re: WorldCom's IP Outages: Whodunnit?I agree with Skeptic about Cisco's track record in bringing a new OS to the market. From experience, however, I expect to see bugs in every piece of software this complex. If you think IOS or JUNOS are bug free, then you're dreaming. What I need is a stable platform, with stable code, that when I have a problem I can get it fixed quickly. To be honest, I don't care how the OS runs... the bottom line is stability in a defined xyz environment. My experience with Cisco is that my problems get fixed and I do speak to very knowledgeable people when I need them.
What I know of JUNOS code is good stability, however the support coverage, from local support to phone support, is not enough. Juniper has a long way to go. Product wise, they are on the right track. In fact I think they are the only viable alternative to Cisco in core routing.
I do have some experience with other vendors, and it was a nightmare.
Proper design, proper staging and testing, and very good support is the key. The comment about TDM circuits being more stable... is a good point. They are more stable, no doubt about it. However they the economics don't work.
We need solutions that improve CAPEX and OPEX in SP and enterprise... CEOs/CTOs/CIOs etc don't buy cool technology any more. They want to know how can solution x make money or save them money. The latest T640 is nice on paper and is very cool stuff... but the market is not there.
re: WorldCom's IP Outages: Whodunnit?Some MSSBU people were suppose to work on this after a reorg a couple of years ago. I don't think much came of it. Chambers asked them to change the name but they ended up keeping it.
re: WorldCom's IP Outages: Whodunnit?The CHAOS software came out of the MSSBU, the old Stratacom group.
And from my reads, the HFR isn't running anything at all because the backplane doesn't even work.
Agree with your comment on Cisco's track record with IOS replacements. IOS is the cockroach of the industry: ugly and old, a questionable factor in the modern ecosystem, but will it ever be forced into extinction?
re: WorldCom's IP Outages: Whodunnit?I've heard CHAOS has the potential to replace IOS because it was written in the late 90's rather than the late 80's, and is fairly solid. But Cisco won't take it seriously because they can't afford it politically either internally or in the marketplace.
Interesting name they picked. I didn't know Cisco had a sense of humor/irony.
Now, will CHAOS support LAN switching gear, now the crown jewels of the company and all of its profits??
re: WorldCom's IP Outages: Whodunnit?That would be the HFR running on the QNX operating system right?
When did you hear that 400-500 people were working on that? The numbers I've heard are a whole lot less.
What I have to say about that project is that cisco's track record in doing IOS replacements isn't very good. And that if I were a service provider, I would give "new" software from cisco no more credibility than software coming from a startup.
===========================
Why Sekptic, you smug dog ;-)
What are the actual numbers of developers, anyone know?
Everyone's sw has bugs, everyone's...the issue is not to debate whicg non OS company writes the best OS!
The issue is builidng embeded systems that are made from the ground up to for HA, systems where you can debug multiple task/threads during development, where you can trace performance and memeory leaks, where you can isolate faults and take corrective measure, where you can trust the rare code that runs in overload conditions, where you can slay and restart any application at runtime without service interuption!
re: WorldCom's IP Outages: Whodunnit?IOS is the cockroach of the industry: ugly and old, a questionable factor in the modern ecosystem, but will it ever be forced into extinction?
------
Statements like these are useless and non constructive. Please state the facts not a propaganda that you're trying to push.
re: WorldCom's IP Outages: Whodunnit?The issue is builidng embeded systems that are made from the ground up to for HA, systems where you can debug multiple task/threads during development, where you can trace performance and memeory leaks, where you can isolate faults and take corrective measure, where you can trust the rare code that runs in overload conditions, where you can slay and restart any application at runtime without service interuption!
-----------
It all depends how you define HA. To me it is a about HSRP, OSPF multipath, MPLS fast reroute etc.. I agree that the OS needs to be robust to handle memory leaks, or corrupted memory, or a bug.. But in the real world... you need a system that is tested and proven.
re: WorldCom's IP Outages: Whodunnit? I think there are more implementations and running code of IPv6 than ISO CLNS, SNA, DECNET, and most others combined, and for sure IPv4 takes the cake including IPX and Appletalk and IPv6 even.
More is known about how to build resilient IP networks than any other networks. There are plenty of companies relying on the Internet for their business and credit card companies and banks are included.
Anyone who says anything differently is living in a shoebox.
Cisco IOS is going to improve, but there will be stronger innovations in other places.
CHAOS is one project that runs on 7200's only. They infiltrated that project with Java Virtual machines, etc etc and it has gone all but down the toilet. I'm not sure if IOS mx/mc/NG was a different project or not, and I haven't even seen 12.2S yet. Cisco has been trying to build better operating systems for their 7x00 and HFR products for a LONG time, and they will continue to try (and possibly fail). The quesiton remains - is it their "core" business and will they be able to keep it that way. I say "no" and "no". Chambers has always said software is their "core" business, but I sure hope he's not only talking about today's (12.0S specifically) IOS.
However, the point of this may not be in fact the operating system, but the routing protocol code itself. There are not 500 programmers working on this, otherwise I'm SURE it would NOT work.
AFAIK there were only about 20 or so coders at Cisco working on the actual routing protocol code over the many years it was developed.
Kirk LougheedBob Thomas Robert WidmerDan Tappan Srihari RamachandraEnke Chen Bob AlbrightsonDave Katz Tony LiWilber Su Kent LeungBruce Cole Paul TrainaJohn Bashinski Ravi ChandraHenry Wang Kevin Paul HerbertPhanindra Jujjavarapu Steven LuongDerek M. Yeung Senthil Sankarappan
Sorry if I missed anybody. There were a lot of important contributions to the TCP/IP world besides Cisco, I think that everybody should understand that concept. And there will be many more contributions, Cisco does not "OWN" TCP/IP like Microsoft "OWNS" Desktop or Server operating systems. It's very different.
re: WorldCom's IP Outages: Whodunnit?Shouldn't the routing protocols that build the routing tables run in user space, with the routing table itself existing the the kernel space ?
Your assumption is that the "kernel" of the router offers memory protection. This is a bad assumption. Lots of telco stuff is still running on old versions of VxWorks where the MMU is used only to increase the power budget of the chip.
On the other hand, some boffins have managed to get Linux running on a Cisco 2500. See http://www.mcvax.org/~koen/uCl.... So maybe there's hope after all.
What I know of JUNOS code is good stability, however the support coverage, from local support to phone support, is not enough. Juniper has a long way to go. Product wise, they are on the right track. In fact I think they are the only viable alternative to Cisco in core routing.
I do have some experience with other vendors, and it was a nightmare.
Proper design, proper staging and testing, and very good support is the key. The comment about TDM circuits being more stable... is a good point. They are more stable, no doubt about it. However they the economics don't work.
We need solutions that improve CAPEX and OPEX in SP and enterprise... CEOs/CTOs/CIOs etc don't buy cool technology any more. They want to know how can solution x make money or save them money.
The latest T640 is nice on paper and is very cool stuff... but the market is not there.