Will New England Ever Have a Cisco?

COMMENTS

Newest First | Oldest First | Threaded View

<< < Page 2 / 3 > >>

paolo.franzoi 12/5/2012 | 3:15:17 PM

re: Will New England Ever Have a Cisco?
The problem that I had with the Wellfleet UI is that I found that getting two brand new routers of the same hardware config to repeatedly do the same thing was nigh on impossible. Cisco had lower performance products but they were more reliable. This was back in the early 90s before the Bay Networks fiasco.

I could deal with cryptic, but I found the Wellfleet software was like a race horse. Fast, but flighty.

Ah the heady days of arguing IP versus IPX versus DECNET versus ARCNET. Let us not forget the LU6.2/PU2.1 and Token Ring. Fine days of chaos. Give me VTAM or give me death!

seven

Reply | Post Message | MESSAGES LIST | START A BOARD

fgoldstein 12/5/2012 | 3:15:14 PM

re: Will New England Ever Have a Cisco? Seven, I up-rated you this time... I did not deal with the UIs and TIs myself. But I worked with people who did, and with some who resold Wellfleet (at BBN). The products started out well but as we understood it, the software did not age well. So by Release 5 or so it was getting crufty. Then they missed a release or so when they realized it needed a big rewrite. (Shades of Mozilla!) This was during the mid-1990s early Internet boom, when the router business was red hot. Buyers shifted to Cisco and Wellfleet/Bay never got back the momentum.

Going back... people bemoaning New England's ability to support startups forget that it used to be "the" place. Look at DEC, for instance, which was a model of the successful venture-funded firm. During the minicomputer years, successful startups were all over the place. I rode DEC's wave up and down, and suspect that other local companies had similar issues. For one thing there was a lot of resistance to cannibalizing one's own markets. So DEC fought PCs far too long. HP, perhaps more typical of the left coast, didn't worry about lower margins; they saw the wave and just figured out how to survive it.

Reply | Post Message | MESSAGES LIST | START A BOARD

gbennett 12/5/2012 | 3:15:13 PM

re: Will New England Ever Have a Cisco? Comrades,
There were two reasons why Wellfleet could never catch up with Cisco in terms of features:

1. Wellfleet started after Cisco, and had fewer engineers working the code. So all things being equal they couldn't catch up.

2. But things weren't equal. While Cisco had a single CPU architecture (albeit with a fancy cache controller), Wellfleet was designing for a multiprocessor platform. That gave Wellfleet a performance advantage, and a more consistent throughput over multiple protocols and traffic patterns, but it meant features took a lot longer to develop and test.

fgoldstein's point about the code going "crufty" after V5 is because Wellfleet was designing the new software (V7) for a new hardware platform (the Backbone Node). While the BN itself turned out OK (if rather late), Wellfleet had committed to its customers that the V7 code written for the Backbone Node would be backwards compatible with the earlier VME-bus platforms. I guess that promise cost an awful lot of engineering time, but the company felt strongly about supporting its installed base, and didn't want the drain of running with two code bases.

Cheers,
Geoff

Reply | Post Message | MESSAGES LIST | START A BOARD

mu-law 12/5/2012 | 3:15:12 PM

re: Will New England Ever Have a Cisco? Every time, the surveys incidicated that the customers prefered the UI (site manager).

Site mangler? Aw crap, we're still in denial!

There was one and only one reason for the demise of Wellfleet, with otherwise technically superior hardware: site mangler. Sure there were other things wrong, but lots of folks succeed in spite of themselves; this was the one fatal strategic error that was somehow traceable to every lost sale, purchasing decision, etc.

This occurrence has served as an object lesson for every packet network element manufacturer that has existed since, and is why a legacy of distrust exists among carriers and broader router buyers alike for standalone GUI based ems systems.

You make a statement when you produce a product that has one and only one facility for configuration and maintenance. (Note that this was true of both Cisco and Wellfleet) And you make an impression when that facility is unstable, inaccessible, poorly controlled, disorganized, and otherwise difficult to use.

Reply | Post Message | MESSAGES LIST | START A BOARD

rjmcmahon 12/5/2012 | 3:15:12 PM

re: Will New England Ever Have a Cisco? There were two reasons why Wellfleet could never catch up with Cisco in terms of features:

Forget about catch up, does Wellfleet exist anymore? Nobody could argue that there wasn't a market for router suppliers as evidenced by the creation and continued existence of Juniper. What happened? Was it a UI problem, maybe some folks decided to cash out, or some other set of reasons? Don't really know but my money wouldn't be on a UI (or a distributed compute problem.)

Reply | Post Message | MESSAGES LIST | START A BOARD

gbennett 12/5/2012 | 3:15:11 PM

re: Will New England Ever Have a Cisco? Comrades,

A couple of posters here have highlighted the lack of usability of the Wellfleet UI as a reason that Wellfleet and Bay Networks failed. I've mentioned the multi-CPU architecture and how it tended to slow down feature development.

I'd like to put it to you that neither of these were the "killer blow" to Wellfleet/Bay. Here's my 2 cents.

October 1993. When Wellfleet and Synoptics merged, we had the #2 router company (with a growing market share) and the #1 hub company getting together. From a product point of view it should have been an ideal marriage. I think there were two critical areas in which the overall company came to grief.

<center>1: Product Mix</center>
OK, so Wellfleet sold routers. Synoptics sold mainly hubs.

Between us we've highlighted some of the Wellfleet product deficiencies in other posts in this thread, and I can tell you there were plenty of others :-)

In the spirit of balance let me take a look at the Synoptics product deficiencies.

At that Synoptics was between its old hub product line and it's new products. The old products were the 3000 series - very widely deployed, stable products, feature-rich - BUT getting old. Competitors had begun to introduce new, entry level "stackable" hubs (I think it was Cabletron who were first here).

Synoptics had new products coming along GÇô the 5000 series (big, data centre hub), and the D5000 (D for GÇ£distributed), which was a smaller stackable that was intended to be deployed along with the big 5000. Both products were seriously delayed. In addition to the usual engineering screwups, Synoptics had to re-think its integrated router strategy on the new products. You see Synoptics had been a big Cisco reseller, and had even integrated a low end Cisco router into the 3000 series hub. Routers were becoming commonplace features in hubs by then, and the flip between possible routing suppliers couldnGÇÖt have come at a worse time.

So the Bay Networks sales folks only had the old 3000 series hubs to sell. And because these were getting old, they were shipped out at very low margins to be competitive.

But Synoptics also had a range of sexy new products in the pipeline. They had developed the worldGÇÖs first GÇ£Ethernet SwitchGÇ¥. The GÇ£LatticeSwitchGÇ¥ was a 10/100Mbit/s design and it looked brilliant GÇô on paper. Unfortunately the designers of the LatticeSwitch had forgotten to put buffers in the data path. So when traffic was sent in a burst from a Fast Ethernet server to an Ethernet workstation, the switch got clogged up, and all the bits fell on the floor. The LatticeSwitch had to be redesigned GÇô not once, but TWICE before it worked acceptably well. By then Cisco had bought Kalpana and the rest is history.

But never fear GÇô Synoptics had also developed one of the first ATM LAN switches - GÇ£LatticeCellGÇ¥. I donGÇÖt want to go into too much detail, but when I left Bay to join FORE Systems in 1995 I realised that the LatticeCell was like a bow and arrow compared to the AK-47 that FORE had developed. By May of 1995 Bay was able to finally brush the LatticeSwitch into the trash after it acquired Centillion. I left Bay to go to FORE in June of 1995 so my first hand knowledge ends at that point.

<center>Sales Strategy</center>
Q4 1993. Sales management at the fledgling Bay Networks obviously wanted to make the merger a success. So they devised a cunning plan. Offer the sales guys 4x accelerators if they hit 150% of targets. Sure enough, the sales guys went mad pulling every scrap of business out of the market. And interestingly enough Bay never set separate targets for the router and hub product lines. It was all one, big number. The sales cycle for a router at that time was 6-12 months, and required a system level knowledge. The sales cycle for a hub was the duration of a phone call and simply required the ability to discount until it hurt.

1994. The first year was a bumper success for Bay in terms of revenue. But to win all that hub business with the old 3000s meant that margins were cut to the bone. And because the former Wellfleet sales guys had been too busy selling hubs to make their massive bonuses, they hadnGÇÖt bother to sell any routers. Bay Networks lost 15% of market share to Cisco in that year.

<center>Company Strategy</center>
Bay should have been the ideal marriage. It turned out to be an ideal MBA case study on how to screw up a high tech merger.

There was a total lack of leadership from the top. Andy Ludwick (from Synoptics) had taken the CEO position, and Paul Severino (from Wellfleet) the Chairman position. Paul is one of the great innovators in the industry with a set of IPO successes behind him. But heGÇÖs not the megalomaniacal type and I think he was ready to settle down and spend some time with his grandkids after a hectic few years. IGÇÖm sure Andy is a very capable guy, but really is a Layer 2 thinker. Neither of them were able to stop their respective engineering and product management groups (East Coast vs West Coast) from squabbling like kids, and building their own empires.

For instance. A hub was the ideal place to put routing intelligence, right? So why was there never a Wellfleet-developed router card for either the 5000 or D5000 hub families?

Wellfleet router code was as solid as a rock by 1995, and yet Bay chose to make the Rapid City acquisition to make up for the LatticeSwitch screwup, and it was the must less stable and much less feature-rich Rapid City code that went forward into next generation Bay products.

<center>Succession Strategy</center>
1995. In late GÇÖ95 the Bay folks realised they had no leadership and brought in Dave House, who had been #2 at Intel. A lot has been written about DaveGÇÖs wild lifestyle, and IGÇÖm trying to stick to first hand knowledge in this post.

Whatever you think about DaveGÇÖs character, he did manage to unload Bay to Nortel for the staggering sum of $9.1B. He made a great deal for a sucker price, and Nortel pretty much frittered away the acquisition.

Since I wasnGÇÖt at Nortel at the time, I wonder if anyone can take up where IGÇÖm leaving off. So what went wrong for the Nortel Enterprise Division?

Cheers,
Geoff

Reply | Post Message | MESSAGES LIST | START A BOARD

gbennett 12/5/2012 | 3:15:11 PM

re: Will New England Ever Have a Cisco? Hi mu-law,

The TI, that's what it was called! I was wracking what's left of my alcohol-addled brain trying to remember :-)

Gosh yes, it was "basic", wasn't it? But I think the design was more suited to a machine-machine interface than Cisco's CLI. I think it's interesting that the industry is still obliged to use CLI as a machine-machine interface when SNMP should have taken over than function years ago.

I'm going to put a different view of why Wellfleet collapsed in a separate post.

Cheers,
Geoff

Reply | Post Message | MESSAGES LIST | START A BOARD

somedumbPM 12/5/2012 | 3:15:10 PM

re: Will New England Ever Have a Cisco? I was at Nortel at the time of the aquisition, but I was pretty much a DMS and AccessNode guy at the time. This was part of Roth's right angle turn *shivers*. The only thing of interest I remember is something being said about some IP code stream that made it all worthwhile - it never did sink in or make any sense to me and I just turned another year older today so maybe I do not remember correctly.

Anywho there are 3 BCNs ticking away where I currently work. And ya the Site Mangler is horrid. All of the circuits have been moved off except 1 PTP T. The other Frame Ts went to a set of Cisco 3745s at diverse locations and PTPs to one or the other, but we cannot convince one end user to move so I can pull the things out and I cannot force them due to a layer 8 (political) issue. They do not trust the new gear - hehe.

Reply | Post Message | MESSAGES LIST | START A BOARD

netskeptic 12/5/2012 | 3:15:08 PM

re: Will New England Ever Have a Cisco? It seems to me that West Coast folks are way ahead in understanding the significance of software component(*), while NE folks are concentrated on moving bits over wire.

Reply | Post Message | MESSAGES LIST | START A BOARD

gbennett 12/5/2012 | 3:15:06 PM

re: Will New England Ever Have a Cisco? Good points.

In fact if you look at Wellfleet's router strategy circa 1993 (the year of the merger), it was kind of in neutral. The only "strategy" was to beat Cisco, and little by little we were starting to gain some market share. By the way, I'm not saying that we were taking that share from Cisco, but in 1993 we were probably taking it from the "also rans" of the router market at that time - the remnants of DEC, 3-Com, etc. I'm pretty sure Cisco was increasing its market share too.

In contrast to Wellfleet, Cisco was busy innovating in the Internet space with BGP. The time I spent asking the Wellfleet US product managers for BGP!!! Don't get me started!

Cisco also innovated, as ozip pointed out, in the IBM space. When I was at Heavy Reading I had the opportunity to interview Cliff Meltzer of Cisco. He's the head honcho for the management BU these days, but back in the early 90s Cliff was helping Cisco to develop an IBM channel interface into the mainframe. I remember Wellfleet's answer to that was "IBM'll never let them get away with that when they realise it'll mean fewer FEP sales". But Cisco did "get away" with it, and turned it into a major success story. I think that strategy took guts, and staying power. Initiatives like that were essentially absent within Wellfleet.

Cheers,
Geoff

Reply | Post Message | MESSAGES LIST | START A BOARD

<< < Page 2 / 3 > >>