The market has been commenting on the rise of Juniper for some years and the fact that Cisco has not been able to stop them growing. Well, we can now see that Cisco was listening to the stock market by spending a lot of time and money developing a new operating system. Oh yeah, and some new hardware to run it on. Read all about the Cisco ASR 1000 here.
IOS performance limits
Its been clear for a long time that the conventional IOS had hit a wall in performance. The monolithic kernel has been subjected to any number of hacks to improve performance (fast switching, CEF) but ultimately, it obviously had to be discarded and a new approach built that supported modularity, improved upgrade, and other serviceability features. This has seen the release of NXOS (Nexus 7000), IOS XR (CRS-1) and now IOS XE for the ASR1000. You can probably also include IOS SX as used in C6500 and other switches as another.
Where now for network design
What might this mean for Network design and architecture ? For the next year or two, it is mostly business as usual. Cisco likes to announce products with only limited functionality in the early stages to mark out their turf, so you probably shouldn’t rush to buy one unless you have some form of passionate belief in the product. For example, BT have announced that the ASR 1000 will revolutionise some part of their network and they are going to buy lots of them. (No doubt with a nice discount and a lot of hand holding by Cisco).
Network Management
One concern is how quickly the Network Management tools will adapt and adopt new measurement and management. The new operating systems means that traditional methods of network system monitoring will need revision. Since Windows Server 2008 might be released soon, budget for software upgrades is going to be tight.
A second concern is whether they will work as advertised. Cisco has made enormous changes to software testing and the number of problems with IOS is a fraction of years gone by, but still……
Better everything
The new IOS versions offer us a lot of benefits. The focus on ISSU for specific software modules, running two versions for roll forward / roll back capability, checkpointing are welcome changes to the IOS strategy. I look forward to spending less time performing upgrades in the middle of night.
Reluctant to criticise IOS – but why so long ?
I won’t criticise IOS for its past limitations, as they were necessary at the time. Remember that IOS was developed at a time of 2 MB flash RAM, 4 MB DRAM and performance was not an issue when 128 kbps was king. I will however, be critical of why has it taken Cisco so long to modernise and update their core business ?
Cisco claims that it took five years to develop this new system. We had similar comments about the Nexus 7000. In my mind, Cisco should have had these products in place about two years ago. Procket was acquired in 2004 and formed the basis of the new ASR1000. Cisco has regularly been feted by the technology press as a leader in buying startups and integrating their assets, but this isn’t necessarily a shining example of speed to market or responding to customer’s needs.
Conclusion
So farewall IOS, its been a ten year relationship and its been sweet, if a little tiring and slow. l am sure to be spending time with the new systems and finding out about all their idiosyncrasies and foibles. I have been dating other network systems though while you were putting on your makeup and I won’t be forgetting them in a hurry.
Loading ...
About Greg Ferro
Yes – should be interesting to see the take-up on this new router. It certainly looks to be filling a big gap above the 7200 in terms of forwarding capacity.
When I saw a presentation about it, I had a concern though – there’s only dual route-processor capability in the 1006. The 1004 and 1002 use a kind of software redundancy – essentially a microkernel that runs two copies of IOS at the same time. If theres some kind of crash of that kernel, you’ve lost your resilience.
I can’t help thinking that one of the big things in this is this QuantumFlow processor. It is extremely powerful and new features can be ‘switched on’ according to your need. What this boils down to is the licensing of features more tightly, fewer code trains and more profit for Cisco. Lets hope it is less trouble for the consumer too.
I agree with you, it doesn’t quite seem to match the HA features that I would expect. I keep scratching at the itch that the Nexus 7000 and the ASR 1000 and something doesn’t seem quite right.
There is always something missing isn’t there. I suspect that the HA features were removed by marketing since they wouldn’t sell so many 7600 chassis.
The rise of marketing at Cisco does worry me. Gosh, when they released the ASR they rode a videoconferencing message over the top of it. As if anyone remotely cares about videoconferencing, the whole technology is a flub and they are hoping that marketing will pick up the numbers
sigh
I know what you mean – I suppose we all want the moon on a stick in terms of resilience, don’t we! I mean, compared to Juniper’s low-end boxes, it has more resilience, and I suppose it has more resilience than we’re used to, so things are moving in the right direction.
Actually – I had a demo of that Telepresence thing a month or so ago – until then, I didn’t care about it either. It was very impressive though – almost like being in the same room – you forgot you were on camera quite quickly. Just wish I could afford one…