Search This Blog

Wednesday, August 13, 2014

Just some dumb broadband

Part two on my ranting about the staff report on the E-Rate Modernization Order: External Connections.

First, I like using "External connections" instead of "Category 1."  Could we make that official jargon?

The definitions in paragraph 15 don't really describe the situation in most districts.  The definitions make it appear that the purpose of a WAN is to deliver data to the Internet.  Even with so many services moving to the cloud,  most WAN traffic is not bound for the Internet, especially in districts with VoIP internally.  Also, many applicants will have at least two aggregation points.

The data about fiber starting in paragraph 17 is confusing need with ability.  The numbers don't seem right to me, but maybe they're close.  Even if they're accurate, they don't indicate a problem.  Only 15% of libraries have fiber?  Maybe.  But do they need fiber?  I'll use my local library as an example.  For $159/month, they could get 100 Mbps over fiber to their building, since Verizon has FiOS fiber running right in front of the library.  Instead, they get 100 Mbps from the cable company for $139/month.  Would it really improve the program to have them pay an extra $20/month for the same bandwidth, just so they could say they have fiber?  Don't talk to me about scaling: FiOS is not part of Verizon Business Global (which does connections of 1 Gbps and greater), so if the library decides to go to 1 Gbps, it's going to mean pulling a new fiber, no matter which company they're using for their current 100 Mbps connection.

"Based on this analysis, we estimate that only 40 percent of schools will purchase at least 1
Mbps per student of last-mile bandwidth in Funding Year 2014, and only 10 percent of schools will purchase the 10 Mbps per student of last-mile bandwidth needed to achieve the SETDA last-mile goal....  EducationSuperHighway has found ... that over 90 percent of schools are currently below the 1 Mbps per student five-year goal."  And based on my analysis, only 1% of schools will actually use 1 Mbps/student in 2014, and 0% will use 10 Mbps/student.  SETDA just pulled some numbers out of the air.  I can't argue with their guess on the need five years from now, but I can say that at present, using 1 Mbps/student as a standard is just promoting waste.

Those numbers in paragraph 32 look pretty conclusive for the power of consortium purchasing.  Network Nebraska dropped member costs from $87 to $1.28 per megabit per second per month.  A savings of 99.985%.  Oh wait, if you read the actual submission, that's over 8 years.  

For comparison, I'll pick a local district here with about 1,000 students.  In 2006, they were spending $568.65/Mbps/month (breathtaking when you look back on those T-1 prices, isn't it?).  In 2014, they're paying $24/Mbps/month.  A savings of 99.958%.  Or they could go to a 1 Gbps connection for $3.90, for a savings of 99.993%; that 1 Gbps connection would make everyone happy, from President Obama on down, but 90% of the bandwidth would be unused, and it would be a waste of $16,200/year in E-Rate funding.

By the way, that district has access to a purchasing cooperative.  If they joined that consortium and contracted with the same service provider, their costs would go up from $24/Mbps/month to over $34/Mbps/month.  Consortium purchasing would drive their cost up 43%.  (The price for 1 Gbps is 173% higher through the consortium.)  Same vendor, same service.

When bandwidth prices have dropped 99% over the last 8 years, showing that your consortium has cut prices by 99% is not demonstrating success.  And Mississippi definitely shouldn't be bragging about lowering costs only 90% in 9 years.

[Side note: Network Nebraska joins Utah Education Network on my list of consortia that are no longer eligible for E-Rate funding under the new definition of "consortium"; they have private universities and museums in their consortium.]

So my advice to E-Rate applicants: ignore the 1 Mbps/student benchmark.  And take a close look at whether that consortium is really going to save you money.


  1. Dear Dan:

    If you're interested, here’s another metric from Nebraska about consortium memberships and aggregated purchasing power.

    Network Nebraska K-12 currently purchases 12.5Gbps of Internet at $1.2844/Mbps/month.

    The only school districts (14) NOT on Network Nebraska this year are purchasing 3Gbps of Internet at $4.00/Mbps/month (the best independently negotiated rate in Omaha, NE, our largest metro area).

    Had they joined Network Nebraska, the difference in potential cost savings ($8,146.80/month) would completely pay for all of their districts’ Network Nebraska consortium memberships, AND leave an additional $3,662/month ($43,944/year) of cost avoidance as a benefit of belonging to the statewide consortium, which is voluntary.

    OH, and yes, we do cost-allocate for public and private higher education on our backbone, and higher education entities are billed for their Internet separately, although their Internet demand is aggregated with K-12 when determining the overall statewide unit cost.

    Tom Rolfes, Education I.T. Manager, State of Nebraska

  2. Tom,
    Thanks for the info. I think you've provided much better evidence for the power of bulk purchasing. It's clear that Network Nebraska is getting much better pricing on their Internet port than are those districts not in the consortium. Just to be clear, I am correct in understanding that in addition to the $1.28/Mbps fee for Internet, members have to pay an additional monthly fee for each location for "interregional transport" and must contract individually for a circuit to connect to Network Nebraska. Is that right? I'd be interested in a cost comparison that included those costs as well. As I've said in the past on this blog, aggregating locations will probably get a lower price on the Internet port, but transport costs will tend to be higher.

    Have you heard from the FCC and USAC that they will continue to allow consortia to cost-allocate the costs of non-public members? That's been the practice in the past, but the new definition of consortium says, "Eligible schools and libraries may not join consortia with ineligible private sector members...." USAC always ignored the similar prohibition that was in the instructions for Item 5a on the Form 471, but now that it's in the code....

    Again, thanks for providing better evidence of Network Nebraska's success.


  3. Dan,
    You are correct. A school district must reach a Network Nebraska aggregation point or an intermediate Educational Service Unit (ESU) sub-aggregation point with their own WAN circuit paid for out of their own pocket after E-rate. This is true in either Internet situation described earlier. So, the "raw" Internet of $1.2844/Mbps/month can be equitably compared to their higher rate of $4.00/Mbps/month because it is an egress from the same ESU address. The total Participation Fee ($218.30/month/entity) and Interregional Transport Fee ($18.49/entity/month) x 14 entities x 12 months PLUS all their Network Nebraska Internet costs is still about $44,000 below what these entities would pay just for Internet if they stay independent of the statewide consortium.

    As for ineligible private sector members belonging to consortia (1997), at least one previous waiver precedent (2000) has been granted by the FCC. With the new emphasis on consortium bulk purchasing, I agree that it is a rule worth re-examining.

  4. Tom,
    Great info, thanks!

    On private sector members of a consortium, I think I found the decision you're referring to:
    Thanks for that, too: paragraph 12 includes a good description of why the private entities aren't allowed in consortia.

    Based on that decision, it seems like every consortium with a private sector member would need an FCC waiver to be funded. The FCC should create a blanket waiver for private colleges and universities.

  5. Dan:
    I don't disagree about the need for a blanket waiver. It makes sense to me.

    One not-so-subtle difference seldom talked about between school districts involved in statewide networks and school districts with "Internet delivered" by an ISP is that statewide networks employ a dynamic routing protocol at their core routers so that every packet of data (e.g. e-mail, videoconferencing, data submission to depts of education, Internet2 traffic, etc...) reaches its destination in the quickest, most efficient pathway, and only traverses commodity Internet when it is the fastest pathway. This is in stark contrast to a school district that relies on commodity Internet for 100% of its data communications. So, as you can see, besides the cost advantages of belonging to statewide network consortiums, there is also likely to be an Internet bandwidth avoidance measure that may constitute between 25-40% of a school district's daily data exchange, never reaching the commodity Internet, because there was a faster Intranet or backbone pathway available. Make sense?