gonna make streaming any faster. DNS just translates names to numbers - and only needs to be done one time for the most part.

Once the lookup is done, the media streams over whatever some other protocols negotiated. If they're using DNS to route, they're frankly doing it wrong - other protocols are better for routing.

I'd really want a fact check on the idea that they're not using something like RIP or BGP to route the packets. This sounds suspiciously like somebody is simply misunderstanding how routing works.
I also desire to signal in your RSS feeds. Thank you as soon as once again and maintain up the great operate! nccma cooler
I used to be more than happy to seek out this internet-site.I wanted to thanks in your time for this glorious read!! I positively enjoying each little bit of it and I have you bookmarked to check out new stuff you weblog post. this thread is amazing i like your work and i appreciate you that you have share a useful stuff thanks for sharing the i shop abatwa
I used to be more than happy to seek out this internet-site.I wanted to thanks in your time for this glorious read!! I positively enjoying each little bit of it and I have you bookmarked to check out new stuff you weblog post.Bookmarking now thanks please consider a follow up post. power sa shop
I think the representation of this article is actually superb one. This is my first visit to your site. Thanks a lot and keep sharing the information. Keep updating the information for all of us. Thanks ZDNet Government was launched as the brand's first industry vertical, with a mission to cater to IT professionals in the public secto I agree with your post. However, do you have any sources I can cite for my paper wheel car com bury
Well welcome, hopefully you can become a vital member of the community and really help to push far ahead of google. Which Im sure the development team would love. This will of course earn you alot points too and get you on the leaders board. z d n e t t h a n k Im not sure i come to an agreement with you on every level, howevor it absolutely was a good posting, many thanks for taking the time to put up your ideas.
Thanks nice info z d n e t I really liked your current article write more..let me add you to its favorite The articles you have on zdnet s i t e are always so enjoyable to read. Good work and I bookmarked it.
Fantastic news about the new release.I positively enjoying each little bit of it and I have you b o o k m a r k e d to check out new stuff you weblog post.Im not sure i come to an agreement with you on every level, howevor it absolutely was a good posting, many thanks for taking the time to put up your ideas
Good day to confirm this comment I would appreciate T h e b e s t o f Z D N e t d e l i v e r e d your website very nice to everyone Yes, Oracle is the only one with shared-disk architecture, but that is there advantage. It means you can add or remove nodes and the database lives on. In a shared nothing architecture, if you lose a node, you lose the system. I'm sure Oracle appreciates EMC highlighting their advantage.I also desire to signal in your RSS feeds. Thank you as soon as once again and maintain up the great operate Awesome post! Thank you very much || thanks for nice content this is really benefit to me.

Join the conversation!

Formatting +
BB Codes - Note: HTML is not supported in forums
  • [b] Bold [/b]
  • [i] Italic [/i]
  • [u] Underline [/u]
  • [s] Strikethrough [/s]
  • [q] "Quote" [/q]
  • [ol][*] 1. Ordered List [/ol]
  • [ul][*] · Unordered List [/ul]
  • [pre] Preformat [/pre]
  • [quote] "Blockquote" [/quote]
ie8 fix
Click Here
ie8 fix

The best of ZDNet, delivered

ZDNet Newsletters

Get the best of ZDNet delivered straight to your inbox

Facebook Activity

White Papers, Webcasts, & Resources
ie8 fix
ie8 fix