dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
1259
share rss forum feed

DrStrangLov

join:2012-03-28

[Exede] Discussion: Current Status Wednesday January 23, 2013

.
.
RE: »Re: Current Status Wednesday January 23, 2013
.
.

said by wm4bama:

You have proven that beam congestion is still a problem with Legacy customers sharing bandwidth with Exede-5 customers.

Negative...upload speed shows no beam congestion, and this data shows that the screwy little rabbit has been ex·or·cised. There is no data to support or refute a conclusion on beam congestion.
»Upload :: 12 Kbps 2 kB/s

Phoenix server would be main bandwidth controller, and each Gateway serviced would have a maximum throughput; but in theory, this server's NMP machine can be programmed in many different ways.

For instance, during load periods, three Exede-5 beams at Riverside could have a fixed throughput (or it could be dynamic, as based upon each beams dynamic loading), and those Exede-12 beams could be set to have proportionally same bandwidth during loading.

The leased fiber lines to each Gateway most likely have a fixed pipe sized 24x7, but I would assume Phoenix's server does not have a Big enough Pipe size equal to all Gateway pipe-sizes, but rather Phoenix has a smaller sized pipe. Translation, when dynamic loading happens, NMP machine may have been programmed in recent past to clamp down more on at least my Exede-5 beam....going from about 12Kbps to 2.1Mbps uploads shows a quantum leap in performance gain, although I experienced these same upload speeds in earlier days during primetimes.

And, btw, on my beam, about 1/2 of it is covered with 3 Exede-12's beams, so those leaving this legacy beam ain't going to signup for Exede-5...trust me

I know that Exede-5 Anik-F2:62 is fully loaded in NW US; but WB-1:163 and Anik-F2:81 are not. I'm not aware of other active Exede-5 beams except Anik-F2:73. In any event, only NW US beam is where complaints exist that I've heard publicly on forums.


wm4bama

join:2012-05-10
Goodwater, AL

Re: [Exede] Discussion: Current Status Wednesday January 23, 201

You missed my point...upload speeds were never a problem on Legacy even during the peak-use hours I got full, or greater than full speeds for upload...there's just not that much of traffic on the upload bandwidth.

It's the download bandwidth that's suffering, and will probably continue to suffer from 5-6pm until after 11pm...that was standard procedure for my Legacy account. Upload speeds were never slowed.
--
Exede12, ViaSat-1, beam 342, Albuquerque Gateway, Denver AcceleNet servers


DrStrangLov

join:2012-03-28

said by wm4bama:

...download bandwidth that's suffering, and will probably continue to suffer from 5-6pm until after 11pm...

Each legacy Gateway had a fixed amount of bandwidth, and NMP policies were designed around this paradigm.

With AcceleNet Servers, specifically Phoenix's AcceleNet Server, where time-zones, along with two different Gateway types (5/12), if you have a limited Internet pipe size, network folks will clamp down and determine who and how much a given customer (5/12) gets. After all, shouldn't beam 358 get the most bandwidth, with it being Wildblue's "Home Beam?"

Exede will add bandwidth when they perceive it is needed, and network folks turn the NMP dials as needed. This has been an industry practice, for years; Cisco's "demons" make it so easy.

In any event, I'm not complaining about download speed...but, I do know the NMP-demon is byting away on Testmy's progress-time-line. I've seen Testmy bar stop, stop, and stop, which means others are being fed the bytes...when it resumes, it can fly.

Spice300
Premium
join:2006-01-10
reply to DrStrangLov

The Network Management Policy still activates on my spot beam sometimes during the day and evening indicating that congestion is still present. However, prime time slowdown is not present every day like it was for 2.5 years prior to Exede.
--
Wildblue Value Pack, beam 31, Riverside gateway


DrStrangLov

join:2012-03-28

said by Spice300:

The Network Management Policy

Actions are based upon:

1. A beam's total bandwidth

2. Bandwidth bought from Tier One Provider

ISPs since early 90s and onward tended to hold off on buying bigger pipes until they perceived it was warranted. You may remember, there were 16 or 32 dial-in lines using a 56Kbps internet pipe. But, back then, NMP demons weren't around, so all users shared equally.

At Riverside Gateway for Anik-F2 users, its rather doubtful maximum bandwidth was bought for every beam when a NMP demon could equalize loading across all beams with a smaller pipe. WE are talking dollars/cents here, big dollars/cents that bean counters cherish, and who get bonuses for being a scrooge. Hence, NMP demon can reduce your total pipe size and server loading via prioritizing traffic; and of course, this reduces your total economic spending.

Hence, in some states, a beam's total bandwidth may never be utilized, and when you have a number of beams at a gateway, they can "share" total bandwidth bought with all users equally via NMP demon prioritizing traffic.

That said, ViaSat's CEO has stated he wants users to have good experiences...so some of the previous practices are being phased out whenever possible.

silbaco
Premium
join:2009-08-03
USA

1 edit
reply to DrStrangLov

I highly doubt the gateways don't have enough bandwidth from the Tier-1's. When a company gets as big as ViaSat, the cost of buying more backhaul from a Tier-1 is pocket change.

WB-1 has a finite amount of capacity ~10Gbps. A certain amount of that is allocated to upload regardless of if it is used. There is likely some more allocated to special purposes. That leaves the remaining for serving customers with downstream traffic. Network management policies or not, that's not much bandwidth. The satellite has probably simply reached it's capacity.


DrStrangLov

join:2012-03-28

1 edit

said by silbaco:

When a company gets as big as ViaSat, the cost of buying more backhaul from a Tier-1 is pocket change.

Ballpark OC-48 rate from one provider, which is enough for one legacy Gateway, runs around $59,999.00/month.
»www.usaccess-llc.com/gige.html

Using a Cisco demon would pay for itself before end of first year of usage....no brainer...and money talks!

silbaco
Premium
join:2009-08-03
USA

$60k/month is pocket change to a company growing as fast as ViaSat, projected to make well of $1bil this fiscal year. However keep in mind the more bandwidth you buy, the cheaper it becomes so we have no way of knowing the exact cost of their bandwidth needs. However an OC-48 wouldn't suffice for bandwidth needed for ViaSat-1.