treats DWDM fiber

Dario Vercelli
Dario Vercelli used Ask the Experts™
on
I have two proposals to diversify the DWDM fiber line now I have a single path with two 23 km fibers crushed together

The first provides two new routes with a stretch of 23 km routes at 35Km and 38KM, are indicative lengths that can be specified as soon as they are joined on the WDM equipment;

the second provides that the possibility of bringing a third path parallel to the two entrances and also in this case completely diversified, the length would reach about 40 KM;

My question is :
Can I have different lengths on two or three fiber ?
Can I have an excessive latency problem?
At the disk I / O level arriving on two different fiber lengths (see above) I can have problems?
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
Ken BooneNetwork Consultant

Commented:
So are these in a LAG group?  Or is the 2nd path just a redundant path for failover?  When you come out of DWDM what is going in to?  A switch, a router, or straight to storage?  What is on each end of the DWDM handoffs?

So far my gut says you won't see a difference.. but I am assuming things are being handed off through a switch or router.  Not sure if this is a direct link between storage devices.
Dario VercelliSystem administrator

Author

Commented:
So are these in a LAG group?
No there is two fiber indipendent

Or is the 2nd path just a redundant path for failover?
all path is active for traffic

When you come out of DWDM what is going in to?
A switch

What is on each end of the DWDM handoffs?
There is a physical WAN connection

(*) it's not a direct connection between two storage

Thks
Ken BooneNetwork Consultant

Commented:
In that case I think you will be fine.
Introduction to R

R is considered the predominant language for data scientist and statisticians. Learn how to use R for your own data science projects.

Dario VercelliSystem administrator

Author

Commented:
then I confirm that the different length of the optical fiber does not involve problems on the disk I / O?

Is there any documentation to confirm?
Ken BooneNetwork Consultant

Commented:
Nope.. just going on gut feel here man.  The traffic is getting switched through the switches.  I think any delay you would see based on 33km vs 40km  would be less than a fraction of  a millisecond.  And the traffic still has to get switched before it reaches the disk i/o components.  But since you are running the fiber and it is a concern for you,  run Dual fiber paths in both directions, then use each dual fiber path as active and the other as failover.  I don't think you are going to see a difference though.
Dario VercelliSystem administrator

Author

Commented:
The cases I can have are always two with active / active fiber:

1) I connect the two sites with a fiber of 24 Km and one of 40 Km, the latency of this difference 16 Km is important?

2) I connect the two sites with a fiber 30 km and a 35 km, with 5 km difference, the latency is lower in this case and I have no problems on I / O?
Network Consultant
Commented:
Ok so to put things in perspective in general single mode fiber will have a latency (in perfect vacuum environment of 4.9 micro seconds.  You have to remember that a micro second is 1 millionth of a second.  So we tend to talk in milliseconds in networking terms which is 1000th of a second.  So 16km will introduce 78.4 microseconds of delay - as in 0.0000785 seconds.  A 5km difference will be a delay of 24.5 microseconds of delay as in 0.0000245 seconds.  At that minuteness I don't "think" it is going to matter.
Dario VercelliSystem administrator

Author

Commented:
thanks for your comprehensive answer
Dario VercelliSystem administrator

Author

Commented:
thanks for your comprehensive answer

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial