2.5 slower with caching?

2.5 slower with caching?

Post by Jeff Bac » Thu, 14 Mar 1996 04:00:00




: I've had a few users who are running/testing Solaris 2.5 with caching.
: They seem to feel that it runs "slow". Not very precise in what is slow,
: but when I turned off caching they reported that it was "faster" or at
: least more consistent.  Has anyone else heard these types of complaints?
: I am caching only application directories (framemaker, etc.) and they
: are running on a Sparc 5/75 and IPXes.  Surely accessing from local disk
: should be faster than going over the network to our server even if I
: have it consistency checking every 30 seconds?

not in every case.

cacheos, in my experience, isn't something that's always useful,
or to be blindly used. it's just like any other tool or component
of an overall system design.

for example, I had several machines here set up with a cache for
their software directories. everyone noticed that those machines
were significantly slower than all the others. didn't make anyone
happy, that. why? because the machine had only one disk. as a result,
when it started paging, it'd flush out both anon memory and software
text... which it'd then have to reload - from the cache.. and in this
case, all the extra overhead of moving the disk heads back and forth
from swap partition to cache partition and back again and again was
overall slower than reloading the software from the relatively faster
server and just swapping to the local disk.

: One other aspect is that with caching on (30 second timeout to test it),
: when they type a command that doesn't exist in their PATH, it takes 10
: or 15 seconds to return for the first mistyped command.  Then its
: instantaneous for a few hours, then it takes 10 or 15 seconds, etc.

turn off consistency checking. :)

-bacon
--
= Jeff Bacon         General Systems Hack, Michigan Technological University =

 
 
 

2.5 slower with caching?

Post by Tom Morni » Thu, 14 Mar 1996 04:00:00



: Surely accessing from local disk
: should be faster than going over the network to our server even if I
: have it consistency checking every 30 seconds?

As i understand it, cachefs is only useful when you have a nearly static
filesystem. Something that needs to be checked every 30 seconds probably
isn't a good candidate.

You might be flooding your network with consistency checking!

This is just my understanding...

-- Tom Mornini
-- http//www.mornini.com

 
 
 

2.5 slower with caching?

Post by David J. Heilan » Thu, 14 Mar 1996 04:00:00


I've had a few users who are running/testing Solaris 2.5 with caching.
They seem to feel that it runs "slow". Not very precise in what is slow,
but when I turned off caching they reported that it was "faster" or at
least more consistent.  Has anyone else heard these types of complaints?
I am caching only application directories (framemaker, etc.) and they
are running on a Sparc 5/75 and IPXes.  Surely accessing from local disk
should be faster than going over the network to our server even if I
have it consistency checking every 30 seconds?

One other aspect is that with caching on (30 second timeout to test it),
when they type a command that doesn't exist in their PATH, it takes 10
or 15 seconds to return for the first mistyped command.  Then its
instantaneous for a few hours, then it takes 10 or 15 seconds, etc.

Any pointers?

-- Dave Heiland --

Lucent Technologies
Malmesbury, England

 
 
 

1. Linux 2.4/2.5 SCSI considerably slower than FreeBSD

Hello,

Marco Flohrer has posted an inquiry to de.comp.os.unix.linux.hardware

Seagate 36ES2 was slow with a DawiControl 2976UW (SYM53C875), only
around 25 MB/s. I have the same observation with a Fujitsu MAH3182MP
with an Adaptec 2940UW Pro which is not much faster. Either bus has an
active LVD/SE terminator.

Single-user mode,
time dd if=/dev/XXX of=/dev/null bs=65536 count=10240
(671,1 MB) linear read.

Table shows throughput in decimal MB/s (M = 1,000,000)

                               2.5  2.4  FBSD        max.
UWSCSI Fuj MAH3182MP  7200/min 32,1 29,4 35,1 TQ     40
UDMA66 Max 4W060H4    5400/min 27,1 26,7 25,7        66
UDMA66 IBM DTLA307045 7200/min 37,2 37,5 37,2 TQ 2.5 66
UDMA66 WDC AC420400D  5400/min 15,5 15,5 15,5 TQ 2.5 66
                               --------------
table is in decimal MB/s.

2.4:  Linux 2.4.19-pre2-ac3
2.5:  Linux 2.5.15
FBSD: FreeBSD 4.6-RC (Tagged Queueing Broken)

The IDE drives are attached to a VIA 82C686 (KT133), the Fujitsu
(actually an U-160 drive) to the mentioned Adaptec.

FBSD gets about 20% better throughput. It's far from perfect, but 90% of
the maximum is probably almost as good as we can get.

Why is Linux SCSI so slow?

--
Matthias Andree
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Curious newbie: How does routing work?

3. Solaris 2.5: IPC gets faster, Sparc 20 gets slower?

4. Palm sync issues with Evolution/Gnome?

5. [2.5] reproducible smp_call_function/_interrupt cache update race.

6. Sox can't compile

7. PC slower with cache !!

8. Apache Problems on FreeBSD

9. Arp cache logic different on 2.5?

10. Q: System w/ limited 256 kB cache, 16MB->32MB = slower linux.

11. squid web cache: the larger the slower

12. PCI cache line messages 2.4/2.5