We haven't made any caching decisions as yet, but have spoken to Qwilt a couple 
of times about their product and think it looks promising. Jeff's and 
subsequent NetFlix SSL information looked like a question that needed to be 
asked.
Qwilt gave a reassuring reply and sent a link to a webex where they should be 
discussing the Netflix announcement and other caching concerns impacting 
universities.
I read the list emails- I seldom question or comment. So I don't know if it 
would be OK to toss the link out. But if anyone wants it, I can email them. And 
if it's OK to just post it here for the curious, I can do that as well.
Mary

From: The EDUCAUSE Wireless Issues Constituent Group Listserv 
[mailto:[email protected]] On Behalf Of Jeffrey Sessler
Sent: Monday, March 23, 2015 2:07 PM
To: [email protected]
Subject: Re: [WIRELESS-LAN] netflix question

 

I don't see Qwilt as a viable long-term solution. As providers move more of 
their content to be delivered over SSL, what's Qwilt's answer? Are they going 
to forge SSL certs?

 

For those using Qwilt, you're OK will all of your traffic being spanned to the 
Quilt server? Are you confident that it's not looking at, storing, or 
transmitting data it shouldn't?

 

Jeff  

>>> On Thursday, March 19, 2015 at 3:47 PM, in message 
>>> <[email protected]>, Frank Bulk <[email protected]> 
>>> wrote:

We use Qwilt, too – happy with it.   Our Netflix cache rate is 59.9%.  It’s 
just amazing how much Netflix content is commonly viewed.

 

And we move a lot more traffic than the University of Alaska. =)

 

Frank

 

From: The EDUCAUSE Wireless Issues Constituent Group Listserv 
[mailto:[email protected]] On Behalf Of Britton Anderson
Sent: Thursday, March 19, 2015 2:50 PM
To: [email protected]
Subject: Re: [WIRELESS-LAN] netflix question

 

This has come up a number of times on the NETMAN list too. I threw a question 
out a number of months ago about caching, and we of course reached out to 
Netflix in regards to acquiring an OpenConnect appliance. Even reached out to 
our ISP some time ago who we had noted were killing us with Netflix traffic 
from their OpenConnect appliance for some help, like a non-transit peer. We got 
nowhere with either.

 

We were kind of stuck and we sought our own caching solution. We went with 
Qwilt. So far I think we are one of 3 Universities in the country that have it 
running. There's an upcoming webinar if you want to learn more about it and 
feel free to reach out to me off list, but as far as the nuts and bolts go--it 
just works. 

 

We offload about 60% of Netflix traffic locally. Apple and Windows updates all 
are non-issues. The biggest thing is perceived speed. It's all transparent, so 
clients don't care where its coming from. They just watch their iOS device 
update to 8.2 in 3 minutes and say "WOW". I was in our student union building 
over lunch last week, and heard two separate conversations about how people 
have thought that the network has gotten much faster because of how fast their 
iPhones have updated. Even apps on my own phone update in a flash. But you can 
clearly see how far and wide Netflix is as the top consumer of streaming video 
for us.

 

I got an Apple TV to test with in our group and I hooked it up to my Netflix 
account and noted how absolutely smooth the playback experience was. HD is just 
ON all the time, no buffering. Fast forwarding, rewinding, to an instant play. 
Like you were watching local content...

 

The raw reports are attached. The numbers are a bit lower for the first one 
since we are now at the tail end of Spring Break, but I pulled the second one 
from the peak time of of the last week that shows the difference of quality of 
experience from content delivered locally versus from the internet.

 

Inline image 2Inline image 1

 

Long story short, we found that we had to help ourselves. I can guarantee we 
pay one of the highest rates--if not THE highest rate--for peering bandwidth in 
the nation up here. A server like this has turned out to be worth its weight in 
gold as we head into tough budget times. It will have paid for itself before 
the year is over.




 

Britton Anderson <mailto:[email protected]>  |

 Senior Network Communications Specialist |

 University of Alaska <http://www.alaska.edu/oit>  |

 907.450.8250

 

 

On Thu, Mar 19, 2015 at 11:02 AM, Lunceford, Daniel <[email protected]> 
wrote:

        
        Technically the user would also have to subscribe to the higher rate
        plan (when last I checked):
        
          SD:  $7.99/mo
          HD:  $8.99/mo
         UHD: $11.99/mo
        
        So technically, the user would have to also be a subscriber to the
        HD/UHD services which might limit your growth a bit.
        
        -drl
        
        
        --
        Dan Lunceford
        Manager of Networking Services
        New Mexico Tech
        [email protected], 575-835-5961
        
        
        -----Original Message-----
        From: The EDUCAUSE Wireless Issues Constituent Group Listserv
        [mailto:[email protected]] On Behalf Of Steve Bohrer
        Sent: Thursday, March 19, 2015 11:38 AM
        To: [email protected]
        Subject: Re: [WIRELESS-LAN] netflix question
        
        An interesting factor in Netflix (and presumably other streaming video)
        is that they will scale their display resolution based on available
        bandwidth. This can make bandwidth planning projections murky. For
        example, from the "Your Account > My Profile > Playback settings" menu
        item for my Netflix account, there are the following options:
        
        "* Auto
        * Low (basic video quality, up to 0.3 GB per hour) * Medium (standard
        video quality, up to 0.7 GB per hour) * High (best video quality, up to

        3 GB per hour for HD, 7 GB per hour for Ultra HD)"
        
        Auto is the default, and the range from 0.3 GB per hour to 7 GB per hour
        is a factor of about 23.
        
        SO, if most of my users are currently getting "Medium" quality at peak
        demand times, I could double or quadruple my available bandwidth, and,
        even if user demand were completely unchanged, all the existing Netflix
        flows could expand to soak up all of the bandwidth increase.
        
        As a rule of thumb for planning, we been assuming "bandwidth demand will
        double about every year and a half to two years." In fact, however,
        Netflix demand can scale up by an order of magnitude with absolutely no
        change in user behavior.
        
        (Presently, we are small and constrained enough that we run an Allot
        NetEnforcer "packet shaper" at our edge, and streaming video gets a
        lower priority than general HTML traffic. Even so, during evening prime
        use periods, Netflix and other streaming video are generally 50% or more
        of our total inbound traffic.)
        
        Steve Bohrer
        Network Admin, ITS
        Bard College at Simon's Rock
        413-528-7645
        
        **********
        Participation and subscription information for this EDUCAUSE Constituent
        Group discussion list can be found at http://www.educause.edu/groups/.
        
        **********
        Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/groups/.

 

********** Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/groups/. 

********** Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/groups/. 

********** Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/groups/. 

Reply via email to