Facebook aims to knock Cisco down a peg with open network hardware

Puffy

Looks interesting.....Facebook aims to knock Cisco down a peg with open network hardware | Ars Technica

Master Of Puppets

Really interesting, I'm looking forward to seeing how this turns out. Thanks for sharing.

DevilWAH

I do like this idea, but one things that concerns me.

Look at the up-time on network equipment 10+ years has been seen on a number of ocassions and 3 or 4 years is not uncommon at the access layer. the reason for this up time is because how locked down networking hardware and software is. When things do go wrong trouble shooting is straight forward due to he structured nature and control.

For large companies that have large network teams an opensource configurable OS to run your networking makes sense, but for smaller companies they could quickly get in to a muddle and introduce complexities that would make it difficult even for an experienced consultant to trouble shoot due to its bespoke nature.

For such a critical service opening it up could actually cause a lot of companies heartache, if they simple go after the cash savings with out relising that a much great level of engineer skill might be needed.

Having said that I look forward to seeing where this goes, just got some new 6500's which I have a 5-6 year refresh cycle on, so I can see networking looking very interesting when I come to replace them

eansdad

Wonder if they will have an article about when it fails and Facebook goes back? Just from a security aspect if all switches are controlled from a single point wouldn't that create a single point of failure? Wanna take down facebook, just hit the infrastructure servers. The whole reason Cisco is as big as it is is because of their reliability. Will Open Compute switches and routers be able to stay up for years? I see to many questions regarding early adoption and possibly adoption in general.

Will be glued to the outcome on this.

shodown

I've been following this movement for a while. I do think for Massive corps this is the way of the future. THey can buy there own hardware and program the routers to behave however they want. This just makes since for them as sometimes they have requirement that just aren't met by the hardware vendors. I had a project recently where I had to setup 2 service provider sized devices for a popular media company just so they could conduct some of there operations. The company has less than 200 people. with the openflow/SDN movement it was possible to buy 2 large servers and do the same functions at 15 percent of the price, it was discussed by the company but deemed too risky at this point.

JustFred

Interesting, looking forward to read more about how this turns out a few months from now

chrisone

not a chance......

networker050184

shodown wrote: »

....deemed too risky at this point.

Not to cherry pick your reply, but I think something like this will always be too risky for the majority of companies. Having a large vendor to fall back on is a huge asset for large companies.

phoeneous

chrisone wrote: »

not a chance......

Agreed. FB needs to stick to social media. How's their phone doing?

DevilWAH

Exactly when you stand to lose £millions for each minute of down time, saving money by buying hardware that increases risk and increases resolution times can be false economy.

It has to be asked, why when Linux can be so much cheaper than Windows to purchase, who it is not the leader in small and medium and large companies. While it is easy to argue it is a better and more flexible OS, the backing of Microsoft and large user base of standardized applications on windows is something companies are prepared to pay for. Banks who are only in it to make profits, still often side with the large vendors, if Linux or opensource really was more profitable in the real world would already be the main players.

What of course is great about this is no matter what it will push the big vendors forward to insure they stay ahead of the curve.

Ahriakin

Openflow...letting the services, and therefore their managers, enforce network policy and COS etc....what could go wrong from the folks that routinely yell every problem is the network, or the firewall or whatever before even checking if their servers are turned on. It is a joke for anything outside of SOHO.

Ditto for this project. If you are Facebook and can afford to hire top programming and network talent you might make it work, just for you. For the masses no. ASICs, newer forwarding techniques, secure software design - these are all beyond the vast majority of companies out there.

FloOz

I like the idea, however I don't think this take off whatsoever in the general enterprise network

shodown

networker050184 wrote: »

Not to cherry pick your reply, but I think something like this will always be too risky for the majority of companies. Having a large vendor to fall back on is a huge asset for large companies.

I get where you are coming from 100 percent. To be honest I was shocked that it was even brought up. the cost of 2 ASR 1006 routers with smart net is way cheaper in the long run than the cost of 2 Dell servers going down and not having a clue what went wrong, or having a established support network in place to ensure problem resolution.

dave330i

No maintenance plan means no critical system on it.

RTmarc

They tried this a while back with their open compute model. It made absolute zero impact on the industry. Why? As already stated, companies don't want to take the risk. At the end of the day, they can always deflect to Cisco or HP or Dell or IBM or whomever without taking full responsibility.

Node Man

If Cisco has about 80% of the global market share and Juniper has about 15%, Facebook is a distant third place in the rankings, at best.

But on a bright side, all the open source/ custom configs, would mean an employment boom for network engineers lol.

SteveO86

I think I'll quit Networking if I ever walk in a DC and see a Facebook logo on any piece of networking equipment...

"There is a routing loop we need 500 likes to solve it!!!"

"Share 100 times and you PC will operate at GB speeds!!!"

They are probably going to try and jump on the SDN bandwagon, this doesn't seen too different from what Google did where they programmed their whole network. I don't about you guys but learning a proprietary piece of equipment that is used so little in the enterprise is not anywhere on the top of the my to-do list (or anywhere on it for that matter)

chrisone

Node Man wrote: »

Facebook is a distant third place in the rankings, at best.
.

Not even close to being 3rd.....FB is not even in the networking top 50 companies. Will never reach top 25! Facebook would have to change their company identity from social media to network technology and invest billions just on "starting" up a network brand. I would buy linksys and netgear before I buy a facebook network device.