Outburst: Brocade – What’s their direction ?

Bought Foundry – rushed it ?

Brocade paid $3.4 billion for Foundry and rushed the acquisition using a lot of cash to close the deal quickly. HP just completed the 3Com purchase for $2.7 billion. To my mind, the 3Com assets are a lot more valuable since they also have firewalls, IDS and a wide range of networking products. Foundry only has Core Ethernet switching products and no access ethernet, no service provider, no wireless.

Of course, Brocade only makes high end FibreChannel switches which are high performance, high complexity and highly proprietary so Foundry makes a nice fit to nice little corner of the market for a tidy profit. But when Cisco announced FCoE and then threw a billion dollars behind it, Brocade’s core products around FibreChannel became obsolete. Oh, I know the FibreChannel will limp along for a few more years, but there is no five year plan for growth anymore. That’s over. Ethernet is the future.

And the Data Centre has become a war zone between HP, IBM, Cisco and now HDS. Where does Brocade fit in ?

Foundry Command Line is a mess

I have also worked on the Foundry Ethernet equipment and find it a confusing mess of interface, syntax and technology differences. It needs a lot of support to keep that equipment running. As a result, it’s hard to like and use the Foundry / Brocade ethernet story. I recently worked on a site that has ServerIron, FastIron and BigIron switches. Each one has a slight, but significant difference in CLI that makes them frustrating to work with (and appreciate how Cisco’s IOS interface is reasonably consistent). I had to constantly refer to the manuals to validate and cross reference every command. That’s a but annoying.

The EtherealMind View


So I reckon Brocade made a big mistake buying Foundry at all, and certainly overpaid for what they did get. Brocade’s share price has dropped a long way, and there has been VERY little information about what they are doing with their Ethernet products. They don’t share anything about future plans or get involved with customers from what I can see.

Where’s the plan ?

I’m not a Brocade customer, and unlikely to ever consider them until I know more about what they are doing. And that doesn’t mean meeting with a sales team or marketing team. I should be able to clearly perceive the future from their public presence. I don’t want closed briefings or NDA roadmaps. Just show me a coherent plan, on your website, and then I can consider you as a serious supplier.

Or is it just me ? Can you see a future for Brocade beyond five years ?

  • LLoyd Christmas

    I think you should give them another quarter or two. You’re being unfairly harsh. let’s see what comes up in the next 6 months as further Brocade and Foundry integration can take place. They made an “oops” last quarter and admitted that. How many CEOs would have done that. He took the blame and accountability. They mentioned they are working on the fix and we should know by the fall if it worked or not.

    Plus, a ton of engineers are being recruited and hired into Brocade. I think it’s only a matter of time until they unveil their strategy and blow people out of the water.

    As far as the cool billion Cisco threw at the Nexus line….they still cannot figure out supply chain, the latency is 3x the values I tested with the Foundry MLX and it cannot do MPLS or hold a full Internet routing table. As far as CEE/FCoE capabilities…look for a forklift upgrade in the hardware or a new chassis altogether. IMO, I will NEVER buy Cisco equipment. Their support (Brocade’s is MUCH MUCH better) has gone downhill and their own engineers that write the code are in India…not US. Talk about a language barrier. No offense.

    You mention the CLI differences. What exactly are you talking about? I think that the CLI will gain parity over time and what about migration from the Cat 6500 to the Nexus? The CLI differences are more drastic than Cat 6500 to Brocade. The Nexus code is build on their FibreChannel code base on the MDS and it 3 years behind Brocade’s according to Gartner. As far as operationally, the Brocade NMS system can discover and report against Cisco equipment. Can CiscoWorks do that? In addition, Brocade has made public that the NMS from fbrechannel will merge with Foundry’s creating one UNIFIED management platform for IP, fibrechannel, and FCoE.

    Another point is IPv6 performance….turn it on on the Nexus line and the performance drops in half. Yeah…I want to be the person to roll out IPv6 capable web hosting to my customer base and then answer questions on why everything got slower. Thanks Cisco! Turn IPv6 routing on the Brocade box….things are still wire speed.

    For all you ISPs out there? Don’t listen to this author. Brocade OWNS the top 3 Internet Exchanges IN THE WORLD! http://www.lightreading.com/document.asp?doc_id=188058 How many 10GB trunks can you link together on a Cisco chassis? Answer = 8. How many on Brocade chassis? Answer = 32. 4x scalability advantage.

    For virtualized datacenters out there. How many MACs can the Nexus handle? Answer = 128K. How many can Brocade handle? Answer = 1 Million. Gee Cisco…thanks for handling the proliferation of VM in my datacenter. Oh wait! You can’t? Hmm….perhaps another vendor will.

    Let’s finally talk about power and cooling! Just look at the facts on the vendors’ respective datasheets. In some cases, it’s startling. From a fibrechannel perspective, Brocade is 1/10th the power….that’s 90% less. On IP/Foundry side, it’s ~30-40% less.

    Not to be harsh on Cisco but I have seen them play dirty pool with marketing and in my clients’ networks. They sell/market/design one solution…then when it doesn’t work, they get arrogant or upsell to something that the customer could have gotten for less cost elsewhere. It just amazes me how many people have the Cisco blinders on. If you look at them from a pure technical perspective, the Nexus just isn’t up to snuff. Now, this article was about Brocade but insert most other well respected platforms out there and Cisco from a technical/architecture perspective doesn’t match up. In some cases, they don’t even come close!

    • http://etherealmind.com Greg Ferro

      First, let me congratulate you on fine rant. Outstanding.

      Second, Brocade should be clearly communicating with their customers – I should not have to wait for themto get around to telling me what’s happening, I need to know now. Tomorrow is too late.

      third, the foundry performance has always been good. The features have always been poor. Limited MPLS, limited QoS, limited load balancing and so on.

      Fourth, Foundry is used in the leading Exchanges because it is cheap, not because it is good. Internet Exchanges ( especially public IXPs ) are highly price sensitive and use almost no features. Flat out performance is what Foundry did well, therefore they sold well to IXP – not necessarily a good claim for product excellence.

      I have seen brocade / foundry do similar sales techniques. Especailly when exaggerating their technical support (before acquisition ) and over claiming features. The customer should always beware sales promises, if you have seen this then it’s your fault for being duped. The capitalist / business process does not allow for honesty,

      My final point is about openness, brocade continues to act closed and separate from their customers. They do not make resources available to consultant,s they are not sharing their product road maps or corporate vision. I can only believe that they do not have one. Here is their chance to open up and communicate.

      Thanks again for your comment. Good stuff.

      • Alex Hoster

        I hardly hesitated to post my comment. I could not resist such a bunch of bad faith. If Brocade is deployed in major IX, it’s because it is future-proof and not because it is cheap. How many vendors have deployed 100G ports using Cisco or Juniper? When I read your article, I can only judge your ignorance and how biased you are. I’ve worked for a decade in the networking industry, and the command line of Brocade IronWare is comprehensible and predictable.

        I have probably deployed about 500 networks in my entire life and about half of them on Brocade devices. There was nothing that the devices did not do right. I have seen them in enterprise networks, in academic networks, in research networks, in financial real-time networks as well as in carrier networks. Some giants Internet services providers do have Brocade routers in their core network. Actually thinking of one of the busiest Internet company 😉

        I could agree with you that the MPLS features are not as rich as some other vendors, but Cisco is far beyond Juniper in carrier MPLS. I have never met any limit of QoS or load balancing on Brocade devices, but maybe you can tell me how Brocade is limited in that regard. That would really be interesting.

        How by the way, what does Cisco actually do? I thought they were a Networking company, but now they seem to do WebEx, video conference, servers, security, wireless, PBX, retail phones, cheap SOHO switches and storage soon… they are now steamrolling on their good old friends such as IBM, had to fire massive headcounts and yet they want to be leaders in all these areas.

        Let’s face it, this article was not to mention something wrong with Brocade, but a mass.. Cisco is losing ground at breakneck speed!

  • Josh

    The company I work for is preparing a network forklift. I had the enlightening opportunity to compare Cisco versus Brocade. The Brocade story absolutely had some compelling factors – it was significantly cheaper, and the number of wire-speed ports was outstanding (especially 10gig ones). But, as the author mentioned, they lacked features and their product lineup lacked any kind of coherency. For their MLX switches, they kept pushing the MRJ-21 cards, which they kept claiming was a “standard” – I see no standards body that confesses to owning a standard for MRJ-21. Arista explicitly excluded MRJ-21 from their design because they didn’t believe it was a standard; I see no gear from HP, Juniper or any other appreciable competitor. Someone is making stuff up, and I’m pretty sure it’s Brocade. The first access switches that Brocade pitched to us were stackable switches – FCX I believe. The price point was awesome! However, there was one significant roadblock for us: we didn’t want to be constrained on PoE power per switch. We will be rolling 802.3at devices, and we didn’t want to have to plan our switch port mapping to make sure we didn’t exceed the available power on any one switch. Brocade didn’t have Power Supplies available to handle that much power, nor were they on any road map that our reps had access to (Cisco has a solution for this with the 3750X switches via the power stacking cable or dual power supplies, what happened to Brocade?). So, next we looked at their chassis-based switches. The single-supervisor SuperX looked pretty decent, but wait: it wasn’t going to support 802.3af power according to our rep. Something about not supporting the required supervisor engine or something like that. Ok, so next up was the redundant-supervisor SX chassis. That would certainly do the trick, but wait: the 802.3af blades won’t be available until later this year. We don’t actually need 802.3af power right away, so we could use standard blades, but we need switches ASAP, so that means some type of RMA or other run-around later to get the appropriate blades. Plus, once we hit this configuration, even with Brocade’s crazy discounts, the Cisco 4506E bundles end up being quite a bit cheaper.

    Come on, that’s just ridiculous! Brocade’s product lineup is so disjointed, and like Greg, I don’t see any sense of direction or comprehension of what customers really want/need. Their vision for the Data Center is practically non-existent. We asked for a discussion about what they are doing compared to Cisco’s data center vision with Nexus 1000v, etc. What we got was some junk about their 10gig CNA – yawn (Not entirely sold on FCoE anyways – iSCSI works pretty well in our environment). Brocade is effectively a niche player in the Ethernet segment, more or less like Arista. Our Brocade rep told us that Brocade as a company expected they were going to have to buy business if they wanted to increase their market share, and their pricing certainly reflected that viewpoint. Based on that, we had serious concerns about the long-term viability of Brocade as a stand-alone company, so we opted to go with Cisco. Cisco knows the business (or at least a significant percentage of it), they at least appear to deliver on what their customers want (although sometimes with smoke and mirrors), and they have a good vision for the Data Center (or at least one that sounds good on paper). Conversely, they really know how to milk their products, and the current blades/fabric modules for the Nexus switches suck. Comparing sheer PPS performance of a Nexus 7k to a Brocade MLX is quite embarrassing for Cisco. Plus, as Lloyd mentioned, the power/cooling requirements of Nexus versus Brocade is absurd – no one could ever accuse Cisco of being “green” in the data center. The wiring closets are reasonably similar as long as you are comparing chassis to chassis and stackable to stackable.

    I too dislike the apparent secrecy of Brocade. I can’t find any decent technical material on their website – only basic marketing and sales materials (seems like quite a few niche players play this game). What’s the big secret, Brocade? Our rep offered to get us access, but somehow that never came to fruition. If Cisco, HP, etc really wanted to know what you were up to, they would just buy some Brocade gear along with a Service Contract and get what they are looking for that way.

  • Me

    Greg – no ser≠vice pro≠vider

    Netiron XMR/MLX/CER (CER is 1 ru) MPLS standards supported for ISP – found on their public web site.

    ï RFC 3031 MPLS Architecture
    ï RFC 3032 MPLS Label Stack Encoding
    ï RFC 3036 LDP Specification
    ï RFC 2205 RSVP v1 Functional Specification
    ï RFC 2209 RSVP v1 Message
    Processing Rules
    ï RFC 3209 RSVP-TE
    ï RFC 3270 MPLS Support of
    Differentiated Services
    ï RFC 4090 Fast Reroute Extensions to
    RSVP-TE for LSP Tunnels; partial support:
    detour style
    ï RFC 3812 MPLS TE MIB
    ï draft-ietf-bfd-mpls BFD for MPLS LSPs

    Greg – no access
    Fastiron WS (10/100 and 10/100/1000 and both available in PoE and non PoE) 802.1af
    Fastiron LS (10/100/1000 with 2 or 3 10g uplinks) – optional stacking
    FastIron GS (10/100/1000 with 2 x 10g uplinks) – optional stacking and PoE 802.1af
    FastIron CX (10/100/1000 with 2 x 10g uplinks) – optional PoE+ 802.1at
    Note : 24 port version powers all ports af 802.1at
    48 port version power 26 ports af 802.1at
    FastIron SX 800 (up to 192 1g ports, support 10g ports support 802.1af on all ports with N+1 power
    FastIron SX 1600 (up to 384 1g ports, supports 10g ports, supports 802.1af on all port with N+1 power

    Greg – Brocade paid $3.4 bil≠lion for Foundry

    Foundry had 2b in the bank so paid 1.4b – a good buy IMO

    Greg – Of course, Brocade only makes high end FibreChannel switches which are high per≠form≠ance, high com≠plex≠ity and highly pro≠pri≠et≠ary so Foundry makes a nice fit to nice little corner of the mar≠ket for a tidy profit.

    Foundry was build on open standards (only two protocols that I know are proprietary are VRRP-E and MRP) If you have used VRRP-E there is no way you would want to go back to VRRP. MRP – everbody has a ring protocol (there is no standard that I know of for this) – so it is a me too thing.

    There are three code trains one for ServerIron, one for NetIron and one for FastIron. This was started under Foundry and finished under Brocade.

    Greg – Ethernet is the future.

    Yes it is. FCoE, DCBS, PFC and Trill etc.
    Loseless Ethernet frabics CEE – Brocade have that
    Trill is the main standard that is a long way off and is the layer 2 routing protocol.

    Brocade OWN their own layer 2 routing protocol (I think it is called FCOSPF) – we will maybe see this used as a pre standard so we can do multihop. Yes I said pre standard as the standard is not expected to be released untill next year. However the code will be in FPGA and can be change to the standard if/when release.

    As a Foundry/Brocade reseller for 10+ years (along with Cisco, HP and 3Com) I must say that you must be speaking to the wrong people – their is lots of information, not to mection many direct customer preso (by SE’s not sales), seminars and my.brocade.com community sites.

    In the first year of the Brocade takeover we have had the hardware platform design from the ground up for server SLB (at up to 70gps), the FastIron CX, NetIron CES and NetIron CER.

    3 New products – one for each space (ISP/Enterprise and SLB). As well as 16 port 10g cards for the BigIron.

    Lloyd also brings up other godd info on network management and the stated goals for it.

    Based on price point AND perforamance The Brocade solutions should be overlooked. But eval to see if the kit meets your needs. Check the Datasheet for the products and see if the feature is supported. Or use a real VAR that can do the work for you.

    • http://etherealmind.com Greg Ferro

      None of this matters until Brocade has a compelling vision. The Foundry product always “ticked the box” when it came to features, but usability and coherence between all products in the family is very, very poor.

      I don’t care whether you use merchant silicon or FPGA’s to get your performance. Give me a usability experience equivalent to Cisco and I would consider your product more favourably.

      If Brocade continues to sell cheap as a feature (mentioned on every comment) they are going to be in trouble.

    • http://www.pothole.com/~dee3 Donald Eastlake 3rd

      Not sure why you say TRILL ” is a long way off”. The TRILL base protocol was approved as an IETF Standard March 15, 2010:
      although there are some codepoints and formats that are still draft.

      Blade Networks has announced a product supporting TRILL:
      and the Cisco FabricPath product is just a pre-standards version of TRIL:.

      Donald (Co-Chair, IETF TRILL Working Group)

      • http://etherealmind.com Greg Ferro

        Well, the post was written six months ago. And most silicon switches will need replacement to support TRILL.

        It’s still a ways off.

    • Ed A.

      On thing to add…

      Cisco offered two versions of switching code – CatOS and IoS

      Cisco offered two verions of code for firewalls – PIX and PIX now maybe something in-between IoS and PIX in the ASA

      Cisco is now offering a whole new code – Nexus!

      So in counting like we are up to 5?

      Shall we continue with….. CM or UM or IPCC or UCCX or CCM? Someone please decide what we are going to call it, please!

  • http://etherealmind.com Greg Ferro

    The previous commenter has failed to disclose. It appears that this comment comes from Brocade premises, and likely that the author is an employee. Treat his conclusions with caution.

    I hate people who use my blog for guerilla marketing. If you want advertising, please talk to me and lets arrange it. Don’t steal my content to promote your agenda.

  • Kyle

    “Foundry only has Core Ethernet switching products and no access ethernet, no service provider, no wireless.”

    OK, OK, I sttoped here.

    Like your entire post, you’re clearly misinformed about Brocade’s products.

    You ask for not using your blog for guerrilla marketing, but that is just what you’re doing for Cisco.

    Do your homework first: http://www.brocade.com

    Playing the Cisco taliban is just pathetic.

    Disclaimer: I’m not a Brocade emplyoee.

    • http://etherealmind.com Greg Ferro

      No has ever accused me of being a Cisco fanboi, I am equally if not more criticial of their products and marketing. On the other hand, Brocade are doing such a crap job of marketing their products that I can only tell you what I know. Their website is unusable, it’s organised like a dog’s breakfast and everything is in PDF format.

      I’ve got better things to do.

      • leigh porter

        I would rather a PDF document than a bloody word document 😉

        What I like about their website is that the complete product documentation is on the same page as the products. This is just fantastic. If any sit eis a dogs breakfast it has to be Ciscos site…

  • Kelly

    Has anyone sucessfully used the Brocade VRRP or network teaming (dual NIC servers connected to two MLX switches for network redundancy)? We purchased two MLX switches with the promise of being able to do so where each server is connected via dual MLX. So far, Brocade has been evasive about being able to use that configuration even though it was promised during the sales pitch.

    • Ed A.

      Kelly – Yes

      If you are using MLX’s you can use MCT (Multi-Chassis Trunking) between the MLX’s and LACP/802.3ad from the servers just as long and the servers are using lAgP for aggregation not PAgP

  • Ed A.

    Greg sir you are a wacko, Please do your homework first.

    They offer the ( non-blocking)

    NetIron – Service Providor and Enterprise

    BigIron – Enterprise and HighEnd

    FastIron – Access layer and Low and Medium sized companies

    TurboIron – Fast Traffic mover 24 x 10Gb

    ServerIron – Load Balancing

    RFS and MC (MobileIron) – Wireless LAN

    All with SFlow for free not just on routers or with an upgrade!

    And now in Jan the VDX/VCS (Clouding and Data Center)

  • http://jimgoings.com Jim Goings

    We’re in the middle of doing a forklift upgrade from Cisco to Brocade. Price was NOT the deciding factor. Our Cisco discount made a Nexus 7000 solution on par with a Brocade MLX16. In the end, the MLX16 has way more performance and expansion.

    We chose Brocade based on features and performance. We also chose the MRJ21 solution to top of our server racks which has worked out beautifully so far… less cable mess for sure! The Nexus top of rack solution uses a powered device (like a switch but not) that trunks up to the switch via fiber with a limit of 40 Gbps (4x10Gb). We didn’t see the need to add another powered device where a simple MRJ21 patch panel would do.

    We are using MCT (multi-chassis trunking) between our core MLX16 switches and our FCX48 access layer switches. Works like a champ in our testing.

    We did struggle to get the HP Servers to use MCT but narrowed the problem down to a lacp issue with the HP driver. Brocade support was awesome in this regard, setting a full system to match ours and proved that the issue was on the HP side. When using Broadcom drives (not HP drivers), MCT worked fine. We are still waiting to hear from HP on this issue.

    We’re a few days away from being completely done, but two months of testing and a week of production use has worked out great.

    • http://etherealmind.com Greg Ferro

      Interesting comment. I’d be interested in hearing what you think in a few months time – sometimes the really good or really bad things only show up after you’ve been running for a while.

      My personal experience with Brocade has been less than exciting in both technology and business. Feedback like yours will challenge me to look again.


      • http://jimgoings.com Jim Goings

        Agreed. I failed to mention that another department in our company has been using Brocades for over a year and we spoke at length with them before choosing. They were very much on the fence saying that Cisco and Brocade both have serious issues. They push their environment very hard so they have experienced some really fun bugs with both systems. Performance goes to Brocade though without question.

        Also, we are currently testing VMWare using MCT… We’re hoping that works out well. You’re right though… the devil in the details remains to be seen in our environment. I’ll post back here when the honeymoon is over.

  • leigh porter

    I also moved to Brocade from Cisco for a new deployment. We have MLX4s MLX8s and CER2024 boxes. The MLX4 platform is a great asset as it combines the good Layer 2 featureset and good Layer 3 performance.

    The MCT works like a dream and has simplified network design by removing a need for multiple subnets across two switches for redundancy. I now have a single subnet between core devices and use MCT for redundancy.

    Brocade support has been fantastic even though w ehave had no issues, their support with LAB time and engineering resource has been better than any vendor I have used.

    We have a very mixed environemt with HP blades and iSCSI storage, VMware, Juniper routers and lots of MPLS. The MLX4 integrates and inteoperates on all levels.