[EHPweb] Akamai billing algorithm

[Robert Simpson]

Hi Lisa,

Here are some definitions from the Akamai website that clarify (a  
little) the 95/5 scheme.

Bob




On Jun 17, 2008, at 12:37 PM, Wald Lisa wrote:

> Web Team et al.-
>
> This email is just for your information.  If you aren't interested  
> in the website traffic and how Akamai bills us for it, you can  
> delete this.
>
> Lorna Schmid, Judy Ferrier, and I had a conference call with Todd  
> Aikers at Akamai list week to learn how Akamai bills us for services  
> and what our options are for the future.  This is a summary of what  
> we learned.
>
> AKAMAI FEE ALGORITHM
> The annual contract with Akamai is 420K.  The Earthquake Hazards  
> Program has been paying 120K of the bill, while E-web has been  
> paying 300K.  The contract covers:
>
> www.usgs.gov
> earthquake.usgs.gov
> quake.wr.usgs.gov
> pasadena.wr.usgs.gov
> neic.usgs.gov
>
> We basically pay for bandwidth per month.  We are allotted 30mbits/ 
> sec average per month using something called the 95/5 algorithm  
> which has something to do with 95%.  In addition we have 400mbits  
> for the entire year that we can draw from if we go over the 30mbits/ 
> sec in any month (termed "bursting").  We are not charged for super- 
> bursting until we reach 4 times our monthly allotment (30 mbit/sec x  
> 4) which is 120 mbits/sec.  At that point we're charged $0.02 per  
> mbyte over the 120mbits.  Note that the units are different, and we  
> don't know exactly how this part works yet.  In any case, we have  
> not had any superbursting charges since this algorithm was put into  
> place (or we should not have had any, in any case, although Lorna  
> thinks we may have paid for superbursting when we weren't supposed  
> to in the past).
>
> EHP AKAMAI USAGE
> In February of this year, our Akamai usage went over 30mbits/sec for  
> the first time; it was 40+ in Feb, 50+ in April, and almost 70 in  
> May.  Most of the increase was from the EHP website, but we're still  
> trying to determine if it is due to simply increasing traffic due to  
> recent earthquakes, or if there is a new or larger file that has  
> caused the significant increase.  Since we've been making alot of  
> changes to configurations on the Akamai end and on our end, it's  
> difficult to pinpoint the cause.  The most recent config change was  
> made at the beginning of June, so we will have to wait until the end  
> of this month to see the significance of the latest change.
>
> BOTTOM LINE
> E-web wants the Earthquake Hazards Program to pay for the percentage  
> of Akamai services that we're using when the contract is renewed (or  
> re-bid) soon, which will be significantly more than 120K, since we  
> use approximately 90% of the service.  No matter who ends up paying  
> for the service, we need to determine the cause of our rising  
> bandwidth and configure our Akamai usage to minimize this number.   
> However, if the increase is due simply to increasing web traffic,  
> there is really nothing we can do aside from optimizing our file  
> sizes, which we have done with some files, and continue to  
> investigate for others.
>
> I am currently do some log analysis with our web traffic stats, but  
> nothing is jumping out at me as the single cause of the increasing  
> bandwidth.  I'll be looking into "last mile compression" this week  
> which could buy us a huge savings all the way around.
>
> - Lisa
> ------
>
>
>
>
> _______________________________________________
> EHPweb mailing list
> EHPweb at geohazards.usgs.gov
> https://geohazards.usgs.gov/mailman/listinfo/ehpweb

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geohazards.usgs.gov/pipermail/ehpweb/attachments/20080617/f799a828/attachment-0002.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Akamai_definitions.rtf
Type: text/rtf
Size: 9135 bytes
Desc: not available
Url : http://geohazards.usgs.gov/pipermail/ehpweb/attachments/20080617/f799a828/attachment-0001.rtf 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geohazards.usgs.gov/pipermail/ehpweb/attachments/20080617/f799a828/attachment-0003.html