Community discussions

MUM Europe 2020
 
WirelessRudy
Forum Guru
Forum Guru
Topic Author
Posts: 3089
Joined: Tue Aug 08, 2006 5:54 pm
Location: Spain

Nstream, the mystery solved?

Thu Aug 30, 2007 4:46 am

I spend two days reading this forum on the nstream topic.
I see supporters as well as hard critizism.
I see ' good for heavy traffic, bad for little traffic' statements.
My confusion only became bigger, what is the best?

We all trive to get the best network running. Most WISP's are getting higher bandwith feeds into their networks and see an increase demand or use for VOIP nowadays.

What we need here is a good explanation from MT, or anybody else, with test results etc. under what conditions nstream should, and when not, be used.

A beginner, and even a mediate skilled MT operator should be able to find an anwer on this forum. I see very little input from MT in most of the topics regarding nstreame.

Why not clarify more, and have a WiKi telling any proposed user some basics as to when nstream should be considered, and when not, and the related settings on polling, framer policy etc.

Most networks have 'hidden nodes' and need polling. Most networks need low latency for VOIP. Apart from that, most networks need fair distributed bandwith to its users.
The users don't care, as long as it works, we can't bother them with explanation on the technics used.
And, in regard of a MT statement found somewhere; Most networks have links that at times generate loads of traffic, but are mostly relatively slack. I think in general we need a solution that works under all circumstances!

Is the issue going to be solved in any future upgrades?
 
0ldman
Forum Guru
Forum Guru
Posts: 1446
Joined: Thu Jul 27, 2006 5:01 am

Re: Nstream, the mystery solved?

Thu Aug 30, 2007 5:15 am

I'd love to hear myself, more in when Nstreme is good vs when it is bad.

I understand no one wants to admit their work is bad or has a weak spot, however, you have to look at it like you don't drive a Dodge Viper to the grocery store for a family of 12. Everything has its purpose, and to try to force a solution into a situation it isn't meant to be isn't fair to anyone.
 
User avatar
Equis
Forum Veteran
Forum Veteran
Posts: 888
Joined: Mon Jun 06, 2005 6:48 am

Re: Nstream, the mystery solved?

Thu Aug 30, 2007 2:36 pm

I have found....

In a point to point link if I need the fastest I can I use ntreame.

In my opinion

Nstreame is faster
Nstreame will keep pings low (with polling) in a fully maxed out link
If the link is not maxed out no nstreame is better (lowering pings)
an Nstreame link is more sensitive to interference
an Nstreame link will disconnect/reconnect more often (see above)
Nstreame is fantastic on long links (20kms +)

I have perhaps 50 p2p backhaul links (all ROS) an use Nstreame on about 10 of them (the busiest ones)

I have tried nstreame on point to mulitipoint bad got less jitter etc with it off (30 clients)

Long story short..
Nstreame is good to squeese ever bit of speed out of a link but does have it's downsides.

:-)
 
yudhi
newbie
Posts: 29
Joined: Mon Mar 06, 2006 9:27 am
Location: World wide web

Re: Nstream, the mystery solved?

Sat Sep 01, 2007 1:38 pm

I have found....

In a point to point link if I need the fastest I can I use ntreame.

In my opinion

Nstreame is faster
Nstreame will keep pings low (with polling) in a fully maxed out link
If the link is not maxed out no nstreame is better (lowering pings)
an Nstreame link is more sensitive to interference
an Nstreame link will disconnect/reconnect more often (see above)
Nstreame is fantastic on long links (20kms +)

I have perhaps 50 p2p backhaul links (all ROS) an use Nstreame on about 10 of them (the busiest ones)

I have tried nstreame on point to mulitipoint bad got less jitter etc with it off (30 clients)

Long story short..
Nstreame is good to squeese ever bit of speed out of a link but does have it's downsides.

:-)
So what kind of situation we better use Nstreme and when do not use Nstreme ? :D
 
User avatar
Equis
Forum Veteran
Forum Veteran
Posts: 888
Joined: Mon Jun 06, 2005 6:48 am

Re: Nstream, the mystery solved?

Sat Sep 01, 2007 1:42 pm

I use Nstreame if the link gets maxed out without it often...
 
WirelessRudy
Forum Guru
Forum Guru
Topic Author
Posts: 3089
Joined: Tue Aug 08, 2006 5:54 pm
Location: Spain

Re: Nstream, the mystery solved?

Sat Sep 01, 2007 2:30 pm

Well, we are getting somewhere,
A conclusion migt be (?):

Use Nstreme on long distance links which,
also carry heavy traffic, and
where there aren't lots of interfering frequencies around.

or,

Don´t use nstreme when:
- lots of clients at short distances.
- in areas where several AP`s working the same band, or other interfering frequencies are around
- low traffic links.

Regarding the traffic: What kind of link is always carriing high levels of data? I think even the busiests links have plenty of ´low use´ time. During such ´low use´ periods the traffic would then actuall suffer from nstreme?

And what about links carriing lots of VOIP traffic? The framer policy is actually interfering with smooth VOIP traffic while setting small frames or ´non´ framing policy is taking away some of the advantages of nstreme?

So, still some questions....
 
tombrdfrd66
Member Candidate
Member Candidate
Posts: 243
Joined: Sat Jan 10, 2009 12:09 am
Location: New Zealand

Re: Nstream, the mystery solved?

Mon Nov 11, 2013 5:07 am

So here we are, six years after this was first raised - by this poster and a similar set of questions from Wireless Rudy at about the same time - and nstream framing policy is still an arcane mystery.

Just the blandest, briefest and most unhelpful skim over the subject in the Wiki, two desperate and unanswered appeals for help in the forum, and the mystery remains.

I've just moved to nstream on one network but as usual with Mikrotik stuff have just left it all at default owing to an almost complete lack of information about the intricacies of its configuration. What situations are the various framing policy settings designed for? Do/can I disable CSMA? The techs at Mikrotik in their ivory tower can't be bothered to tell me, so busy are they inventing even more bells and whistles most of us won't have a clue how to implement.

This is why, if I was building a network all over again, I would not use Mikrotiik gear. It's good hardware, technically it can be brilliant, but it's so far up itself - so elitist - that us poor admins just out of networking 101 can only rely on the charity of our fellow users to help us along.

This is the reason I have ony just moved to RoS v5 from 3.30, and it's going to be a long, long time before I venture into v6. I'll let you lot do all the beta testing on your own clients and will see if there's anything sufficiently useful to upgrade for when you've sorted out all Mikrotik's problems for them.

Oh, and I guess I'll have to leave my nstream framing policy at 'none'.
The more I know, the more I know I don't know.
 
ejansson
Member
Member
Posts: 301
Joined: Fri Oct 21, 2005 4:09 pm
Location: Manitoba, Canada

Re: Nstream, the mystery solved?

Mon Nov 11, 2013 7:49 am

Best Fit : fits in as many packets as it can then sends the combined packets, most often used
Dynamic size: Gathers packets till timer elapses and sends packet, have not found much difference to Bet fit, but in theory should have lower latency as there is know waiting for a full load
Exact size: creates a packet the size of the frame limit even if it has to fragment packets, will maximize through put but packets can be fragmented.

Most cases I have used the 3200 default, but under interference conditions thing may improve with smaller packets as they do not get corrupted as easily

CSMA, seem to be a better under most conditions
 
tombrdfrd66
Member Candidate
Member Candidate
Posts: 243
Joined: Sat Jan 10, 2009 12:09 am
Location: New Zealand

Re: Nstream, the mystery solved?

Tue Nov 12, 2013 3:16 am

Thanks for that.

Are you saying it's better not to disable CSMA?
The more I know, the more I know I don't know.
 
ejansson
Member
Member
Posts: 301
Joined: Fri Oct 21, 2005 4:09 pm
Location: Manitoba, Canada

Re: Nstream, the mystery solved?

Tue Nov 12, 2013 4:16 pm

Most of the time I do check the box. I've seen marginal differences if any under most conditions.

Who is online

Users browsing this forum: dmcdon, llubik and 40 guests