Network Latency

Hi all,
This is my first question here, i am managing small airport network about 100 users iam using two mikrotik RB433 one receives 3.5 Mb point to point line and the other is 1.5 Mb both of them connected to cisco switch 2960 I made 2 vlans one for each line all departments i send signal to them by fiber converters because switch i used does not support gigabit at the end there are some D-Link and TP-Link switchs and wireless.
The problem is when i changed the service provider the speed of the net is not fast comparing with the previouse ISP althgou the bandwidth was less than the avialable one.
I attached some snapshots when i ping and when i see my tx and rx.
Any help will be appreciated.
Who is Participating?

Improve company productivity with a Business Account.Sign Up

Norm DickinsonConnect With a Mentor GuruCommented:
There are three or four "pipelines" for the Internet you must access which vary depending on your internal equipment and your provider. The first pipeline is your own internal network, which has not changed so we assume that is not the issue. The second pipeline is from your network to the local ISP, which you claim is offering you more bandwidth, so once again we assume that this is not the answer. The third pipeline is from the ISP to their upline provider. The fourth and beyond are from that provider to other top level providers. It is one of these upper level pipelines where I suspect your problem lies - and as such, it is beyond your control.

There are a variety of speeds available to end user, which to most of us represent bandwidth. However, just because you get a fast speed to the ISP does not mean that the ISP has a particularly fast speed to the web, or has not oversold his speed and saturated his bandwidth. This was often the case with cable modems vs. DSL. While the cable modem ranked higher speeds in most cases, it also was a shared line so that all users on a certain leg had to suffer when one user was using especially heavy bandwidth, and caused huge swings in actual available bandwidth. DSL on the other hand offered a slightly slower speed but it was a dedicated line to each end user, so was more stable.

My guess is that your "faster" ISP is connected using a slower - and cheaper - main feed into the upline system, or perhaps they are connected into another lower tier provider which gives you either more hops or slower hops once you get to their border routers. The older ISP that you switched away from was probably more expensive because they were paying for more bandwidth and it may be best to switch back to them and buy more bandwidth from them if this kind of speed issue is going to be a problem. Sometimes you get what you pay for.
gsmartinManager of ITCommented:
I would have to agree with 'tgfdotus' analogy, which is typical.  When selecting providers you need to find out what Tier level provider they.  If they are not a Tier 1 provider then you need to get information on their topology and providers they rely on; hopefully only Tier 1.  However, this matters very little if they oversubscribe/over sell their bandwidth.

You need to run a few traceroutes to a few different high bandwidth web sites.  The ones I usually use to get a variety of different results (google, yahoo, intel, Cisco, YouTube, MTV, Microsoft) to name a few.  I would expect higher latency from YouTube, Google, and MTV.  However, Intel and yahoo I usually get very good results.  The point here is to isolate and rule out which routers/providers are increasing your latency.  

Compare your results with the other traceroutes and pinpoint where the root cause of your latency.  

Also, you can use to identify what your download and upload speeds as well as latency you are receiving from local servers.

FYI... I've seen different paths to google with higher latency then others.  This is because they have multiple severs/Data Centers around the country and world.  However, local servers should be resolving the majority of your requests.
Latency and bandwidth SLAs aside, when you switch providers, something that's often overlooked is speed/duplex mismatching.  Verify that your interfaces from the providers equipment has the best negotiated connection, like 100mbps Full Duplex.  If you see that you only have a 1/2 duplex connection, it can severely impact performance if the provider's equipment is set to 10mb FULL duplex.  Any mismatch at all can cause bad performance.  Verify what the provider actually has their equipment set for (not just what it SHOULD be, but what it IS set for) and then hard set your interfaces to match.
While all of the suggestions are certainly valid, there isn't truly enough information to really know.  Judging from the interface list, it'd appear that you are 100% wifi for your internal network.  Per your original post, you mention 100 users locally.  Though it's not the question that you're asking, it does beg the question of your internal design.  I am not familiar with your particular wifi radios, but the thought of one primary wifi radio handling 100% traffic for anwhere near 100 clients gives me an uncomfortable feeling that your latency could be from an overloaded radio.  I know that the prior ISP felt faster and that your internal network hasn't changed, but unless that Microtik has some magic, there's got to be some opportunity to improve your network topology.

If you care to explore the possibilities, let us know.

1 question for you--the MTU for your outside interface is 1526bytes.  Did you force that value or is it auto determined?  Sounds a few bytes high.  What is the largest ping packet you can send before the pings start fragmenting?  Ping -l xxxx, keeping increasing the -l value until you get fragmentation, then back it down a notch to find your real max transmission unit size.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.