[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
20030612: data feed problems (cont.)
- Subject: 20030612: data feed problems (cont.)
- Date: Thu, 12 Jun 2003 15:23:48 -0600
>From: "Chance Eppinette" <address@hidden>
>Organization: University of Louisiana at Monroe
>Keywords: 200306111400.h5BE0HLd016841 IDD
Hi Chance,
>ULM's I2 pipe is 3Mb (386KB). The policy on our firewall for Tornado is to
>allow at most 386KB but to guarantee at least 200KB if needed. I bumped the
>guarantee up from 100KB. However on average we are only seeing about 30KB
>of average use. I did notice that around 11:30a the traffic did pickup to a
>little over 60KB, but since 1:15p it has dropped back to about 45KB.
I think that something you are doing to limit traffic to tornado is
slowing down the reception of data and causing the high latencies that
Adam/Eric are experiencing for model data. This, in turn, is actually
limiting their ingest of data to levels far below what you have set.
The datastream that is composed of the model data has up to 25000
products per hour. The sizes of these products varies from a few
hundred bytes to up to 500000 bytes with an average being around 15000
bytes. All of which translates into a lot of packets that are probably
being examined.
>I don't think the new guarantee is going to help anything, but I wanted to
>be sure we didn't restrict the traffic too much.
Is there any way that you could open your firewall to port 388 and then
monitor the activity on that port for tornado so the receipt of data
would not be slowed? If the bandwidth use exceeded your desired limit,
you could ask/demand that Adam/Eric reduce their data request. Such
limiting is easily done through user configuration of the LDM requests.
>Chance Eppinette
>Technology Support Manager
>ULM Computing Center
>address@hidden
>318-342-5021 (phone)
>318-342-5018 (fax)
Tom Yoksas