HP NNMi9 vs 7.53 and earlier

[Narj]

Hi all,

I'm looking at piloting HP NNMi9 and was wondering whether anyone else is running this in their environment (preferably 1000+ node predominantly Cisco/Juniper setup). I'd really appreciate some feedback on functionality/stability and how it compares to previous versions.

I've had experience with implementing/maintaining versions 6.x and 7.5.3 (the latter being a complete nightmare sometimes, hence the pilot and my eagerness to upgrade). To say I've learned a *lot* of lessons and had to counter many pitfalls is an understatement.

Reading through the deployment documentation for NNMi9, it seems that HP have seemingly taken notice of a lot of customer concerns and sorted out/rewritten much that has been problematic for me and others in the past. I'm just wondering if it it's as good as they make it sound?

Your views much appreciated!

 

[Narj]

NNMi is pretty solid compared to 7.x and earlier (it was completely rewritten for nnmi8).

Thanks for the reply. NNM7.x seems a bit of a mess really with separate databases and all this clunky stuff in the middle that fails to work properly half the time... Managing it sometimes feels like firefighting.

The scaling is out of this world - and it has a decent user interface (no more home base or that dreadful MOTIF gui - though I spose that would only be applicable to you if you've got a unix system).

That's good to hear - I must say that I quite like using the X GUI for event customisation and whatnot, but that's more out of years of habit than anything else.

What OS are you deploying it on? You going to be polling or just trapping you managed estate? SNMP3 in your place? IPV6? And are you monitoring network kit only - or storage/servers/hardware alerts?

We deployed 7.53 on Solaris 10, but are looking at Linux for NNMi. According to HP, Solaris has now dropped way down the pecking order for performance and they're recommending Linux.

The SNMPv3 SPI with 7.53 is a bit dodgy in my experience. I'm hoping that the new HP SPI integrated into NNMi is much less problematic. No IPv6 just yet, but perhaps in the near future.

We'll be polling as well as dealing with traps. I know some swear by traps and minimal polling, but I'm concerned that traps may not always make it to the receiver in certain conditions. The estate isn't huge, under 2000 devices, although there are quite a lot of big switches with tonnes of interfaces in that lot.

Do you have anything else you need to integrate with (ciscoworks, troubleticketing or manager of mangers solution etc.)
Do you have any other HP OpenView/BTO software deployed that you will integrate with?

The only other products worth considering in this space are Specktrum (but its CA - gah!), or Nagios (but like all ITSM tools you have to put a lot in to get anything good out!)


EIDT: Sorry for the q's, but its hard to comment on it generally other than saying it that it works for most use cases.

We don't really do much integration right now, although it's something I'll probably have to look at.

Agreed btw - there doesn't seem to be any network management "Holy Grail".

I've been reading the deployment documentation (you can tell HP have a fortune to spend on getting documentation done properly) and must say the ongoing discovery makes so much more sense. A bit of me died inside when I first found that 7.5x relied on a scheduled ET discovery, it's a killer when things change quite often in your network.

Do you find that management overhead of NNMi is significantly lower compared to previous versions? Also, did you do any Syslog integration previously with ovsyslogcfg, and how are you dealing with the loss of that functionality now?

Cheers!

 

[Narj]

Thread ressurection but I thought I'd mention that I ended up rolling out NNMi9 (with patch 1) and am really happy with it. MUCH less admin overhead and it seems solid enough. A few minor bugs but HP are pretty quick with their deveoplment on this product and fixes seem to come out quick.

Only problem is it caused a load of work in the first week because it picked up quite a few issues that we never knew we had. Not to mention a load of kit I didn't even know was there.

The clustering works a treat as well, and the discovery is pretty quick. We could see what a local site were doing with re-jigging of cables and cards within minutes of them doing it. excellent solution.