x
Routing

Appliance Vendors Don't Fear Cisco

Even as Cisco Systems Inc. (Nasdaq: CSCO) trots out a new router to consolidate telecom boxes, the appliance vendors continue to say the big router firms simply can't compete with their standalone products.

Earlier this week, Cisco introduced its ASR 1000 platform, a router that's meant to also do the jobs of deep packet inspection (DPI) gear, session border controllers (SBCs), and firewalls. (See Cisco Takes Hold of the Edge.)

Juniper Networks Inc. (NYSE: JNPR) has also been pushing the idea of moving intelligence deeper into its products by more closely integrating session management and policy control in its products. Alcatel-Lucent (NYSE: ALU) might be getting into the game as well. (See Juniper Pushes Services Policy, AlcaLu's Edge Upgrade, and AlcaLu Identifies Deep Packet Potential.)

The argument goes that by letting routers handle these functions, enterprises and service providers can do away with standalone network elements that only perform one or two specialized tasks.

"Obviously customers and service providers want fewer boxes in their networks," says IDC analyst Eve Griliches. "Everyone wants appliance functionality moved into routers or switches."

But DPI and SBC vendors are skeptical that embedded functionality can completely replace standalone products. For one thing, they question whether an integrated device can perform at the same level as a standalone device.

"The burden is on the newcomers to prove that this works," says Kevin Mitchell, director of marketing at SBC vendor Acme Packet Inc. (Nasdaq: APKT). "There's a whole host of functions that, combined with performance, calls for a standalone network element."

It's debatable whether customers even want an all-in-one device, says NextPoint Networks Inc. chief marketing officer Mark Pugerude.

"The god box has fallen out of favor," Pugerude says. "When carriers are looking at applying new applications across their networks they tend to buy opportunistically. They use best-of-breed products for critical applications and combine less important applications into one platform."

Perhaps most importantly, standalone vendors say that a player like Cisco probably won't be able to support the developing needs of customers, because it lacks their level of specialization.

"It comes down to core competency and focus," says Arbor Networks CTO Rob Malan. "If a company is 100 percent focused on a market, it will build better products than those focused on the general purpose market."

— Ryan Lawler, Reporter, Light Reading

Page 1 / 2   >   >>
douaibei 12/5/2012 | 3:46:05 PM
re: Appliance Vendors Don't Fear Cisco the telco carriers are experiencing the network neutrality, the bits pipe is becoming more dumb than before, and the telco is developing a business model which is more independent of the hardware infrastructure. so the importance of hardware vendors are going down, but the terminal system is emerging to be the key killer offer in the market.

we have seen some carrier are releasing the bid for huge blad server system which terminate all the service intelligence, so we do not need to worry the cisco will be any threat to they competitors. the reason is even cisco themself is not care how bits pipe, what they want is to control or lead the end user service experience which rely on cisco offer. the offer can be tangible or intanbible.

I has been looking for the relation between the overall teleco market and the terminal, business model information in the market. hopefully some one can help here.

ex-procket-jock 12/5/2012 | 3:46:02 PM
re: Appliance Vendors Don't Fear Cisco I find it interesting that folks would draw comparisons between Cisco's ASR and a god box. I certainly don't see this as a god box. What Cisco has done is combine a few services-oriented features at the edge and given providers a toolkit to create their own "service differentiation" feature sets as required. This strategy is not at odds with IAB principles as it still keeps the core doing what it does best - processing packets as fast as it can.

On another point, what is the point of a fast processor if you cannot load it up with services? Of course there is a penalty for doing so which is why Cisco created the honking QFP in the first place.
mr zippy 12/5/2012 | 3:46:01 PM
re: Appliance Vendors Don't Fear Cisco Maybe the question to ask is what have Cisco fundamentally done. I think it is two things:

1. they've combined together fairly unrelated functions into the same "box" (yes, these functions all look at packets, but what they do with them and to them are significantly different. e.g. there'd be very little if any code duplication between packet forwarding code and DPI code). I think the benefit here is primarily in data center space savings / rack space.

2. these functions were and are typically performed by separate boxes, interconnected with a non-proprietary network / bus i.e. Ethernet/IP. By combining these functions within the box, they've eliminated the non-proprietary bus. A valid technical reason to do that would be for performance - if the functions need to communicate at > 10Gbps, you can't use commodity Ethernet/IP, and other interconnects like Infiniband aren't as commodity and ubiquitous. A valid commercial reason to do that is to bind your customers to the proprietary bus, so that future purchases must be from you.

It seems to me that if you go down this path, you're trading off flexibility. Because you've bought into many functions in the same box, if one function fails, you can't replace only that function. Function failure impact other pretty much unrelated functions operation. Because you've bought into the proprietary bus, maybe even to continue to justify that earlier decision, the available options for future functions is reduced - for example, you've limited the options to solving your problems to Cisco solutions - and they may not always be the best ones.

I'm starting to realise that properties of good networks are similar to properties of good programs, namely,

* loose coupling between functions,
* tight cohesion between related sub-functions.

The way to achieve those things is,

* modularity,
* minimal and simple interfaces between functions that have to communicate

By placing separate functions in physically separate chassis (the plural), you typically and inherently create modularity. By using a non-proprietary and simple interconnect like ethernet/IP between those boxes, you typically and inherently create a minimal and simple interface.

Summing this up, that means:

* design your network, the functions you need to perform, and the equipment you buy to build it to be as loosely coupled as possible,

* stick to dumb and non-proprietary interconnects between the functions and therefore equipment that performs them,

which results in future flexibility.

If you need a box like an ASR1000, because it saves you space, or appears simpler to operate, just make sure you're aware of the trade offs you might be making by tightly coupling functions, and going with a proprietary network/bus between those functions.
mr zippy 12/5/2012 | 3:46:00 PM
re: Appliance Vendors Don't Fear Cisco "Obviously customers and service providers want fewer boxes in their networks," says IDC analyst Eve Griliches. "Everyone wants appliance functionality moved into routers or switches."

I want less boxes, but I don't want appliance functionality moved into routers or switches.

Actually, I don't really want "less boxes". I want just enough boxes to do the job; no more, no less. Just enough boxes in the network to make it available, scalable, adaptable, general purpose, high performing and cost effective to operate.

Either too many "boxes", or not enough "boxes", means compromising on one of those key network properties.
rjmcmahon 12/5/2012 | 3:45:59 PM
re: Appliance Vendors Don't Fear Cisco Mr Zippy,

I'm pro internet, open access, etc and would like to see a data infrastructure that supports innovation, free speech, et al. I'm also for folks coming up with workable properties that engineers can use. Chemistry's periodic table reveals that these models can be extremely worthwhile and necessary for progress.

I don't know anything about Cisco's ASR 1000 and the "dumb bits" guideline may apply here. But the challenge I perceive is to be careful and not to over generalize.

In your programming analogy, taken to the extreme, every function would be put into it's own memory space and all shared data constructs would be realized via IPC mechanisms (e.g. shared memory.) What we've seen is that things like threads have become prevalent in practice and serve a purpose too. Part of the designer's job is to discern when these guidelines apply and when they don't.

Good luck with it.
materialgirl 12/5/2012 | 3:45:58 PM
re: Appliance Vendors Don't Fear Cisco We keep dancing around the dumb-and-open network, that helps the economy but has no current viable business model, and the smart-but-expensive-and-controlling network that works for stock holders.

The CSCO box clearly supports the latter. This is because their customers such as ATT, CMCSK want expensive, closed, controlled networks for their business purposes. What they have shown, is that the SECOND they get hold of your data, they either give it away or shut you down on a whim. That this is also likely to hamper freedom of idea sharing and hence the economy over time is none of their concern, since they have their stock options lined up.

However, for the "rest of us", as soon as these boxes go in, we have to trust our service provider not to abuse them, since the difference between "management" and "censorship" is so small. At a minimum, they will increase our cost of access.

However, without a viable alternative financing structure for open access, this is what we will get. IMHO, these boxes are bad news for everyone other than a few managers at big service providers who are getting rich on overly generous stock payouts.
rjmcmahon 12/5/2012 | 3:45:57 PM
re: Appliance Vendors Don't Fear Cisco On investment, the 1031 exchange Tenant in Common stuff seems to be a booming industry. The TIC with a master lease by the sponsor appeals to many retirees needing a steady cash flow. The sponsor guarantees a relatively conservative rate of return and takes all the upside above that rate. I wonder if there are any lessons from there that the data network providers could benefit from?
materialgirl 12/5/2012 | 3:45:55 PM
re: Appliance Vendors Don't Fear Cisco Dear RJ:
Sounds line a plan to me. The problem today is that the dumb pipe telco model is the DRAM business or the airline business all over again. Zero retained earnings after decades of operation.

Bandwidth is either a valuable monopoly or a valueless commodity with too many players. The problem is how to manage the monopolist beast.
materialgirl 12/5/2012 | 3:45:55 PM
re: Appliance Vendors Don't Fear Cisco Dear RJ:
Sounds line a plan to me. The problem today is that the dumb pipe telco model is the DRAM business or the airline business all over again. Zero retained earnings after decades of operation.

Bandwidth is either a valuable monopoly or a valueless commodity with too many players. The problem is how to manage the monopolist beast.
rjmcmahon 12/5/2012 | 3:45:54 PM
re: Appliance Vendors Don't Fear Cisco The problem today is that the dumb pipe telco model is the DRAM business or the airline business all over again.

I can only be grateful that the airline industry was built out by folks like William Boeing and not Warren Buffett. I think our country would be a shell of what it is without this strategic industry.

No government air mail contracts, no airline industry. No US post, no good roads. No municipal fire departments, no water infrastructures. No street lighting and electric street cars, no power infrastructure. Come on folks, every other meaningful infrastructure project required both a public and private component to get the job done.

Why would this one be any different?

Today's leaders need to wake up and understand that a quality data communications infrastructure is strategic to the country. It's pretty obvious, even with today's substandard internet, that peoples lives have the potential for much improvement with a modern communication's network with built in storage.

Bandwidth is either a valuable monopoly or a valueless commodity with too many players. The problem is how to manage the monopolist beast.

Agreed. I perceive two monopolies, one established (voice/TDM) and one not (data/best effort). Conflating the two may be part of the problem.

FCC chairman Martin's recent speech at Stanford shows how the elites in our country aren't meeting basic standards.

http://tinyurl.com/29eavl

Martin recognizes that the broadband definition of 200Kbs is insufficient and needs to be raised. He than says he thinks it should be raised to at least 768Kbs and advanced services to 1.5Mbs. The audience doesn't bat an eye. Later he says that removing regulations requiring copper loop unbundling has stimulated fiber access network investments. Yet all one has to do is look at the GLW's annual reports to see this is false.

Why we put up with this crap is hard for me to understand. I thought folks at Stanford were educated? A child left behind seems to apply even at our elite universities these days. Sad.

We've got to crack this nut folks and we're not getting help from those in positions of power. At least not from what I can tell.
Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE