What Cisco Has to Do to Win the Blade Server Market

Over the past several months, there has been a lot of discussions about how 2011 is the year Cisco will become a leader in the blade server space.  There’s no doubt that there are a lot of customers who have moved to UCS,  but in reality there are a few other things that Cisco will need to do to win the top spot.  Today I’m going to discuss a few of these things.

The following are my recommendations / suggestions for Cisco.  The intention is not to bash anyone at Cisco, but to provide an outsider’s perspective of what I think needs to happen if Cisco wants to increase their UCS blade server market share.

Focus on Product Branding.  When you ask people what “Proliant” is, a majority of the people know it is an HP server brand.  On the contrary, ask people what “UCS” is and fewer people know.  Why?  Partially because HP has been in the server space for many years, but it’s also due to a lack of branding by Cisco.  Sure, everyone knows the name Cisco – even commoners (people outside of the data center) recognize the logo from TV shows like 24 and movies.  Cisco does a GREAT job of getting the company name out, but needs to focus on getting the UCS brand out there a little more.

Create Public Configuration and Reference Tools.  I’m a hands-on type of guy, and when I want to see what my server options are, I want to go to the manufacturers’ web sites and configure the systems myself.  IBM has a nice stand-alone product configuration tool, while HP and Dell have great web-based tools that can help you design a blade server system.  When it comes to reference tools, HP has QuickSpecs while IBM has a very informative Redbooks collection.  However when we look at Cisco, the details and list price of a UCS configuration is held within the arms of the Cisco sales teams or Certified Partners.  I don’t quite understand the reasoning of this, but if Cisco shared their tools and materials to the general public it could increase the interest in the product.

Change the UCS Partner Certification Requirements.  The last time I checked, for a business partner to become a full-blown certified partner with Cisco for the UCS product, they had to have a single engineer that was Network, Storage, Virtualization and Server certified.  While I get the idea that you want one person to be able to do everything, the reality is that is a challenging request.  For partners who have invested into IBM, HP and Dell products, getting certified on Cisco could be costly, as you have to ask your server/virtualization guy to become a network guy.  IBM, HP and Dell have similar requirements, but they allow for someone to have network certification, someone to have storage certification, etc.  Perhaps Cisco should follow suite and allow for the partner to have one person for each area instead of all-in-one.  This would attract more business partners to sell Cisco UCS which leads to more opportunities for Cisco to grow their market share.

Focus on SMB Market.  Who is the target market for UCS?  Obviously we see UCS making traction in large environments like co-location hosting facitilities where there is a large amount of virtual hosts, but what about the Small-to-Medium Business (SMB) Market?  When I look at the UCS model, I see a good fit where there are 8 or more VMware / Hyper-V hosts needed, but does Cisco UCS work in environments with 3 or 4 virtualization hosts?  Since Cisco UCS was first released in early 2009 the messaging was focused on “the Enterprise” data center.  If Cisco wants to take market share, they need to find a way to attract the smaller SMB space.

Get Bloggers Involved.  This last suggestion is based on the success that I’ve seen with HP.  They have created multiple “Technology Days” where they have invited bloggers of all sorts to participate.  In this event they provide access to the engineers who create their server and storage products.  In return, the bloggers are able to write up articles that discuss their opinions about the HP products.  I know that Cisco is looking into doing this, but hope they seriously consider it if they want to promote growth of the UCS line.

Innovate.  Need to add in a PCI Express card into a UCS blade server and you’ll be out of luck.  Cisco created the c-class SERIES rack server with intentions to offer customers the ability for larger I/O expansion.  They also recently added the capability to manage the c-class series rack servers with the Cisco UCS Manager allowing one management system for both the rack and blade server environments.  But what if Cisco created a PCI Express blade like IBM and HP offer.  Sure, this would take up a blade server slot, but it would allow for those users to keep it all in a UCS chassis instead of branching in a rack platform.  If Cisco didn’t want to create a PCI Express blade, what if they created a rack-based “shell” that allowed a user to populate with B200 blade servers.  The shell could be enabled to allow a user to have a couple of PCI Express expansion slots therefore providing the customer with the ability to standardize on the blade platform.  Don’t get me wrong – I love what Cisco has done with the Extended Memory on the B-250 blade server, but in order for them to grow in the market place they are going to have to innovate a lot more.

That concludes my thoughts on what Cisco needs to do to win the blade server market.  Let me know how you think they could do it in the comments below.

30 thoughts on “What Cisco Has to Do to Win the Blade Server Market

  1. Jeff Allen

    Disclaimer: I work for Cisco
    Great article Kevin! I agree with much of what you have said and look for some of these to be tackled during 2011. The only one we won’t be doing is adding “c-class” to UCS since c-class belongs to HP! :) UCS uses “c-series” instead :)
    You are correct that if a customer needs a special PCI card, they should use one of the growing number of c-series rack servers and then manage them via UCS Manager. 99% of all blade customers have rack servers already so this is nothing new to them. And because Cisco’s rack and blade servers are managed from the exact same console, you can move Service Profiles between the two form factors. Unlike any other vendor out there, the mgmt UI and tools are identical between the two types of servers.
    I originally agreed with your thoughts on a PCI blade when I first arrived at Cisco. But seeing 1st hand how smoothly rack servers integrate into UCS Manager, I think it would be a very small seller. If they need a PCI card, they’ll buy a server capable of delivering that feature without creating a one-off design inside their blade chassis. HP and IBM need this to fit within the same chassis because the tools used to manage their blades (VC, VCEM, OA, etc) are unique to their blades and blades only.

    Lastly, don’t forget that every PCI expansion blade in an HP or IBM chassis takes up 2 switch ports in whatever interconnect you chose. If those ports are on a FlexFabric module, you are talking about $1100 per port – no thanks for me.

    Anyway – like I said, great article.

  2. Pingback: Kevin Houston

  3. Pingback: Scott Hanson

  4. Pingback: unix player

  5. Pingback: Eric Markwardt

  6. Pingback: Kevin Houston

  7. Pingback: Toru OZAKI

  8. Jed Scaramella

    Nice insight as always.
    Interesting comments on the brand, from where I sit at IDC Cisco entering the server market was the talk of the industry for a year. Even their competitors wouldn’t stop talking about them… But valid point that “UCS” may not have brand equity.
    I will say that in my conversations with customers and financial analysts, they are top of mind (along with HP) in “Convergence”

  9. Pingback: Back From the Pile: Interesting Links, February 4, 2011 – Stephen Foskett, Pack Rat

  10. Pingback: Mark S A Smith

  11. Dick Diver

    I really appreciate some of the innovative features of the UCS platform, but think that this move is undermining Cisco as a whole by battling HP head-to-head. Ten years ago everyone thought Sun would usurp HP in the data center with their servers, but then Sun was beaten down to almost life support. Then the market believed that Dell was going to beat HP with their innovations in PC supply chain, and they too got flipped by HP. Now the story is that Cisco will beat HP in the data center. The reason this won’t happen is simple – Cisco can’t draw enough oxygen in the profit margin space that HP thrives in. Cisco was accustomed to commanding a super-premium for their networking gear and were virtually unchallenged. The x86 server space offers razor thin margins and always will. I don’t know what Cisco was thinking moving into this war zone, but it brought HP into the Networking space. Now Cisco’s profit margins are deteriorating (see yesterday’s CSCO quarterly results, and the previous quarter, and the previous quarter) rapidly. Let’s say Cisco makes 75% (arbitrary) margin on their networking gear – their bread and butter. HP looks at that and says they’re happy with 10% profit margin on networking. And that’s when they have you. Cisco is forced to sell UCS close to cost, or worse, drop proof-of-concepts all over the country. Meanwhile, as networking bids come due, HP puts 3com in front of the customer and can offer them cost savings that Cisco isn’t capable of absorbing. Someone needs to go back in time and tell John Chambers never to wrestle with a pig – you both get dirty and the pig likes it. Anyway, I applaud the new technology as it makes the blade space better as a whole. Really looking forward to the next gen enclosures that all of the players will be bringing out soon.

  12. Pingback: Tomas Novak

  13. Bow Bow

    Kevin, i agree with you – ProLiant is a brand name most data center customers understand, with other names like iLO and Smart Array. Compaq was already quite successful in this even before hp acquires them.
    P:S: it is ProLiant and not Proliant ;)
    Cisco does bring some new ideas which provides creative competition to the industry, a good thing for customers, HP and other vendors who are serious in enterprise servers market. However, one thing I personally dislike about some Cisco marketing/bloggers is the amount of marketing FUD and distorted information they gave in an attempt to overly discredit HP or any other blade vendors, that earns them no respect in my book for a new comer to the blade market.

  14. Mark


    Disclaimer, I work for Cisco.

    The Cisco/HP “head to head” battle was going to happen regardless if Cisco entered the server market or not. The fact is blade servers changed the network access layer (the largest component of any network measured by ports). In many cases, non-Cisco switches grew market share in the blade switch market. Also, blade server vendors offered organic switches for their blade servers. Blade servers are becoming a significant share of x86 servers, HP has gained significant blade server share (approaching 50%), and Virtual Connect has a significant share of access layer switching for the C-Class. HP led with BNT blade switches with many customers on the P-Class and IBM led with BNT blade switches with many customers prior to Cisco’s release of UCS.

    HP was a strong player in Ethernet networking with ProCurve (especially in higher education and state and local government customers) prior to Cisco releasing UCS, and prior to HP acquiring H3C.

    As for “razor thin margins”, while this is true in generic, undifferentiated rack-mount servers, this is not true in blade servers, nor in blade server network switches. HP C-Class enclosures and Virtual Connect access network switches command Cisco-like margins. Blended blade server, enclosure, and blade switch margins are significantly higher than rack-mount server margins. And don’t get me started on the ink and toner cartridge margins.

    HP has been driving towards and end-to-end IT vendor for some time. It is not new. Rumors of Dell buying a switch vendor like Force 10 were floating around three or four years ago. While it did not happen then, t still could happen.

    This is the new order in the IT world. There are three vendor models emerging. The first is the integrated server vendor and outsourcer vendor. That was originally done by IBM with IBM Global Services. HP flirted with this model under Carly but did not fulfill it until Hurd acquired EDS. This led Dell to acquire Perot. The second is the integrated server/network vendor. Cisco and HP fit this model. The third is the integrated server and middleware vendor. IBM also follows this model with DB2 and WebSphere, and Oracle is now engaged in this model after acquiring Sun.

    The jury is still out on what product and services areas to partner vs. where to integrate. Margins around specific products are less relevant than overall company margins (networking and software command higher margins than servers and services).

    But, due to blade servers and virtualization, the server and the access layer have effectively merged. It only makes sense for network vendors and server vendors to align, either by organic R&D (HP Virtual Connect, Cisco UCS), or through acquisition (IBM and BNT).

  15. Dick Diver


    Thanks for the thoughtful response. Disclaimer: I sold out my CSCO stock several years ago (at $24 and change, thankfully) when they moved into the server space.

    HP is a man-o-war, slow to turn around, but once its guns are trained on the target brings massive firepower. The effects are easily seen in Cisco’s last three quarterly reviews – they are getting knee-capped on profit margins. What does this matter for Cisco in general, and their servers in particular? Cisco is priced in the market as a high margin equity; any threat to those margins radically alters the market price. Playbook goes three ways: 1) maintain the margin but lose market share, 2) gain market share but lose the margin, or 3) decrease the cost side of the equation in order to support margin- that means cut R&D, make the employees of Cisco eat the cost, slower times to market, and etc. When stocks go seriously south (like CSCO’s 14% drop last week), instability results.

    In other words, it really doesn’t matter how well Cisco does on the server front because they are suffering a rear-guard action on margins in their core business (networking). From your perspective, I recognize that you can only do the best you can from where you’re at and hope that management can escape what appears to be a carefully laid trap.

  16. Kevin Houston

    EXCELLENT points regarding #Cisco in the x86 market. You are right in that “x86 server space offers razor thin margins” – this is something Cisco has never had to experience, so they will either change the industry and allow for the x86 market to get more margin (like in the late 90’s) or they’ll crash and burn under the pressure of high profit requirements from their shareholders. I guess only time will tell. Thanks for your comments, and thanks for reading!

  17. Kevin Houston

    Mark, thanks for putting a lot of thought into your comment regarding #Cisco vs #HP. I have to disagree with your comments about high margins on blade servers. Yes, you can get to higher margins with blades, in comparison to rack servers, but to do so you have be Elite certified, get new customer registration, deal registration, etc. These types of programs take time to develop and until Cisco is able to create a profitable partner channel, I think it’s going to be challenging for resellers to make large margins under the Cisco UCS umbrella. Thanks for reading!

  18. Pingback: Calvin Zito

  19. Lee Johns

    Full Disclosure from me also. HP Employee. I also think it is a thougtful post and so are the comments. I think it is interesting that the post from Jeff Allen tries to make the case that HP is expensive for networking ports. Oh the irony!

  20. Pingback: Blades Made Simple™ » Blog Archive » This Day in History: “HP Claims Cisco UCS Will Be Dead in 1 Year”

  21. NookinDeNiro

    Gartner results for CQ1 x86 server market came out last week, Cisco did not even make it into the top ten list. Still lots of hype around Cisco but very little traction in almost 2 years since launch, would not be surprised if its one of the groups John Chambers has said will be cut.

  22. Pingback: my pc login

Comments are closed.