Cisco bug and bankruptcy

[wintermute000]

https://www.bloomberg.com/news/articles/2016-09-08/cisco-s-network-bugs-are-front-and-center-in-bankruptcy-fight

[dlots]

Over at  http://www.businessinsider.com/cisco-bug-blamed-for-machine-zone-bankruptcy-2016-9 you'll see that they had multiple outages over 8 months, not just 1 big outage. If you are getting 4 Million dollars a month and your required to have 99.999 up-time, and this one contract is 80% of your business, then IMO you should be able to loose an entire Data-Center and not have an issue, much less 1-2 Nexus 3Ks.

[NetworkGroover]

I only have an outsider view - does anyone else who regularly works with their products feel like they're going downhill or is it just me?

[deanwebb]

I only have an outsider view - does anyone else who regularly works with their products feel like they're going downhill or is it just me?

Did you see the list of products affected by the CMS telnet vulnerability?

Seriously, though, Cisco stuff seems to be made under the motto, "That'll do." The old stuff works well enough, but that new kit... I'm not so sure about it all... IOS 12.2(55) is a very comfortable rev of IOS. ASA 9.5.2 is where the least bugs are at. Beyond those, it can be a little iffy about what the gear will do. The thread here about ASA 9.7 code is illustrative of that.

I think you've brought up a general concern. I don't plan to moonlight as a marketing writer for Arista, but I would bet that their collective jobs would be pretty easy if they put the list of Cisco vulns up against Arista vulns.

[dlots]

I would defiantly say they are going down hill.  We just replaced some gear in a data-center with ~6 years of up-time (to long I know), but within 4 hours we hit a bug and took down a bunch of stuff.  I was at my last job for ~5 years, and we didn't have all that much gear there, but what we did have we had alot of config on the boxes, and I don't think there was ever a point in time where we didn't have in a ticket (most of the time a lvl 1 ticket) about something that turned out to be a Cisco bug... Seriously how does "you can't have QoS on 2 different types of interfaces" (Gig/T1 for this example) get though testing, much less stay a bug for ~9 months.

[NetworkGroover]

Yeah I know as much as it seems like it, I'm not trying to, at least directly, make this a Cisco vs. Arista argument/shameless plug.  Just trying to put a feeler out there for current temp in and out of the data center.

[Otanx]

It isn't just Cisco. All of the incumbents seem to be having more and more issues. Some of those issues are old legacy code bases that are left around for compatibility. Some of it is the pace of new feature release, and what I think is reducing in house testing to speed up that release schedule. Smaller companies/open source projects that are not carrying that legacy baggage can move faster.

-Otanx

[wintermute000]

everyone thinks Cisco's code quality has gone down the drain.
There's a combination of factors here

- they have a LOT of stuff out there
- they have to support a LOT of stuff for a LOT of time
- spaghetti code base
- legacy code base
- too many acquisitions (see spaghetti and legacy code base) to integrate properly
- reducing costs for testing
- offshore development
- increasing complexity and feature-sets (see reduce costs for testing)
- massive user-base putting their kit through their paces in every conceivable scenario and combination

Having said all that, unless this N3K bug was hitting multiple switches close together, how can 1 switch failure bring the whole thing down (let alone 10 hours WTF) and if it did why can't you just swap the damned thing out. They're 1/2RU things right? not even the excuse of a big monolithic N7K/6500 with hundreds of leads going every which way but loose / weighs a ton


I was talking to a SE from another hawt SD-type up and comer and told an anecdote from some ex-Cisco management: when they were pushing for resourcing to start a ground up rewrite of teh code behind XYZ as the current stack was obviously busted and could not be duct taped further in a sane or scalable manner, their request was knocked back on the logic that their revenue for segment XYZ was still increasing so why change anything. And you wonder why a horde of once startups are starting to eat their lunch in heaps of market segments

[deanwebb]

Wintermute's analysis is dead on the money. 

(I just realized how awful that must sound to someone who isn't familiar with English slang... for the readers that weren't familiar with that phrase, it means "absolutely correct.")

Cisco *used* to be the gang that was coming up with neato new nifty stuff, but now they're like a big car company. They need people to keep buying what they're making and innovation that isn't backward-compatible won't sell to their core audience.

Any company that grows and innovates more for its acquisitions has already peaked. It can still survive, but not with the growth it once had. Ask Microsoft.