[Last Call] Learn how to a build a cloud-first strategyRegister Now

  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 433
  • Last Modified:

Unamanaged Dell switches. Need a good test for saturation. One segment is having problems.

The problem is users in building 2 lose their connectivity through quantum servers to our database. Intermittant issue. Usually happens in the morning when it is busiest.

300 users. 3 buildings.
Building 1 hosts 20 servers on GB backbone. 150 users there with 10/100/1000

Building 2 100 users 10/100/1000 linked with 1gb fiber to the Dell 5448 switch in building 1.

Building 3 50 users 10/100/1000 1gb fiber link to building 1.

During the morning we lose our database connection. Our DBA swears nothing is wrong with the software. But the quantums go offline and our DBA suspects network saturation.

so the question is how can I guage if a network segment is saturated on unmanaged switches???

I do a ping -t from building23 to the servers and get <1ms always.

The servers are always below 5% nw utilization.
1 Solution
Does the problem happen at the same time every morning?
pkwillisAuthor Commented:
not at the same exact time. no.

A quantum is an exe that communicates for the clients request to our secure DB. The weird thing is each quantum is an exact copy, yet each day it is a different quantum that goes offline. This leads me to suspect it is in the DB logic, most likely a timeout.

In order to prove this and rule it out, I need to look at the segment in building one and see how much traffic is flowing.
Do these unmanaged switches support any SNMP polling? If so, you might be able to at least look at the uplinks.
Failing that, choose a non-involved workstation or server at each building (if these are Windows or *nix boxes) and use a free little app called iperf. You can set an instance of the app at each box running and test true bidirectional traffic. It's not exact, but if you ran a bidirectional test running at say 50mbit/sec between the two - and said test didn't stop and/or fail during the quantum communication loss - then you could at least say 'well, I can pass a nonstop 50mb/sec stream on that segment without problems.'

SNMP is  great tool, alternative would "manualy" trying each link with bandwith mesuring soft- like this one
It does need two PC on each end, receving end CAT5 you can move across diffrent ports, and run test , that will show if any port "craping out" and does not perform well.
In attach pic, I was ttesting firewall performance on one of the machine.
Before the test make sure PC are good in both direction, you can do that by connecting them directly with reverse CAT5.
Aaron StreetInfrastructure ManagerCommented:
First of all you wont get any SNMP on an unmanaged switch, hence the name unmanaged. you cant assing an IP address to an unmanaged swith so there woudl be no way to retrive SNMP info.

one other way to do this is to purchase a 1gig hub or a dedicated wire tap. (or if you can enabale managment ment on one of the core swihcs you may be able to set up a mirrored port) and look at the data witth an application such as wire shark.

I know for instance the 5448 is a managed switch, so you should be able to set up this to be monitored via using snmp.

but if uyou run a ping 24/7 with out a single droped packet and constently <1ms responce then its highly unlickly your link is over uterlised. pings dont have any piority over other traffic so if the network is conjested enough to affect the servers then you would expect to see it also affect your pings.

iperf is a great tool and if you dont like Command line interface look for Jperf which is a jarva GUI frount end. for it


Featured Post

A Cyber Security RX to Protect Your Organization

Join us on December 13th for a webinar to learn how medical providers can defend against malware with a cyber security "Rx" that supports a healthy technology adoption plan for every healthcare organization.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now