Netflow engine 5.0.2 and sizing server

[j.canastero]

I installed the NFC version recently, in order to collect the flows of around 160 routers, and I had to create near 884 filters and aggregators. My question is: if I have a Solaris server with 2 processors of 400Mhz and RAM 2 GB, it is possible that the equipment has interruptions or stop the services

When I enter the option of reports, is too slow.

What are the memory values that would be the optimal ones to setup in the file /config/nfcmem?

What other parameter is necessary to review of OS and the application netflow?

thanks in advance

J.Canastero

[didyap]

Exporting a large number of flows can degrade the router and Netflow Collector performance. To decrease the load, refer to http://www.cisco.com/en/US/products/sw/iosswrel/ps5207/products_feature_guide09186a00801a7618.html

To aggregate the flows, refer to http://www.cisco.com/en/US/products/sw/netmgtsw/ps1964/products_implementation_design_guide09186a00800d6a11.html#wp1031105

[Jan Nejman]

Dear Mr. Canastero,

can you more closely specify what amount of traffic is monitored or how many flows generate

your routers... Our company developed a Linux based Netflow monitoring software. We recommended 3Ghz Pentium 4 with 1-2GB RAM for processing 800 mil flows/day. (number of flows depends on network traffic character and traffic volume).

Jan

Caligare Co.

http://www.caligare.com

[j.canastero]

Hello,

We are collecting around 1720000 flows/day

Jesus Canastero

[Jan Nejman]

Hello,

I think that your processor cannot be utilized by netflow traffic. Try to find which process uses your CPU or disk.

Jan

Caligare Co.

http://www.caligare.com

[j.canastero]

hello Jan

Using prstat on solaris the nfc process use almost 50% from CPU0, and I have configured the nfcmem file as follows

collector min=640M max=768m

Webmem min=256 max=512

cnsxml min 256 max=512

genreport min=256 max=512

re_mem min=256 max=512

UDP_READER_POOL_ENTRIES 70000

Jesus Canastero