Texas Hold’em and the IETF – Did Brocade bet against TRILL?

For the last two post, which you can find HERE and HERE, I’ve knocked Cisco around. For those who don’t know me, I should warn that I am an equal opportunity offender. With that in mind, let’s take a look at Brocade’s implementation of TRILL.

As most of you should know, TRILL uses IS-IS on Layer 2 to identify the shortest path between switches, and load balance across those paths. Since this is happening at layer 2, not layer 3, it does away with Spanning Tree, which means more bandwidth and faster fail-over using the same number of ports, fiber paths, cables, and switches.

Of course, despite the fact that we all understand the above to be true, Brocade decided to go their own way and replace IS-IS with FSPF or Fabric Shortest Path First.

If you haven’t done much work in SAN environments, you may not be familiar with FSPF. Brocade created FSPF in 1997 to answer bandwidth concerns in Fiber Channel SANS. It has since become the standard path selection protocol in Fiber Channel fabrics.

With that understanding, let me back up and rephrase. As TRILL utilizing IS-IS was being developed by the IETF, Brocade a member of the IETF, decided to implement their own version of TRILL utilizing FSPF.

Brocade along with Cisco are both offenders. They both claim to be working with the IETF, yet at the same time both have released competitors to TRILL. Are we to believe that Brocade worked to make TRILL the best possible solution at the same time that they were creating a competitor to it? What about Cisco and FabricPath?

Both companies claim that their solution “extends” TRILL with additional features.

Were those “extended” features brought up in meetings when the TRILL standard was being discussed? Did the IETF choose to ignore those suggestions? I doubt it.

Cisco, Brocade, and most like every other vendor sat at the table the same way a poker player does during a game of texas hold ’em. No one showed their cards, but everyone watched the flop, river, and turn cards, to see what they could create with their own hands to drive the other players off the table.

Make no mistake, TRILL did not benefit from Brocade, Cisco, or any other vendor’s presence on the committee. Their involvement was for their own purposes, not the benefit of customers.

Cisco is SCARED! Why Cisco won’t release an emulator.

Greg Ferro posted on his blog another plea to Cisco to play nice and give network engineers tools for testing, verifying, and learning new technology. If you’ve missed the recent debate on the matter, it’s OK. Crawl back under that rock, you won’t miss a thing.

I generally read Greg’s posts while nodding my head like some sick bobble headed doll, with an occasional grunt in agreement. However today, my head stopped bobbing when I realized something…

Cisco is AFRAID of the virtual switch/router.

Let that sink in for a minute.

I know what you’re thinking. “They don’t have anything to be afraid of. That’s crazy talk.”  I’m sure that people said the same about Dell and HP when ESX was first announced. “They don’t have anything to worry about. No data center could ever virtualize all of their servers. That’s just crazy.” Only, it did happen. Right now I am sitting just a few hundred feet from 100 servers that would be over 500 servers if it wasn’t for vmWare. Think of the lost revenue to Dell and HP.

But, you say, “what about the Nexus 1000v”. What about it? Cisco had already lost sales because all of those virtual servers didn’t need individual switchports. That was Cisco’s way of getting some of that revenue back. It wasn’t about extending network engineer’s control into the virtual environment. It was about lost port revenue.

Imagine with me for a moment. What would happen if you could virtualize the Edge and Core layers of your network all onto a single HA cluster. (Maybe a couple of Dell or HP servers.)

Firewalls, Check
Routing, Check
IDS, Check
VPN, Check

Where is the need for 10GB, 40GB, 100GB, TRILL, or Fabric Path? What about all of the other technologies that Cisco will sell us over the next 10 years, forcing us to replace existing hardware?

Outside of the HA cluster, you would need a couple of switches for Distribution, and you would need your normal Access layer switches, but how many components of the network would be cut? Not only routers, firewalls, and switches, but adapters, redundant power supplies, wireless controllers.

It’s already been done. Look at Cisco Call Manager. A router, switch, and server that do the work of racks and racks of PBX equipment.

“But, I just want them to release it so that I can test.”

Cisco has three choices: 1. Stick fingers in their ears and hum loudly. (Current tactic) 2. Release a good virtual network platform, and wait for everyone to ask, “wait…why can’t we virtualize this for real?” 3. Release a crippled, barely working virtual platform, and then get derided for their poor product.

No matter how Cisco looks at it, they lose.

Suddenly I am asking myself. After IPv6, what is the next big thing to happen in networking? Could virtualization change networking the way it changed servers?

Is Cisco getting back on track?

Cisco’s big-man-in-charge, John Chambers, sent out an email to all employees this week, which outlined a few important things:

-Cisco has lost focus
-Cisco was caught off guard by certain movements within the Networking community (openflow, new products from other vendors, etc)
-Cisco makes it difficult for new product to make it to market
-Cisco has to focus on the core business components, rather than continuing to diversify into low margin consumer markets
-Most importantly, Cisco shareholders, employees, and customers are not happy with the current direction that Cisco has taken

The message is a great read, and gives me hope that Cisco can get back on the ball, and address some of it’s core issues. Kudos to the Cisco team for taking a hard look at where they are, and making decisions to correct their wandering trajectory. Here’s hoping they follow through!

http://blogs.cisco.com/news/message-from-john-chambers-where-cisco-is-taking-the-network/

Microsoft meets the first snag in plan to purchase IPv4 addresses

As you should now be aware of, Microsoft is planning on hoarding purchasing a large huge block of IP addresses from Nortel. Now ARIN chief, John Curran, has made it clear that if the plan does not meet the current ARIN requirements for transfers, the IP address space can be reassigned. Here are a couple of relevant quotes:

Companies that are allocating their address to a third party can ask for compensation if they want to, he said. However, the acquiring party is required to show an immediate and appropriate need for the addresses, he said.

Existing transfer policies allow up to 12-months worth of address space to be transferred from one entity to another, he said.

So, that brings up the question, can Microsoft show a need for 666,000 in the next 12 months?

Link: http://www.computerworld.com/s/article/9215091/IPv4_address_transfers_must_meet_policy_ARIN_chief_says

Microsoft sets the price of IPv4 space post IPocalypse.

The resellers market for IPv4 address space has just heated up. Microsoft has puchased 666,624 IPv4 addresses from the burned out shell of Nortel for the tidy sum of $7.5 million dollars. That works out to a little over $11 per IP address. That’s pretty expensive when you consider how much Nortel paid originally. This also make me worry about how quickly prices will inflate on the remaining address space that is floating around out there.

On another note, I guess Microsoft realized that since they had built poor, if any IPv6 support into their products for so many years, they better make certain they have enough v4 addresses.

via http://downloadsquad.switched.com/2011/03/25/microsoft-buys-666-624-ipv4-addresses-from-nortel-for-7-5-milli/

RSA hacked? No, really.

In an open letter to customers, Art Coviello from RSA admitted that attackers had gained access to some of their internal information regarding the RSA SecurID products. While there was no customer information lost, it says that RSA is working with customers to provide “immediate steps for them to take to strengthen their SecurID implementations.”

I think I’ll go for a walk now…