LVM questions regarding performance

Question

Over the holidays, I'll be redoing my server (running Arch) using LVM.  The situation is this: I have two internal HDDs (80GB apiece) on the IDE bus, and an external USB2.0 drive sitting around not being used. The current install is also partitioned in such a way that there are some partitions that are nearly empty. So naturally I want to put things into a big volume group so that I'm not wasting space, full well knowing that there may be a slight performance hit (that I can justifiy).
I'll probably do it this way, using /boot because of the legacy GRUB limitation, with swaps outside of the LVM to make sure they're contiguous:
/dev/sda1     /boot
/dev/sda2     swap 1
/dev/sda3     LVM partition #1
/dev/sdb1     swap 2
/dev/sdb2     LVM partition #2
/dev/sdc1     LVM partition #3
Anyway, my questions are these:
1) Do certain filesystems perform better when housed on volume groups?
2) As the two internal IDE drives are faster than the external, I would like to "prioritize" those two logical volume partitions. Is it possible to do this?
I know they seem like simple questions. I've searched, but can't come up with anything searching the obvious search strings. So sorry if I'm overlooking something.

Answer 1

thanks for your fast reply!
i moved from a replicated to a distributed cache:
<?xml version="1.0"?>
<!DOCTYPE cache-config SYSTEM "cache-config.dtd">
<cache-config>
<caching-scheme-mapping>
<cache-mapping>
<cache-name>*</cache-name>
<scheme-name>example-distributed</scheme-name>
</cache-mapping>
</caching-scheme-mapping>
<caching-schemes>
<!--
Distributed caching scheme.
-->
<distributed-scheme>
<scheme-name>example-distributed</scheme-name>
<service-name>DistributedCache</service-name>
<backing-map-scheme>
<class-scheme>
<scheme-ref>unlimited-backing-map</scheme-ref>
</class-scheme>
</backing-map-scheme>
<autostart>true</autostart>
</distributed-scheme>
<!--
Backing map scheme definition used by all the caches that do
not require any eviction policies
-->
<class-scheme>
<scheme-name>unlimited-backing-map</scheme-name>
<class-name>com.tangosol.util.SafeHashMap</class-name>
<init-params></init-params>
</class-scheme>
</caching-schemes>
</cache-config>
everything is now much faster, which is fine.
however, if i try to put 200k instead of 100k objects in the cache (i do this in putAll batches of 10k each) i get a java.lang.OutOfMemoryError: Java heap space.
as you can see the object has 2 ints only ... so the raw data of 200k objects is 1,6mb only.
any suggestions?
update:
i increased the java vm heap memory to 1gb, now it works.
however, i notice that for each 100k objects (2 integers each) jave consumes ~80mb heap memory. at about 700mb heap memory the gc kicks in.
how can this be opimized? 8 bytes of payload is not that much, i doubt that i'm able to put REAL data records into the cache this way.
thanks!
Message was edited by:
druid99