Cisco IPS Manager Express performace tuning

A few weeks ago I did a clean install of IME on a virtual XP with 2GB of RAM. The user experience was okay, after it loaded the gadgets, but I just had to check out the system resources graph.
The java process used around a 100MB of RAM, and the MySQL used under 15MB. While I had a lot of free RAM as buffer, from my experiences it never hurts to give MySQL a lot of RAM.
The config file is "my.ini" under the base installation folder, and this is how it looks like after some tweaks:


[mysqld]
basedir=C:\Program Files\Cisco Systems\Cisco IPS Manager Express/MYSQL
datadir=C:\Program Files\Cisco Systems\Cisco IPS Manager Express/MYSQL/data
port=47007
myisam-recover=DEFAULT,FORCE,QUICK
bind-address=127.0.0.1


# performance options
key_buffer_size=128M
table_cache=256
sort_buffer_size=32M
read_buffer_size=8M

Now it uses around 120MB, and once started, it loads the gadgets noticeably faster.
I tried to add some InnoDB options too, but the options are not recognized. I guess that InnoDB support is not compiled in, or something.

Cisco MARS 20 upgrade

I'm not sure where can you find performance evaluations for MARS models, but our MARS 20 definitely had issues with managing 100+ switches and our main ASA cluster and FWSM logs. Since it's EOL soon, we don't want to buy a bigger one. But it's just a PC, so we looked into upgrade possibilities.

The MARS 20 has a SuperMicro P4SCT mainboard and 1GB of RAM, so our options are:

• A new CPU, possibly with Hyperthreading. Problem is, the kernel is not SMP compatible, and I don't really want to compile a new one.
• More RAM. That's always good, the maximum supported is 4GB DDR-400 in 4 sticks of 1GB.
• A new HDD. Our old one is, well, old (40k+ hours), so I decided to replace it with a 160GB standard IDE Western Digital drive.

In the end, we opted to upgrade the RAM and replace the HDD.

The RAM upgrade was no problem at all, and I used a RescueCD USB stick to 'dd' the MARS OS between drives. The new drive is a good 2 times faster, and the new memory usage stats are eye-candy:

[root@voyager bin]# free -m

             total       used       free     shared    buffers     cached

Mem:          3772       3764          7        210         98       2141

-/+ buffers/cache:       1523       2248

Swap:         1027        851        175

Overall, the responsiveness of the system is way better in the web interface, and cron tasks are much faster.

Kaniak interview

In case you guys missed it:
http://www.galomagazine.com/artculture/accent-pieces-aleksandra-kaniak-on-making-a-place-for-polish-in-hollywood/

Catalyst 3500XL layer 3 frozen

I'm not sure about the reason (most likely a broadcast storm), but one of our 3500XL switches stopped processing L3 traffic. Layer 2 switching was normal, and since it's our distribution switch (Cisco Modular Campus Design, here we come!), I didn't really want to reload it.

Consoling to the switch provided me with a responsive CLI, so no trouble there, either. CPU utilization was normal too.

After a few show commands, it turned out that it can't even resolve IPs with ARP.

I tried clear arp-cache, clear mac-address table, clear int vlan 10; no joy. I had to delete the entire vlan interface (no interface vlan 10), clear the caches, and re-create the vlan interface to make it work again. Of course it's running the latest IOS out there (12.0(5)WC17).

A new kind of sexy: Huawei Quidway S2309

During winter break I managed to acquire a new baby to my collection:

a Huawei Quidway S2309 switch.

It's a layer2, carrier CPE grade box, with a lot of features:

http://www.huawei.com/ucmf/groups/public/documents/attachments/hw_093972.pdf

It runs the Huawei VRP software, which is almost like the 3Com Comware, so I can use this baby to practice for my HP ASE exams.

Stay tuned for updates on this baby...