Note on SFP and CFPs

In the data centers, and MTSO’s and Central Offices, and POP’s, the equipment uses SFPs and CDPs.

I didn’t think much of them. Just a little module, no big deal to me.

I was talking to a vendor a minute ago. 10 gig SFPs sell for as much as 3500 for name brand ones, generics for as little as 1500.

100 gig CFPs sell for about 25k.

You need a minimum of two per fiber circuit, typically 4 per circuit.

It is common to have 8, 16, or 32 SFPs per circuit card in a RODAM and several cards per RODAM, and several CFPs per RODAM.

Due to the cost and due to design flexablity these are not typically installed when the box is turned up, rather when a circuit is turned up.

I am guessing, if data center build outs stop, the demand for SFPs and CFPs will continue for some time afterwards.

Cheers
Qazulight

1 Like

Do you mean SFP (small form-factor pluggable transceiver) and CFP (C form-factor pluggable)?

2 Likes

Do you mean SFP (small form-factor pluggable transceiver) and CFP (C form-factor pluggable)?

Yes.

I am sorry about all the gobbly gook. I don’t have time right now to write a good post.

Cheers
Qazulight

That’s correct. But what stock are you talking about?

tj

That’s correct. But what stock are you talking about?

tj,

No sure, I thought I saw one mentioned, was hoping someone else would remember.

Cheers
Qazulight

The Aaoi thread got me to thinking about it.

I will have to reaearch AAOI, and then maybe look a little deeper.

I swapped a card out on an old SONET unit the other night, it was my first time to plug and unplug 32 connections. With the new proceedures requiring cleaning and scoping of each fiber every time it is disconnected, I can see where multi connections will be coming soon.

I remember the splicers fusing one fiber at a time in 1999. Today with ribbon cables and self aligning fusion machines they are doing many at a time.

I expect to see simular things happen with the fiber. There is just too much that can go wrong one fiber at a time, when a tech has to handle 32 just to swap a card.

Cheers
Qazulight

2 Likes

Hi Qazu,

You need a minimum of two per fiber circuit, typically 4 per circuit.

That all depends on where the circuit goes. If it is a local circuit then yes it could be 2 per circuit but if they are sending it around town it could have 8 or more. If it is going national then the circuit could have many more. Its really hard to state just how many would be used in each circuit without knowing the drawing. But besides that the communication providers will have many more in stock. I typically have at least 50 Sfp+, XFP, and sfp’s on me. Many of them are vendor specific but I also have many that are generic.

Due to the cost and due to design flexablity these are not typically installed when the box is turned up, rather when a circuit is turned up.

That all depends on the vendors equipment. When we place a card in the ciena 6500 equipment we generally place the sfp’s and fiber out the ports. It takes less man power to run the fibers all at once. But these are generally in the central offices. In the field we will generally run the fibers for the equipment but place our sfp’s one at a time. In the office we are generally adding customers at a higher rate then the field so it makes more sense.

I am guessing, if data center build outs stop, the demand for SFPs and CFPs will continue for some time afterwards.

The CFP’s are going away and being replaced by QSFP28 but if you saw my last post the Data centers are upgrading about every 3 years. The data centers are where it is at. They are growing and moving to higher data speeds more and more. I think they will keep all of the communications companies (cable, telco) running to keep up with them. The two companies I like in this market are acacia, Which does not make their own sfp’s but contract them out, it has a high market with ZTE. They are down a lot this year because China slowed down. But they also have customers in the data centers. The other is AAOI which does make its own sfp’s (vertically integrated) and has a high placement in the data centers (Amazon) but has a smaller market in China.

http://www.fiber-optic-transceiver-module.com/what-are-the-d…

Andy

3 Likes

Sorry Andy,

Can only give one rec.

In my MTSO we are not building out the SFPs. The tech is changing so fast that we may have to pull them and put new ones in.

Also, contractor I was talking to called the 100 gig units, the backhaul unit CFPs and the 10 gig units SFPs. I haven’t had time to study it, I am still decommissioning the 2G system. We should be completely breaker off before Christmas and start wrecking the equipment out by January. In our office we need to hurry so we will have room for the 5G equipment coming in.

Cheers
Qazulight (So yes in fact I AM a dinosaur. Kids today can’t even tune a 600 channel analog system.)

1 Like

Hi Qazu,

Also, contractor I was talking to called the 100 gig units, the backhaul unit CFPs and the 10 gig units SFPs. I haven’t had time to study it, I am still decommissioning the 2G system. We should be completely breaker off before Christmas and start wrecking the equipment out by January. In our office we need to hurry so we will have room for the 5G equipment coming in.

You are right Qazu, the older 100 gig modules were CFP’s but the new ones that are coming in will be about the same size as a sfp+ they run on less voltage and less heat. I expect your 5g equipment will have these in them. Pretty much everyone is using sfp as a generic name now but the 1 gig’s are sfp’s and the 10 gigs are sfp+. You have to look close at them because they are exactly the same size but there will be a label on them stating their bandwidth. But you are right this industry is really moving fast.

Let us all know when the 5G equipment is in place and what they are feeding it with. I am curious because I have run a lot of circuits into the cell sites and we have been running 1 gig circuits in right now. I am wondering if they are going to put in 10 gig. Some of the cell sites are still using sonet and have 500 meg at the most. Many of them 100 meg.

Andy

1 Like

Let us all know when the 5G equipment is in place and what they are feeding it with. I am curious because I have run a lot of circuits into the cell sites and we have been running 1 gig circuits in right now. I am wondering if they are going to put in 10 gig. Some of the cell sites are still using sonet and have 500 meg at the most.

I may or may not be in a position to say.

The RAN engineering courses I took explained that we are removing the base station from the cell site in this iteration.

Today, in 4G LTE, the radio head is in the antenna at the top of the tower. Fiber connects the base station to the antenna. In 5G, because the wavelengths are so short, the coverage by sector, will be very small. So we are moving the base to MTSO’s in some areas and to existing cell sites in others.

So (and there is still a huge turf war going on here) the 5G equipment in the MTSO will belong to the cell tech crew. This means not enough work to justify a MTSO tech.

As such I have applications out for cell tech jobs and a Core Backbone Network Specialist job. (Baby sitting a hundred miles of fiber.)

I do not think we will be pumping up the backhaul beyond a GIG for the first year. However, you have me curious, I will see if I can find the engineers working the 5G beta in Austin.

Cheers
Qazulight

1 Like

Thanks Qazu,
Some people were stating on the boards that everyone was going to get a gig on 5g but we both know that would be impossible. We have fiber going directly into most of the towers where I am at so we could drop them 100 gig if they wanted it but I am curious what they plan.

I hope you get the cell job. Those guys have to work some OT but it seems like a great job.

Take care,
Andy

Andy,

Some people were stating on the boards that everyone was going to get a gig on 5g but we both know that would be impossible.

1 gig plus at the air interface IS the target spec. However, we are in Beta testing now. Austin and now one other market are being deployed. Not only that, the final 5G specs are not published yet. So while we will be pushing deployment as early as next spring, all of the target specs will not be met in any significant part of the network before 2019.

The thing I really find fascinating as these are supposed to be all white box deployments and virtual servers and self organizing networks.

The idea is that if you have a disaster like we had in Baton Rouge a year or so ago, the network can automatically spin down in the affected server center and spin up elsewhere on the fly.

Think T-1000 Terminator.

Cheers
Qazulight