• 05/30/2014
    11:45 AM
  • Rating: 
    0 votes
    Vote up!
    Vote down!

Ethernet Switch Market Drops Almost $1B in Q1

Dell'Oro Group cites "pause" in datacenter switching spending due to Cisco Nexus 9000 switches as one reason for the decline.

The market for Layer 2 and 3 Ethernet switches fell by nearly $1 billion in the first quarter of 2014, from the fourth quarter of 2013, to just over $5 billion, according to a report from research firm Dell’Oro Group.

At the same time, sales of 40GbE and 100GbE switches rose sequentially in the first quarter as enterprises upgrade their networks to the faster switches from 1GbE and 10GbE models, according to Dell’Oro Group’s Ethernet Switch Quarterly Report.

There are several reasons for the overall decline, says Alan Weckel, vice president of Ethernet Switch market research at Dell’Oro. First, deployment of Ethernet switches on campus networks has declined as enterprises opt for more wireless connectivity.

Second, there was a “pause” in the datacenter switching market in the quarter, due to the transition to industry leader Cisco's new 9000 series switches, which Cisco introduced in November 2013. At the same time, Cisco unveiled its version of a software-defined network platform, Application-Centric Infrastructure (ACI). It touted ACI and the 9000 switches at its Cisco Live 2014 user conference last week in San Francisco.

Because the 9000 line is still so new, potential customers were still just evaluating the products and hadn’t committed to buying anything, Weckel says, adding that some of the new switches haven’t begun shipping yet.

“It gets announced, they get beta units, and it takes three to six months, sometimes longer, to play with it in their labs with a bit of production traffic before they are willing to buy a significant quantity,” he told us in a phone interview. “So during that time, they obviously don’t buy the old stuff either.”

Cisco declined to comment on the Dell’Oro report, but did say that sales of the 9000 series are promising.

“Cisco’s Nexus 9000 and our ACI solution has been in the market for less than two quarters and is off to a very solid start with 1,000 customers in the pipeline,” the company said in a prepared statement. “To date, Nexus 9000 grew to a user base of 175 customers, up from 20 customers last quarter.”

Weckel says another factor contributing to the first-quarter loss was that anticipated seasonal decline in networking equipment sales was “stronger than usual.” The first quarter tends to be a slow quarter because there is a surge of spending in the fourth quarter of the previous year as many IT departments want to spend their budgets before the year is out or lose it, he notes. The first quarter also is typically spent planning for networking projects, but not purchasing equipment.

The 40GbE market was stronger than the overall networking market, but it’s still early in the adoption of 40GbE, so a modest increase in unit sales shows up as a large number percentage-wise, according to Weckel.

“But also, all the newer platforms have 10 gig to the server with a 40 gig spine or core, and those are very attractive on a cost and deployment basis.”

While still No. 1 in market share in the Ethernet switch space, Cisco’s share declined to 62%, down from 66% in the first quarter of 2013, according to Dell’Oro. HP stayed in second place with an 11% share, unchanged from the year before. Juniper passed Dell to rank third in market share at 4%, up from 2%, and Dell slipped to fourth with a share of 2% from 3% in the year-ago quarter.


SDN impact

Earlier this week, Infonetics released a report in which it cited "SDN hesitation" as slowing down service provider spending on routers and switches in the first quarter of this year. It will be interesting to see if this trend continues in Q2.



Could SDN Be the Issue?

I think a lot of IT Pros are still trying to figure out how SDN will mature and what its ultimate role in the IT landscape will be.  To that extent, perhaps no one wants to rush into anything too fast.  At least not yet anyway.

Re: Could SDN Be the Issue?

AbeG> I think there's some truth to that. In particular as mentioned above, I think some people are sitting back waiting on the Cisco ACI play to come fo fruition. As the largest incumbent, Cisco undoubtedly has customers hesitating on upgrades because they aren't sure whether to wait for the full ACI solution before kicking off their refresh cycle.

If you're a Juniper switching customer, what's stopping you upgrading? I don't really believe that most customers are sitting there saying "let me see how this whole white box /SDN switching things goes"; those that want to go white box are more likely already testing or deploying, and they'll have a decent idea of the timelines on which they thing they'd be able to put these into deployment. In reality though, most white box switches seem to be going to the huge players in the market, not Joe Enterprise, who really wants an 'out of the box' (no pun intended) solution I think.

The fall will continue

I've been discussing this with customers and predicting something like this. 

The question is really how much further sales will fall and there are many factors. 

1. Whitebox and commodity switches sell for one fifth the price of vendor branded products. The TAM is sharinking rapidly as vendors drop prices to compensate. 

2. virtualisation means that there are less ports in use for the same number of physical servers. 

3. campus networks are being replaced with wireless networks for laptops. 

4. tablets and smartphones are replacing desktops & laptops in the enterprise

5. physical network devices have been feature complete for the last 3 years. There is limited demand for new devices. 


For these reasons, the question is how much further the market value will fall in the next few years. I cannot see many reasons for Ethernet switches sales to grow except for SDN in the Data Centre and even that will sell very few devices because of virtualization. There is also the negative impact of cloud providers further eroding device sales. 

Open to suggestions but I'm expecting to see this discussion every quarter for the next year or two. 

Re: The fall will continue

Wired connections will always be faster and more reliable than wireless connections.  To that end, there may be room for growth within certain niche markets.  Some ideas/examples that may be worth discussing are:

Internet/Network access in public areas.  Public transportation vehicles such as airplanes, greyhound buses, and aamtrak trains offer electric outlets and wifi internet access.  An onboard media server could offer high quality HD content more reliably and efficiently over an ethernet port vs. wifi. 

The free WiFi offered by these providers is typically slow enough as it is.  The connection speed is only made worse when all users are connected via WiFi, perhaps at legacy Wireless-G speeds.

Internet/Network access at home.  Telecommunications providers will typically run phone and cable lines throughout your home at your request.  Why not offer to run ethernet cables along with an optional switch?  At the very least, I could see students living off-campus being very interested in this and landlords being responsive to those requests.

IPv6 Support.  It seems as though the best advice for organizations of all sizes with respect to IPv6 is to get ready ahead of time so as to avoid any potential panicks down the line.  No one will ever be able to guess if or when IPv6 will be required as opposed to an option, but for some organizations, the gamble of waiting is not worth the risk.

Re: The fall will continue

SDN definitely plays a predominant role especially in any distributed protocol by decoupling control plane from the data plane. Just for analogy, In OSPF protocol RFC, most of the content describes about the control plane(how to manager the stuff and et al.,) and heart 'n soul of the protocol (i.e, Algorithm) is just described in few pages of it. It may not be a surprise, If SDN eliminate's the need for having separate protocol specific distributing intelligence for any future protocols design. More you can go to search fiberstore on google.