Live data from Hacker News

Introducing “6-pack”: the first open hardware modular switch

code.facebook.com

41–50 of 97 posts

Re: Introducing “6-pack”: the first open hardware modular switch

#41
post #22

Earlier quoted context omitted.

Is anyone building a copy of that to sell to other people besides Facebook? Similar to how you can buy a Backblaze Pod ( https://www.backblaze.com/blog/backblaze-storage-pod-4/ ) from http://www.45drives.com/

Apparently juniper is with their OCX1100[1] model. [1]: http://forums.juniper.net/t5/Data-Center-Technologists/Junip...

All of these things are based on the Broadcom Trident II chip. If you want it cheap, don't go to juniper - get one from Quanta. You can have a 32x40G switch for $6,000.

Re: Introducing “6-pack”: the first open hardware modular switch

#42
post #37
post #20

Earlier quoted context omitted.

You probably don't need 16 40Gbps channels at home, either.

Heh maybe not need, but want? I definitely want. Looks like a 40 GbE NIC is just $500: http://www.amazon.com/MELLANOX-CONNECTX-3-56GBE-1PORT-MCX313... Just need an SSD RAID to feed it...

Less than $500 actually, you can get them for about $350.

Re: Introducing “6-pack”: the first open hardware modular switch

#43
post #14

I was expecting a link to something like this, but more advanced http://opencores.org/project,mac_layer_switch and some links to digilent dev boards for hardware, however the linked project is a bit more ambitious, all full of custom ASICs and such. An open source switch made out of COTS FPGA dev boards would be interesting. So you'd use something like http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,400... B…

That NIC seems quite overpriced for 4x1G...

Re: Introducing “6-pack”: the first open hardware modular switch

#44

Earlier quoted context omitted.

Apparently juniper is with their OCX1100[1] model. [1]: http://forums.juniper.net/t5/Data-Center-Technologists/Junip...

All of these things are based on the Broadcom Trident II chip. If you want it cheap, don't go to juniper - get one from Quanta. You can have a 32x40G switch for $6,000.

thanks, good to know!

Re: Introducing “6-pack”: the first open hardware modular switch

#45
post #22
post #5

Earlier quoted context omitted.

Facebook's Wedge 1U switch is what this builds on. https://code.facebook.com/posts/681382905244727/introducing-...

Is anyone building a copy of that to sell to other people besides Facebook? Similar to how you can buy a Backblaze Pod ( https://www.backblaze.com/blog/backblaze-storage-pod-4/ ) from http://www.45drives.com/

I'd imagine folks like Quanta or Hyve Solutions will be selling something like this soon.

Re: Introducing “6-pack”: the first open hardware modular switch

#46
post #17

Isn't this just wrapping for the COTS switching ASICs or am I missing something big? Almost all "open" projects seem to hit a point where the line between open and closed is quite arbitrary, even if they choose not to see it that way.

"Wrapping commodity ASICs" is worth $4B: http://www.arista.com/ If Facebook wants to give me that value for free I won't complain.

And the founder/chairman/chief development officer of Arista, Andy Bechtolsheim, is a board member and founder of the Open Compute Project.

Re: Introducing “6-pack”: the first open hardware modular switch

#47
post #14

I was expecting a link to something like this, but more advanced http://opencores.org/project,mac_layer_switch and some links to digilent dev boards for hardware, however the linked project is a bit more ambitious, all full of custom ASICs and such. An open source switch made out of COTS FPGA dev boards would be interesting. So you'd use something like http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,400... B…

That NIC seems quite overpriced for 4x1G...

It's not a NIC; it's an FPGA that can be programmed to do things that NICs have never dreamed of.

Re: Introducing “6-pack”: the first open hardware modular switch

#48
post #14

I was expecting a link to something like this, but more advanced http://opencores.org/project,mac_layer_switch and some links to digilent dev boards for hardware, however the linked project is a bit more ambitious, all full of custom ASICs and such. An open source switch made out of COTS FPGA dev boards would be interesting. So you'd use something like http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,400... B…

That NIC seems quite overpriced for 4x1G...

Good thing it is not targeted at people looking to buy a 4x1G NIC, but at people who want a nicely integrated FPGA dev plattform with 4x1G and fast connection to a PC.

Re: Introducing “6-pack”: the first open hardware modular switch

#49

Earlier quoted context omitted.

Apparently juniper is with their OCX1100[1] model. [1]: http://forums.juniper.net/t5/Data-Center-Technologists/Junip...

All of these things are based on the Broadcom Trident II chip. If you want it cheap, don't go to juniper - get one from Quanta. You can have a 32x40G switch for $6,000.

On a similar note, I just recently became aware of Cumulus[1] Debian based switches (but the good bits are closed source) from among others edge-core[2] (via a presentation by PaaS-provider http://zetta.io) -- eg:

http://whiteboxswitch.com/products/edge-core-as4600-54t

[1] http://cumulusnetworks.com/support/faq/

[2] http://www.edge-core.com/

Re: Introducing “6-pack”: the first open hardware modular switch

#50
post #37
post #20

Earlier quoted context omitted.

You probably don't need 16 40Gbps channels at home, either.

Heh maybe not need, but want? I definitely want. Looks like a 40 GbE NIC is just $500: http://www.amazon.com/MELLANOX-CONNECTX-3-56GBE-1PORT-MCX313... Just need an SSD RAID to feed it...

While I wasn't aware you could get them that cheap, I think I'd still look into second hand infiniband for home use. Allows the use of copper for short (pc-to-pc, point-to-point) distances. While optic is absolutely cool, as far as I can tell a single 1m patch cable will cost ~80 USD -- quite a lot considering the cost of the NIC... And even in IBM's sales brochure for their optical switch, infiband comes out a little ahead:

http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=S...

Post reply on HN