12

For example, let's compare IP Base vs. Advanced Enterprise Services. Let's say you have advent installed, but you're only actually using the features relevant to ipbase. Other than flash memory space, would running advent when you only need ipbase decrease the router's performance in any way?

Or to put it another way, would the router's performance be enhanced (however slight) by running only the IOS feature set that you need instead of the full advent image?

2
  • Also, I'm talking pure performance only. I understand that from a security perspective, an IOS image with fewer features presents a smaller attack surface. Commented Jun 24, 2013 at 18:14
  • Short answer: going from IP Base to Adv Ent services does not consume more CPU... Adv Ent services obviously consumes more DRAM because the image has to hold the code for all those features Commented Jun 24, 2013 at 18:28

2 Answers 2

13

Strictly speaking, yes, the unused features do cost you some performance. However, "measurable" loss is questionable. Most features are initialized even if they aren't used, so that will eat into memory, and any periodic "house keeping" tasks will still get CPU time. The major features (ipv6, telephony service, routing protocols, ids/ips, etc.) have to be explicitly enabled and configured; and yes, they're very much measurable.

Is advent 10% slower than ipbase at doing "ipbase"... No. It'll take a lot longer to boot, but day-to-day operations will see very little difference. (assuming you have the memory to run advent in the first place)

4
  • This is not true. CEF path per-interface is nowadays built dynamically and doesn't contain any features that are not activated. It was true 5-6 years ago (performance-wise). Commented Jun 24, 2013 at 19:51
  • Who said anything about CEF? Or packet forwarding at all? I'm talking about hardly measurable CPU usage due to the processes for features not actively being used. Compare the output of sho proc from ipbase and advent if you doubt this. Commented Jun 24, 2013 at 21:24
  • By performance I understand pps. On a router, what kind of other performance you may think about? In IOS processes have limited set of time share of CPU, and then they're stopped to do packet processing, so the difference between IP Base and Advanced Enterprise Services from the router performance perspective is zero. Commented Jun 25, 2013 at 7:13
  • Do you have any sources to qualify this claim? Commented Mar 30, 2016 at 21:49
7

Larger images will consume more RAM - it does not impact the CPU once loaded although it will also incur a longer boot time since it of course has to be decompressed into memory and that takes time.

On more modern platforms with speedy Intel Xeons, the decompression difference isn't very significant. On older MIPS or PowerPC processors, it is.

3
  • Just curious, which line of Cisco routers have Xeon CPUS? Commented Jun 24, 2013 at 20:06
  • @WaxTrax some nexus7k and asr9k line-cards & SUPs use Xeons...I can't remember which ones off the top of my head but if you dig around ciscolive365.com enough you'll find the information Commented Jun 24, 2013 at 20:41
  • recent RSP/SUP cards do - the 6500 got the Sup2T which moved away from MIPS to Intel Xeon. to my knowledge, the ASR range has always been Intel based. Commented Jun 24, 2013 at 22:37

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.