Re: [Bloat] Flent-farm costs

Re: [Bloat] Flent-farm costs

[Rich Brown]

Dave wrote:

> Costs on the "flent-farm" continue to drop. Our earliest linode
> servers cost $20/month and our two latest ones (nanoservers) cost
> $5/month. For "science!" I've been generally unwilling to
> update/change these much ...

I've been running netperf.bufferbloat.net (the netperf server that we most publicize) for several years. It's a modest OpenVZ VPS from RamNode in Atlanta. It has two failings:

- It costs ~$16/month (I don't mind this expense, but $16/month >> $5/month for the nanoservers)
- About ever third month, its traffic goes over the 4TB/month limit, and RamNode shuts the server off. I regularly run a script to find heavy users and block their IP using iptables. (Many people are running a test every five minutes for days at a time.) But that's a hassle. And buying an additional terabyte per month from RamNode is $10/month, which gets expensive.

To address this, I stood up up a new (KVM-based) VPS with RamNode (also in Atlanta, presumably in their same data center) that will permit more in-depth iptables rules. My goal would be to look at connection frequency, and if someone is trying to do every-five-minute testing, limit their bandwidth to 10kbps. 

This raises a host of questions:

- For Science - the current netperf.bufferbloat.net is atl.richb-hanover.com; the new server is atl2.richb-hanover.com. Do you get similar performance from both servers?

- Is this plan to bandwidth-limit abusers realistic? Will it be possible to design rules that exclude abusers while allowing legitimate research use? (I'm concerned that running five tests in a row in 10 minutes might look like an every-five-minute abuser...)

- Should I use one of the Linode nanoservers?

- Should we move the netperf.bufferbloat.net name to use the existing flent server farm machines?

- Are there other approaches to supporting netperf.bufferbloat.net?

Many thanks!

Rich

Re: [Bloat] Flent-farm costs

[Dave Taht]

Given that donations cracked 160/month (thanks everyone), I would like
to refund yours, and possibly netperf.bufferbloat.net into a linode server in
atlanta. Not a nanoserver since it is loaded more than
anything else.

However I'd like to add you to the admin list (pls send ssh public key?)
and also borrow your iptables scripts!

As for the TB/month limit. I *think* our linode cluster is the sum total
of all linodes. We rarely crack 10% bandwidth usage/month currently.

Today's stats:

371GB Used, 17629GB Remaining, 18000GB Quota

As "for science!" I am currently unaware of anyone actually doing
science against any of the bufferbloat.net flent servers presently.

That said, I've always thought having stuff in "other clouds" made
sense, it's just that figuring out how to "use" and especially
*maintain* other clouds - like aws and google's that's always been a
problem.

anyway I booted a new $20 dual-cpu linode in atlanta just now. As I've lost
track of how to update bufferbloat.net (toke?) it is temporarily up at
flent-atlanta.taht.net. (or dns should propigate in an hour or or so).

It's such a joy to type a few apt commands and be "done" setting up a
theoretically bloat free node..... I'll do a few tests over the course
of the week. I have to admit the inital tests I ran were kind of
disappointing, losing for example a ton of udp. some of the "science" we
should have been doing was tracking what our cloudy provider was doing
underneath us and on the path!

http://flent-atlanta.taht.net/flent
from that node to yours:
                              avg       median          # data pts
 Ping (ms) ICMP   :        10.52        11.60 ms              347
 Ping (ms) UDP BE :       141.51         0.53 ms               40
 Ping (ms) UDP BK :       134.33         0.53 ms               25
 Ping (ms) UDP EF :       150.47         6.99 ms               70
 Ping (ms) avg    :       109.21        11.49 ms              350
 TCP download BE  :       221.67       221.47 Mbits/s         299
 TCP download BK  :       234.83       232.08 Mbits/s         299
 TCP download CS5 :       228.35       220.72 Mbits/s         299
 TCP download EF  :       194.07       193.26 Mbits/s         299
 TCP download avg :       219.73       224.68 Mbits/s         301
 TCP download sum :       878.92       898.15 Mbits/s         301
 TCP totals       :      1180.11      1189.00 Mbits/s         301
 TCP upload BE    :        83.56        80.19 Mbits/s         266
 TCP upload BK    :        80.84        79.49 Mbits/s         267
 TCP upload CS5   :        81.78        81.67 Mbits/s         267
 TCP upload EF    :        55.01        54.85 Mbits/s         263
 TCP upload avg   :        75.30        75.53 Mbits/s         301
 TCP upload sum   :       301.19       301.11 Mbits/s         301



Rich Brown <richb.hanover@gmail.com> writes:

> Dave wrote:
>
>> Costs on the "flent-farm" continue to drop. Our earliest linode
>> servers cost $20/month and our two latest ones (nanoservers) cost
>> $5/month. For "science!" I've been generally unwilling to
>> update/change these much ...
>
> I've been running netperf.bufferbloat.net (the netperf server that we
> most publicize) for several years. It's a modest OpenVZ VPS from
> RamNode in Atlanta. It has two failings:
>
> - It costs ~$16/month (I don't mind this expense, but $16/month >>
> $5/month for the nanoservers)
> - About ever third month, its traffic goes over the 4TB/month limit,
> and RamNode shuts the server off. I regularly run a script to find
> heavy users and block their IP using iptables. (Many people are
> running a test every five minutes for days at a time.) But that's a
> hassle. And buying an additional terabyte per month from RamNode is
> $10/month, which gets expensive.
>
> To address this, I stood up up a new (KVM-based) VPS with RamNode
> (also in Atlanta, presumably in their same data center) that will
> permit more in-depth iptables rules. My goal would be to look at
> connection frequency, and if someone is trying to do every-five-minute
> testing, limit their bandwidth to 10kbps.
>
> This raises a host of questions:
>
> - For Science - the current netperf.bufferbloat.net is
> atl.richb-hanover.com; the new server is atl2.richb-hanover.com. Do
> you get similar performance from both servers?
>
> - Is this plan to bandwidth-limit abusers realistic? Will it be
> possible to design rules that exclude abusers while allowing
> legitimate research use? (I'm concerned that running five tests in a
> row in 10 minutes might look like an every-five-minute abuser...)
>
> - Should I use one of the Linode nanoservers?
>
> - Should we move the netperf.bufferbloat.net name to use the existing
> flent server farm machines?
>
> - Are there other approaches to supporting netperf.bufferbloat.net?
>
> Many thanks!
>
> Rich
>
>
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat