Thuderbolt to 10gbe?

[sfxguy]

I know the current Thunderbolt to ethernet is gigabit.

Anyone know of any plans for one to 10gbe without having to go through an external box with a 10gbe card in it, (which are already available)

An adapter just like the current one with 10gbe would be amazing.

 

[ThePreditor]

That would be called Thunderbolt? 10Gbps ethernet connections are EXTREMELY rare unless you are enterprise level or extremely high-end post house, so I see that need highly unnecessary.

 

[carestudio]

Interesting. I am wondering if the 10Gb would saturate the entire Thunderbolt bandwidth and cause display not function properly? just guessing.
On the other hand, I am particular interested in 4-port gigabit to TB.
check the smalltree

 

[theSeb]

That would be called Thunderbolt? 10Gbps ethernet connections are EXTREMELY rare unless you are enterprise level or extremely high-end post house, so I see that need highly unnecessary.

No, that would be called 10GbE. Thunderbolt is not 10 Gb Ethernet. As our devices become more interconnected in the home and Internet connection speeds are becoming faster, 10 GbE is becoming something that is actually necessary in the home and it shouldn't be too long until we start seeing consumer devices that support it. 100 GbE is now being used in the most demanding applications and development of Terabit Ethernet is under way. It's not hard to imagine that 10 GbE will start moving into the consumer space.

http://www.theregister.co.uk/2012/05/02/ethernet_alliance/

http://www.techpowerup.com/162027/Intel®-Ethernet-Controller-X540-Brings-10GbE-to-the-Masses.html

----------

Interesting. I am wondering if the 10Gb would saturate the entire Thunderbolt bandwidth and cause display not function properly? just guessing.
On the other hand, I am particular interested in 4-port gigabit to TB.
check the smalltree

There is a separate channel for the display data. You can read Anandtech's Pegasus Promise R6 review where they put in 6 SSDs and achieved data throughput of nearly 10 Gb/s, but the display etc still worked fine.

 

[murphychris]

100 GbE is now being used in the most demanding applications and development of Terabit Ethernet is under way.

InfiniBand is what's being used there, not 100GigE. Not even 40GigE. And 10gb/s InfiniBand is cheaper than 10gb/s ethernet. There are problems getting ethernet to scale at higher speeds including 10gb/s, so it seems increasingly clear there will need to be topology changes - and in effect ethernet is being reinvented but given the same name.

 

[theSeb]

InfiniBand is what's being used there, not 100GigE.

I am not quite following. Are you suggesting that 100 GbE is not being used in demanding applications?

http://en.wikipedia.org/wiki/100_Gigabit_Ethernet#Routers_and_switches_with_100GbE_interfaces

Cisco's first deployment of 100GbE at AT&T and Comcast occurred in April 2011.[45]

and so forth.

Maybe you took exception that I didn't say Infiniband and 100 GbE is being used in the most demanding applications? I apologise for the lack of love for Infiniband in my post.

 

[murphychris]

I am not quite following. Are you suggesting that 100 GbE is not being used in demanding applications?

It seems that I am, although in retrospect it's incorrect to categorically say 100GigE is not being used at all. And actually it's somewhat misleading to consider it ethernet at all because the link layer is totally different. Already 10GigE consumes a lot of CPU dealing with protocol overhead - it's just not scalable. InfiniBand doesn't have this problem. I don't think it's unfair to consider 100GigE implementations to be something of a closed-loop trial.

Maybe you took exception that I didn't say Infiniband and 100 GbE is being used in the most demanding applications? I apologise for the lack of love for Infiniband in my post.

Right, instead I should probably have taken more exception to the notion that 10GigE is something presently, or in the very near future, necessary in the home.

 

[theSeb]

It seems that I am, although in retrospect it's incorrect to categorically say 100GigE is not being used at all. And actually it's somewhat misleading to consider it ethernet at all because the link layer is totally different. Already 10GigE consumes a lot of CPU dealing with protocol overhead - it's just not scalable. InfiniBand doesn't have this problem. I don't think it's unfair to consider 100GigE implementations to be something of a closed-loop trial.

As long as the Ethernet Alliance keeps calling it Ethernet, then I don't really care.

Right, instead I should probably have taken more exception to the notion that 10GigE is something presently, or in the very near future, necessary in the home.

I am not sure why you would think that. My near future is about 2 years and there is definitely a need to go beyond 1 GbE for home users, considering how the world is moving. Sure, over here in the UK we are behind the rest of the developed world when it comes to broadband speeds, but that does not mean the rest of the world has to suffer.

 

[murphychris]

I am not sure why you would think that. My near future is about 2 years and there is definitely a need to go beyond 1 GbE for home users, considering how the world is moving.

Home users have only incidentally arrived at GigE because the economics of GigE over Fast Ethernet made it compelling to provide it before there was widespread need. Most home "modems" for DSL or cable are at best 100Mb/s, many are still at 10Mb/s. Also home users are substantially more likely to use wireless connections, which is significantly bandwidth constrained even compared to GigE let alone 10GigE. I don't at all see the use case beyond niche.

Sure, over here in the UK we are behind the rest of the developed world when it comes to broadband speeds, but that does not mean the rest of the world has to suffer.

South Korea is the world leader in internet bandwidth provided to the most people. Most of this has to do with government policies. Even there, they are nowhere near 1000Gb/s speed. Experiments have started with GigE speeds, but the average is 17Mb/s with a peak of almost 50Mb/s. [1]

Yes there's desire but I just don't see the money there to broadly role out internet connections inside of the next 10 years that will utilize 10GigE connections. We'd need a significant leap in performance for essentially no cost increase to get there.



[1] http://en.wikipedia.org/wiki/Internet_in_South_Korea#Internet_speed

 

[theSeb]

Home users have only incidentally arrived at GigE because the economics of GigE over Fast Ethernet made it compelling to provide it before there was widespread need. Most home "modems" for DSL or cable are at best 100Mb/s, many are still at 10Mb/s. Also home users are substantially more likely to use wireless connections, which is significantly bandwidth constrained even compared to GigE let alone 10GigE. I don't at all see the use case beyond niche.



South Korea is the world leader in internet bandwidth provided to the most people. Most of this has to do with government policies. Even there, they are nowhere near 1000Gb/s speed. Experiments have started with GigE speeds, but the average is 17Mb/s with a peak of almost 50Mb/s. [1]

Yes there's desire but I just don't see the money there to broadly role out internet connections inside of the next 10 years that will utilize 10GigE connections. We'd need a significant leap in performance for essentially no cost increase to get there.



[1] http://en.wikipedia.org/wiki/Internet_in_South_Korea#Internet_speed

I am not suggesting that Internet speeds will all of a sudden jump to 1 Gbps and that's not at all the main reason why I think we'll start to see 10 GbE moving into the consumer space. It's more to throw data around the house like streaming etc from and to multiple devices. Wi-fi is a big pain and I'll reserve my judgement on 802.11 ac until we see some real world tests. I can put my computer right next to a 450 Mb/s router running on an un-congested 5 GHz channel and I still get around 11-12 MB/s.

 

[murphychris]

It's more to throw data around the house like streaming etc from and to multiple devices.

I don't buy it. Not yet anyway. 10GigE is in the realm of bandwidth required for raw uncompressed video editing. Not pushing files around the house. Most high end homes now aren't even being build with ethernet anymore. In big houses, I'd expect to see WDS before Cat 5e/6.

 

[jb510]

a little o/t

I actually think one of the driving forces that will bring 10GbE to office/home is the broder adoption of SSDs and faster NAS/RAIDs.

1GbE (125MB/s) is roughly equivalent to what a single HDD is capable of over SATA, so until very recently there hasn't been much point in having faster networks since not much could make use of them outside the server farm. SSDs however have leap frogged ahead to 3-500MB/s, and that means a single SSD can easily saturate 1GbE. Where the concern used to be saturation across the router by multiple client now there is again a concern of saturating a single link, and the switches will need to catch up.

 

[murphychris]

I actually think one of the driving forces that will bring 10GbE to office/home is the broder adoption of SSDs and faster NAS/RAIDs.

Certainly even the amateur photographer would like to have better than 100+MB/s to their NAS, which then makes NAS a more clear win. It's better than direct attach storage in every respect except performance over GigE. (Not that it's all that slow, it's about that of FW800 if you use async NFS, but it does suffer a lot more if you use AFP or SMB.)

Many professional photographers bulk up on DAS, and not just for fast local scratch/working files, but to store entire libraries - rightly because NAS is too slow for this. They see it as either/or. Rather than getting small but fast local storage for work files, and push/pulling their completed files up to a NAS. It's not that slow, but slow enough they'll risk their whole library to local storage (and file system corruption as a result of a crash or power failure).

HD video people cannot use fast local storage if they're in a collaborative environment. The files are so huge even over 10GigE it doesn't make sense. So they use 10GigE and live edit over the network to avoid push/pulling huge files around entirely.

So there is very clearly a workflow need for 10GigE for amateur and pro photographers, but even there it's not such a need that they are willing to buy 10GigE. Overwhelmingly they do not buy. It's still the domain of HD video (and enterprise for busy servers).

But the idea of home users pushing huge files around? I remain 99% unconvinced. Even wireless can stream the files they use, which are not big, and are highly compressed.