What Does HP Networking Need to Do Be Successful ?

It’s not the first sale of a product that defines your success in networking products, its the second – EtherealMind.


If you want to summarise Cisco’s networking dominance then perhaps it’s the simple fact that after you have bought, administered and managed their equipment, there is no reason not to purchase Cisco again. And then they become entrenched in your processes and procedures. At this point, I feel that Cisco doesn’t have to be better, they just have to be good enough to continue succeeding.

But what does HP have to do to be successful ? What can they do to appeal to the networking engineers, administrators and designers who are going to use their products everyday ? They are going to be concerned about career development and advancement, and will need to make a significant commitment to relearn or adapt their skills. How can HP overcome these objections to using their products ?

Here is my opinion after the HP Network Tech Day in Sacramento in August 2010.

Establish a world class education program

Many people have invested significant personal time and effort to master Cisco networking and passing various certifications. These education programs build much more than just vendor loyalty, they build lasting value for customers who have networks that:

  • have more features and capability
  • work more reliably
  • are more adaptable
  • are fixed more rapidly
  • build employee loyalty and retention
  • lower support and operational costs for HP

These outcomes are why business invest significant amounts in developing their human resources.

Career Development

Many engineers are dependent on Cisco certifications for career development. And managers have responded by recognising certifications as milestones for financial recognition.

Cisco certified engineer will resist any use of HP equipment after they have committed significant time and resources to completing certifications. The use of top up or cross certifications that credit existing Cisco Certifications is move that encourages people to undertake HP certifications.

Note that this cross-certification also devalues the HP Networking branding of the exams and courseware.

Range of educational choices

It needs more than just a training course and an exam. The learning ecosystem needs textbooks and exam preparation materials from training companies. They should consider approaching Cisco training business that are disenfranchised with the current Cisco revenue model for training partners.

Overcoming the experience gap

Many people will object to using HP Networking on the basis that they have no experience on using the software. Of course, they aren’t going to use the software until you use the software. Although education programs can cross some of this gap, it doesn’t not address some vital operational challenges.

Many organisations have adopted ITIL processes concerning change management. This is usually results in pretesting and validation of any proposed change as a requirement and results in insidious damage to network upgrade requests by slowing down their ability to change the network.

For Networkers there is rarely a lab or test area available, compare this with MS Windows where anyone can install as many copies of MS Win2K3 as needed to test and prove a change. As a result, many network changes are not allowed or hard to get approved.

When engineers do not have the confidence to develop and use the advanced features, the value of your product is minimised and your sales diminish.

Release an Emulator

To break the cycle, HP Networking should release a version the Comware platform for use in some form of emulator that can be used to accompany training.

And don’t feature limit the software. Provide the full package with all the features. By all means cripple the performance for forwarding, but don’t limit the feature sets. Without access to the entire feature set, people aren’t going to use it.

Use Case

Let me provide a use case: lets say you have large Core LAN and you want to change an 802.1d Spanning Tree configuration to use 802.1s MST. Such a change can easily shutdown the entire LAN network but the SLA means you cannot have an outage. With an emulator, you can develop a design, perform the migration until you are confident that you understand the technology and rehearse the changes.

At this point, you can get a change request approved because validation has been performed and execute the work with confidence. The engineer has fully engaged with HP Networking product and added significant value to the company by using enhanced features. You have also created customer loyalty and engagement on an enduring basis.

Its expensive providing loan equipment

Executives who doubt this suggestion should remember that providing loan equipment is an expensive proposition. An emulator is an instantly scalable solution that removes customer objection from the smallest to the largest customers.

Examples

There are two examples of successful emulators. The Cisco emulator using the dynamips software (neither endorsed or supported by Cisco) has created tremendous opportunities for people to develop skills, learn and study. Many organisations use this software to pre-test network changes today.

Also look at Vyatta, who provide an open source router that can easily be run in an emulated environment and is key value proposition for many customers.

Technical White Papers

The value of technical marketing is often misunderstood. The typical comment is that White Papers are never actually used because they don’t match the final deployed network. Counter-intuitively, this is exactly correct. No two customers are ever the same, and white papers are unlikely to address a spectrum of requirements.
However, network architects and designers use the white papers extensively as the first point of research for new features and capabilities. Human nature doesn’t like to admit that they don’t know something – pride / face etc so the real value is often obscured.

Let’s just say that competitors produce white papers and continue to do so despite the cost and time taken to produce them. They wouldn’t keep doing it if it wasn’t paying off.

And Don’t Hide Them

These white papers should not be secret. Don’t release them to partners only, or certified people, or some self selecting group. Don’t make people login to receive them (so that you can track who downloaded them).

Consider the following:

  • Companies have policies that require any competitive research to kept private.
  • Designers may be acting without approval and cannot flag their intention to Sales people
  • It takes time to understand the problem before you engage the vendor. You can’t ask intelligent questions of a vendor until you have enough knowledge to be intelligent.(Catch-22)
  • Many companies will regard such requests as leaking security details.
  • It’s offensive

Make them freely available to anyone who wants to access them and keep them updated. Then we will feel confident that we understand your strategy.

Vitally important is that all documentation must be available on HTML and PDF format for searching and reading.If there is one thing I hate, it not being able to search online for content, but I also need the PDF file for offline storage and review.

Networking is a bit different

There is a divergence between networking products and server products. One of the larger challenges for HP will be to internally comprehend that difference and not get misaligned.

Because Cisco is challenging HP in one of their core areas of the servers, and because of the Virtual Connect products (being pseudo networking elements), it is possible that business management might perceive HP Networking products to be part of their server strategy and/or defending the server business.

However, Networking products have a key difference – software . The Comware and Procurve software platforms from the A and E series respectively, come bundled with and are intimately identified as “HP Products”. Compare with Microsoft Windows Server or Linux, where the server is sold as a hardware item only and the support and maintenance are not comprehensively HP product. OK, so the integration and compatibility has to be there but HP does not strategise, develop and release the software.

In this sense, Networking products are more similar to storage products where the division sells a strategy, roadmap and develops products internally. Although HP doesn’t tend to innovate on developing storage products, and prefers to buy established products such R9000 (Ibrix) and R4000 (Left Hand) and even possibly 3Par, they do continue to maintain the software.

If HP executives fail to grasp this difference (and I think they might), then HP Networking is likely to do less well.

Be Serious

Networking is an entire market segment not an “extension”. People will notice a lack of attention and stay away if HP doesn’t not attack the Networking market directly and engage fully. By all means sell the relationship that exists with the Server and Storage teams, but don’t fail to recognise that Networking is it’s own market place.

Price Reasonably

Don’t get inflated ideas about your product and attempt to price according to the Cisco RRP. Just because Cisco can make 80-90% profit margins on most networking products; this does not mean that HP can achieve anything similar.
Secondly, Cisco has much more to lose than HP. Cisco’s share price is intimately linked to its oversized profit margins. Any competitive selling price will result in Cisco reducing prices and thus validating HP Networking as a viable competitor.

Products

In a certain sense, the features on the products aren’t vital if all other areas are addressed. The success of the ProCurve networking products is proves that simple, cost effective and reasonably stable products will work for many customers. HP is going to get reasonably good sales just by offering a product (same as Cisco is getting server sales just by offering a server product, some people align completely with certain vendors).

The key factor is that they work reliably. For example, Huawei has been criticised many times for poor reliability and bad software, and there is growing resistance to using their products. It’s isn’t usually vocalised by customers as “your product is rubbish” but people can find any number of reasons and methods to not buy a specific product.
Given that I haven’t had a detailed look at the products, I can’t make any comments about whether they are worthy contenders. Certainly, the product range is wide, although short on firewalls and some other elements.

The EtherealMind View

HP has a solid foundation on which to build a successful business unit. The people I met have a strategy that addresses concerns that I would have at a business level, either as a reseller partner or as a customer and HP has a reputation for delivering products that customer will use. They may not always be “sexy shiny exciting“ but they are one the biggest server vendors because they deliver good products at good price.

While the potential is clear, the next step is to see whether the products are worthy. And that’s going to take some time to see how people’s experiences are shaped, and how many ‘war stories’ we hear, and whether the products continue to develop. The devil is now in the detail, and for engineers like me, execution matters.

Disclosure

HP Networking Tech Day is sponsored the flights, travel and accommodation for this event. I am not required to write or podcast about the event and am free to share my own thoughts and perceptions. As if I’d do anything different. I am not paid for my time and do not receive any payment for service.

About Greg Ferro

Greg Ferro is a Network Engineer/Architect, mostly focussed on Data Centre, Security Infrastructure, and recently Virtualization. He has over 20 years in IT, in wide range of employers working as a freelance consultant including Finance, Service Providers and Online Companies. He is CCIE#6920 and has a few ideas about the world, but not enough to really count.

He is a host on the Packet Pushers Podcast, blogger at EtherealMind.com and on Twitter @etherealmind and Google Plus

  • Mike Kaciuba

    The free training emulator would be a great way for HP to get a competitive edge over cisco. Nice article.

    • Tyler

      I have to agree here. We have a TON of ProCurve at the edge, but if I had a free emulator using actual production images, i would feel much more inclined to use them in production environments and throughout the LAN.

      Another thought to break into the distribution and core layers of the network…offer the layer three routing for free instead of requiring a license.

  • http://www.cisco.com Ed

    I work for cisco, and have access to various emulation tools that you’re probably aware of…but I think Dynamips is incredible. And it works because it uses the same images you actually deploy. Debug images and emulators are normally for developers and not relevant to engineers – do you care if ip_function_x returns 1, and do you really know if that’s a problem? Dynamips fills a hole that if filled by the vendor would probably be too restricted to be actually useful.

    In the end though, the real problem is that emulators just prove the functionality of code (yes, bgp will establish and exchange routes) but don’t address scale (can this router cpu handle a flapping neighbor). I’d pnly test my config on the same image I’m going to deploy, and test scale with the actual platform.

    • http://etherealmind.com Greg Ferro

      Not every company can afford a lab of networking equipment, especially Cisco equipment. In this case, an emulator might be the only alternative. As someone who once worked extensively in medium companies where experience was hard to get, an emulator would have been an invaluable resource to getting upgrades and changes to routing, or switching done correctly the first time.

      If your company needs to test at scale, then the only correct solution is to use the actual equipment and have the money to fund such an activity. In the real world, that doesn’t happen often.

      Emulators are vital for delivering results to the business.

    • Oliver Gorwits

      I think that the days of Dynamips may be numbered. There have long been limitations in switching platform support but now IOS (15+) and NX-OS are running on platforms it cannot emulate. Soon many network engineers will be much happier migrating to another vendor that they can officially lab up in emulation. It will be a big win for any vendor that pushes emulation publicly. An excellent point, Greg.

  • Mike Kaciuba

    Hi Ed,

    Your points do make sense in that emulating some environments for production testing may leave too many critical issues (cpu, throughput, etc) untested. A little off topic but how do you feel about an emulated but capped version of IOS just for training purposes? You may have seen Greg’s petition?

    http://etherealmind.com/petition/

    ?

  • http://wikibon.org/blog Stu Miniman

    Excellent post – networking people are wary of “outsiders” which they consider HP (servers) and Brocade (storage networking)

    • http://etherealmind.com Greg Ferro

      Thankyou.

      We should remember that twenty years ago, all networking was multivendor. And while the vendors build stacks in an attempt to sell solutions on perceived value of low risk, history tells us that it’s likely to be just a temporary trend to have a single vendor.

      It’s all different now.

  • HP has an uphill climb

    I have friends that work for HP and they wouldn’t touch the networking gear (not talking 3com stuff) at all. Product has tons of problems and the support organization is terrible.

    HP should of let Dell buy 3PAR and bought brocade. There was already a relationship between HP and Foundry, and brocade is a great player for SAN pieces. That would of made HP more competitive then buying 3par.

    • http://etherealmind.com Greg Ferro

      The ProCurve gear has been good enough for SME use, or maybe even Access switching. I wouldn’t use ProCurve for more than that because stability has not been proven. The 3Com/H3C reputation appears to be solid in the Asia Pacific region where they have an substantial market share ( Cisco is not totally dominant in Asia excluding Australia and Taiwan).

      There are lots of reasons that HP did not buy Brocade. First, Brocade is only a data centre product not an enterprise networking company. Second, Brocade is having trouble making the transition from Fibre Channel to Ethernet. After purchasing Foundry they haven’t released any products, and their share price continues to drop as people stop buying FibreChannel products. The Foundry product wasn’t a very good buy, and they overpaid for it.

      So buying Brocade isn’t good for anyone. At some point it will get cheap enough to make it worthwhile for someone, I guess. Most likely, that a couple of years away since their revenue is dropping so fast.

      • http://www.affirmedsystems.com JOseph

        HP Procurve is at least as stable as the Cisco. IMHO.

        I use Procurve at many trading firms and market maker’s colocated datacenters at different fiancial exchanges.

        never had any downtime.

        I really only use Cisco switch products for 6509E type deployments where its obviously the best option- but for a few servers or iscsi sans the HP Procurve 3500yl is the way to go

  • Mike

    Also it would be very helpful if HP didn’t have a bug the requires all MAC’s to be globally unique in the CAM table regardless of the vlan the device is in. This counts them out in a lot of my installs for the small to medium business. I haven’t found an easy way to identify whuch models are single FDB instances and which aren’t.

    From –
    http://www.hp.com/rnd/support/faqs/5300xl.htm

    : Are there any interoperability restrictions or issues when interconnecting a 1600m/2400m/2424m/25xx/4000m/8000m switch and a Series 5300xl switch?
    These two switch families have significantly different architectures. The 5300xl series switches use a multiple forwarding database utilizing a single MAC address and multiple VIDs. Switches using a multiple forwarding database can properly interpret multiple occurrences of the same MAC address received from the same device, but on different ports belonging to different VLANs. However, switches having a single forwarding database do not have this capability and interpret multiple instances of a MAC address on different ports as a “duplicate MAC address” error condition. As a result, using a switch with a single forwarding database may have interoperability restrictions when deploying features such as multiple VLANs and XRRP.

    • Jeroen van Ingen

      The gear with a single (shared) forwarding table Mike describes has been obsolete for years… I have to admit, we still have some 11yo HP 4000 switches in production with this design, but all switches after the ancient Voyager series (1600m/2400m/2424m/4000m/8000m) had separate forwarding tables per VLAN.

      I’m responsible for about 400 HP ProCurve switches in our university; while they don’t match the feature richness of eg Cisco, I really like the products as a whole.

      Yes, we’ve seen several bugs, but that’s the same for every vendor; we’ve had our share on Cisco gear too. And the support organisation is getter better and better. In short, I still like working with HP’s ProCurve line, especially for access switching.

  • Pingback: evilrouters.net » Net Field Day ñ Intro and Day 1

  • http://www.affirmedsystems.com JOseph

    Uh, reality check…

    HP is already successful in networking… I, and many other network integrators already use them more than cisco for all small business switch deployments (under 100 users).

    Cisco is like 4 times the price for a 24 or 48 port PoE switch, and clients will simply not pay for Cisco when they really only need simple gigabit PoE networking.

    This year I have have probably installed 50 sites with HP switches and 2 with Cisco.

    HP Procurve is pretty much my standard.

  • Pingback: HP ExpertONE certifications announced. ñ My Etherealmind

  • http://www.celikalper.com Alper

    great post again !

    But one question; if you compare HP with Cisco only from support point of view, what are your comments?

    • http://etherealmind.com Greg Ferro

      Only time will tell. On support, HP is talking a good game, but I don’ t have any personal experience of it. I’ve only heard bad stories, but no one talks much about the good ones so I’ll reserve judgement for sometime in the future.

      • http://www.celikalper.com Alper

        Thanks for the answer Greg. I can only say it is not easy to compare services since all vendors have different ìresponseî and ìrepairî time commitment and it is most of the time “apple to pear” comparison. Well, I agree with you on the fact that the time will tell us more about this competition but hopefully the real winner will be the customers.