Socket read write on Solaris 8 too slow.

Hi
We have an application which consists of several server instances.
The front-end is web-based , using JSP/Servlets. We are using Tomcat 4.0.
The servlet makes several connections to the underlying servers. But the read/write operation is too slow. Same setup runs much quicker on Windows.
We are running jdk1.4.1_02 (stable ?). Would much appreciate any help.
cheers
Projyal

tomcat version --3.2.3
j2sdk version----1.4
plarform---------solaris
solaris version--8
using the aboue configuration i need to deploy JSP codes on the solaris platfrom..can any technical expert guide me on how to do this..would be grateful if anyone could suggest a good site where i can download free tuorials of "JSP" for development in HTML files format or HELP format]
thank you
regards
brijesh

Similar Messages

Hard drive read write speeds are too slow

I was having an issue with 17inch macbook pro. was getting 4.0mb/s read write speed tops. and smart status failure.
So i replaced the drive. The new drive is a 7200 RPM 500GB WD drive. I'm getting 12.0mb/s on top speed. Computer feels a little quicker but still slow overall.
I ran this test on a fresh install of snow leopard.
Is it just me or do those read/write speeds on the new drive still seem slow?
2007 17inch macbook pro with 2gigs of ram.

That does seem slow - on the other hand, what are you using to measure it?
My 1Tb 2.5" SATA drive performance shows as follows: http://grab.by/7zau
Xbench Stats: http://grab.by/7zau
Anything in System Profiler? http://grab.by/7zaR
Link Speed at 3 Gb?

Slow nfs read/write from solaris 8 to solaris 10

there is a bug:4852269 which sums up my problem well.
I have a solaris 10 sparc with a big nfs shared raid5, and most of the other suns I have run solaris 8. the access time from sol8 to sol10 is awfull.
Is there some config/patch I can apply to the solaris 10 box?
I have put on the latest 10_Recommended.zip

it's a ufs filesystem
If it helps, here is my /etc/default/nfs
# ident "@(#)nfs 1.10 04/09/01 SMI"
# Copyright 2004 Sun Microsystems, Inc. All rights reserved.
# Use is subject to license terms.
# Sets the maximum number of concurrent connection oriented connections.
# Default is unlimited and is obtained by not setting NFSD_MAX_CONNECTIONS.
# Equivalent to -c.
#NFSD_MAX_CONNECTIONS=
# Set connection queue length for the NFS over a connection-oriented
# transport. The default value is 32 entries.
# Equivalent to -l.
NFSD_LISTEN_BACKLOG=32
# Start NFS daemon over the specified protocol only.
# Equivalent to -p, ALL is equivalent to -a on the nfsd command line.
# Mutually exclusive with NFSD_DEVICE.
NFSD_PROTOCOL=ALL
# Start NFS daemon for the transport specified by the given device only.
# Equivalent to -t.
# Mutually exclusive with setting NFSD_PROTOCOL.
#NFSD_DEVICE=
# Maximum number of concurrent NFS requests.
# Equivalent to last numeric argument on nfsd command line.
NFSD_SERVERS=16
# Set connection queue length for lockd over a connection-oriented transport.
# Default and minimum value is 32.
LOCKD_LISTEN_BACKLOG=32
# Maximum number of concurrent lockd requests.
# Default is 20.
LOCKD_SERVERS=20
# Retransmit Timeout before lockd tries again.
# Default is 5.
LOCKD_RETRANSMIT_TIMEOUT=5
# Grace period in seconds that all clients (both NLM & NFSv4) have to
# reclaim locks after a server reboot. Also controls the NFSv4 lease
# interval.
# Overrides the deprecated setting LOCKD_GRACE_PERIOD.
# Default is 90 seconds.
GRACE_PERIOD=90
# Deprecated.
# As for GRACE_PERIOD, above.
# Default is 90 seconds.
#LOCKD_GRACE_PERIOD=90
# Sets the minimum version of the NFS protocol that will be registered
# and offered by the server. The default is 2.
#NFS_SERVER_VERSMIN=2
# Sets the maximum version of the NFS protocol that will be registered
# and offered by the server. The default is 4.
NFS_SERVER_VERSMAX=2
# Sets the minimum version of the NFS protocol that will be used by
# the NFS client. Can be overridden by the "vers=" NFS mount option.
# The default is 2.
#NFS_CLIENT_VERSMIN=2
# Sets the maximum version of the NFS protocol that will be used by
# the NFS client. Can be overridden by the "vers=" NFS mount option.
# If "vers=" is not specified for an NFS mount, this is the version
# that will be attempted first. The default is 4.
#NFS_CLIENT_VERSMAX=4
# Determines if the NFS version 4 delegation feature will be enabled
# for the server. If it is enabled, the server will attempt to
# provide delegations to the NFS version 4 client. The default is on.
#NFS_SERVER_DELEGATION=on
# Specifies to nfsmapid daemon that it is to override its default
# behavior of using the DNS domain, and that it is to use 'domain' as
# the domain to append to outbound attribute strings, and that it is to
# use 'domain' to compare against inbound attribute strings.
#NFSMAPID_DOMAIN=domain
#

Newbie socket read write question

Hello all,
I need to develop a simple client application in java using socket
programming. The client will connect to a server implemented in VC++.
The server part has been implemented so I need to develop only the
client part.
Could anybody help me with a simple program that sends data to the
server and then prints out whatever the server sends back for the
given request.
The client and the server communicate using a specified format of bytes as shown below. I do not know how to read those bytes using the sizes as the token.
I know how to connect to the server but I do not know the proper way of sending and receiving data.
connection = new Socket(destination, port);
outStream = new DataOutputStream connection.getOutputStream());
//send the ID 1 to the server
outStream.writeByte(1);
//send the total packet size
//send the data
Thanks in advance
Pertheli
The client and server communicates with the following packet format
shown below. Each packet will have header, size and data as shown
PACKET
offset(byte) contents
0 byte PACKET ID
1 byte PACKET length (motolora format)
2 byte
3 byte DATA * n (motolora format)
4 byte
5 byte ~ DATA
n byte
Now for a For say PACKET ID = 1
we send the data to the server like
0 byte (motolora format)
3 byte
Then from the VC server we recieve the data in the format such as
0 byte Name
43 byte
44 byte Status

Hi!
Try this...
Socket connection = new Socket(server,port);
// use a Buffered*Stream as often as possible, because
// of the IP Packet Lengths...It Increases the
// network performance (but you DON'T need it)
// you'll send bytes or messages to the server in
// the OutputStream and receive answers from the
// server in the InputStream
int packetOutputBufferSize = connection.getSendBufferSize();
int packetInputBufferSize = connection.getReceiveBufferSize();
BufferedOutputStream bos = new BufferedOutputStream(connection.getOutputStream(),packetOutputBufferSize);
BufferedInputStream bis = new BufferedInputStream(connection.getInputStream(),packetInputBufferSize);
// sending data through the socket
// create your packet in a byte array
byte[] output = new byte[<packet-lenght>];
output[0]=<packet ID>
output[1]=....
bos.write(output,0,output.length);
// receiving data from the socket
byte[] input = new byte[<max-packet-length>];
int read=0;
int length=0;
while ((read=bis.read(input,length,(input.length-length)) > 0) {
length+=read;
if (length>=input.length) break;
// now your server output is in the bytearray 'input'
If it isn't possible to set a max packet size, use a bytearray as buffer and write it content into a ByteArrayOutputStream, thats something like a resizable write-only byte array. (you'll get the full content of it by calling the method ByteArrayOutputStream.toByteArray())
Maybe you want to check the packet content during receiving the bytes from the socket: Then you should use some read() method from the InputStream, but thats not very powerful.

The write operation is too slow, why?

Hi,all
I have two DS, the version is 5.2_patch_6, and I have enabled the MMR. I find one of the two DS is response slow when I do write operations, as follows:
[05/Mar/2009:11:38:57 +0800] conn=38445 op=24167 msgId=707927 - MOD dn="uid=shzadmin,ou=People,o=xxxxx" authzid="dn:uid=amAdm
in,ou=People,o=xxxxx"
[05/Mar/2009:11:39:35 +0800] conn=38445 op=24167 msgId=707927 - RESULT err=0 tag=103 nentries=0 etime=38 csn=49af4f38000000010000
[05/Mar/2009:14:01:59 +0800] conn=38445 op=24393 msgId=710798 - ADD dn="uid=lhttestadmin,ou=People,o=xxxxx" authzid="dn:uid=amAdmin,ou=People,o=xxxxx"
[05/Mar/2009:14:02:37 +0800] conn=38445 op=24393 msgId=710798 - RESULT err=0 tag=105 nentries=0 etime=38 csn=49af70be000000010000
[05/Mar/2009:14:09:24 +0800] conn=38445 op=24471 msgId=711018 - DEL dn="uid=lhttestadmin,ou=People,o=xxxxx" authzid="dn:uid=amAdmin,ou=People,o=xxxxx"
[05/Mar/2009:14:09:36 +0800] conn=38445 op=24471 msgId=711018 - RESULT err=0 tag=107 nentries=0 etime=12 csn=49af727b000000010000
I don't know why. And more, I find the slow DS's error log have warning log as follows:
[26/Feb/2009:18:08:35 +0800] - WARNING<10307> - Replication - conn=-1 op=-1 msgId=-1 - Configuration warning Tombstone purging int
ernal search is unindexed in replica o=xxxxx
Does the warning result the slow? Anybody can help?
Thanks in advance!
Shen
Edited by: shenhongzhou on Mar 5, 2009 12:09 AM

[26/Feb/2009:18:08:35 +0800] - WARNING<10307> - Replication - conn=-1 op=-1 msgId=-1 - Configuration warning Tombstone purging internal search is unindexed in replica o=xxxxx
This is a known problem (bug 6175472) and the fix is already available in 5.2p6. Look for the script called <serverroot>/bin/slapd/admin/bin/ActivateFix6175472. Run it as:
# <server root>/bin/slapd/admin/bin/ActivateFix6175472 <server-root>/slapd-whatever
This will shutdown the instance, modify dse.ldif, define and create a vlv index (for the tombstone entries) and start the instance back up. Beware that if any of your entries have the objectclass definition in mixed case (objectClass etc), the script might not work as expected.
Easiest way to check is:
# cd <server-root>/slapd-whatever/config
# grep -i objectclass dse.ldif | grep -v objectclass
If nothing is returned, then you're all set to run the script. If not, you'll have to contact Sun Support (quoting bug 6692944) for a fix to the script that handles entries in a case insensitive manner.

Read Encoder Position is too slow, why?

I am controlling a telescope using close-loop servo mode of 7342. What I want is to use get current position of my motor using encoder and since the position value is critical to my telescope motion system I need a fast read back of that value. However, the action is very slow, about 20ms a cycle. Is there any specification on the speed of this function? Is there any solution to speed it up? Are there any new products which is faster than this card?

The NI-735x family of motion controllers allows you to read the position much faster. The position is read from a shared register that the motion board updates. Since it is a register read, the host takes very little time to get the position information.
The NI-733x/4x families of motion controllers do send a command to the motion board that the motion board has to process and then return the data. Typically this should only take a few milliseconds or less for a read. But it is dependent on what else the motion board is doing. Things that can effect the time it takes to respond to a host request are number of axes used in arc moves, and short bursts of multiple commands sent to the board. The board has a FIFO that is 16 commands deep. If you send down multiple commands followed by a read position, each command in front of the read command has to be processed first. Which could explain why you are seeing a 20 ms read.
In NI-Motion 6.1.x we have changed the driver software to help load balance the host request with the rest of the processes running on the board. In most cases this helps performance.
What version of our driver software are you using? How fast do you really need to get the position information?
Thanks,
Rodger S.

Formatted on XP - EXTREMELY SLOW read/write on the Mac !!!!

Man this iPod is wack!! Im using the iPod mini in Disk Mode to transfer bits (updaters since my XP machine is NEVER connected to the net) back and forth between my PowerBook and my XP Intel based machine. Obviously i set up and formatted the iPod Mini on the XP machine.
Observation, file read/write times are unacceptably SLOW on Mac and the PC. Did Apple deliberately make this thing unusable in this situation? This thing has a transfer speed as if it were connected to a USB 1.1 port but it is NOT, on any of my machines. I format my USB Flash Drive (SanDisk Cruzer Titanium) on the same XP machine, use the same ports and it's read/write times are at least 5 times faster than the Mini !!!
When i format and setup the iPod Mini on the Mac then install MacDrive on the PC astonishingly the read/write times significantly increase on BOTH Mac and the PC. As i suggested i think there is something drastically wrong with Apple's PC iPod setup, there is NO way my other devices formatted on the PC have such abysmal performance . Man i click on a folder in the Finder of the PC formatted iPod and it take about 3 seconds until the files inside are displayed...this is also the case on the PC.
Anyone else experienced this?
I was going to buy another iPod specifically for transferring data between my Mac's and my colleagues PC's, but there is no way i will with such severely degraded read/write performance.
Best

Apparently this is not a single isolated case.
Scott Aronian Posts :
"On Mac FAT32 Format Slower Than HFS "
My 60GB iPod came in Windows FAT32 format which I liked since I could use it in disk mode on both Mac & Windows. However it seemed slow to start playing video and when switching between video files.
I did some timing tests and on the Mac FAT32 was much slower than HFS.
FAT32
Start playing movie = 12s
Stop movie and restart at same location = between 12-30s
HFS
Start playing movie = 5s
Stop movie and restart at same location = between 6s
If you have a Mac and your new 5G iPod is formatted for Windows, you may want to use the iPod Updated to restore and make the drive HFS.
There is NO such performance problems with USB Flash drives when formatted to Fat32 on PC's i have tested many and they perform similar on BOTH platforms.
Something fishy goin on here

Oracle 10g direct path write too slow

Hi All,
We have Oracle 10g on a Solaris virtual server, VMWare ESXi being the host. Data files are on RAID1, internal storage on a HP DL585 with VMFS partition at ESXi level. Problem is that DB writes for a CREATE TABLE as SELECT... statement is way too slow. To create a table which is 0.5 GB, DB takes 9 minutes which amounts to 1 MB/s. When we check for FTP or file copy at Solaris level with same size file (0.5 GB), it flies through in less than a minute. This is Oracle 10.2.0.4, 8K data block, 2 vCPU assigned to the Solaris VM. Have checked with VMWare support for any known issues and also have SR open with Oracle for any param changes that can help speed up things. Any clues or pointers from you all will be of great help.
Thanks,
Nikhil

Here's the output from tkprof for waits
Elapsed times include waiting on following events:
Event waited on Times Max. Wait Total Waited
---------------------------------------- Waited ---------- ------------
single-task message 1 0.17 0.17
SQL*Net message to dblink 150 0.00 0.00
SQL*Net message from dblink 150 0.04 0.32
SQL*Net message to client 1 0.00 0.00
direct path write temp 4003 1.16 804.93
direct path read temp 2563 0.14 35.86
SQL*Net more data from dblink 126967 0.17 11.81
SQL*Net message from client 1 17.73 17.73
Direct Path write temp has total waits of 804.93. Also, I am NOT looking to tune a particular SQL. Database is overall slow on VMware and I am looking for any gotchas for running Oracle 10g within a Solaris VM.
Thanks,
Nikhil

Alert: Logical disk transfer (reads and writes) latency is too high Resolution state

Hi
We are getting following errors for my 2 virtual servers. We are getting this alert continuously. My setup Windows 2008 R2 SP1 2 node Hyper V cluster. Which is hosted 7 guest OS out of am facing this problem with to guest os. Once this alert started
my backup running slow.
Alert: Logical disk transfer (reads and writes) latency is too high
Source: E:
Path: Servername.domain.com
Last modified by: System
Last modified time: 4/23/2013 4:15:47 PM Alert description: The threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded.
Alert view link: "http://server/OperationsManager?DisplayMode=Pivot&AlertID=%7bca891ba3-e9f2-421f-9994-7b4d6e867b33%7d"
Notification subscription ID generating this message: {F71E01AF-0BE6-8377-7BE5-5CB6F5C037A1}
Reagrds
Mahesh

Hi,
Please see if following helps
Disk transfer (reads and writes) latency is too high
The
threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded
If they are of no help, try asking this question in Operations Manager - General forum since alerts are generated by SCOM.
Regards, Santosh
I do not represent the organisation I work for, all the opinions expressed here are my own.
This posting is provided AS IS with no warranties or guarantees and confers no rights.

Muti-channel read/write through FP data socket posible?

Hello,
I'm trying to talk to cFP-1808 module from desktop PC real-time host over Ethernet. I did not find any Fieldpoint driver for desktop RT so I'm forced to use data sockets to communicate with cFP-1808. It seems that one can only read or write to one FP data channel per socket. With typical latency of tens of milliseconds per such read/write and couple of hundreds channels present, I'm looking at SECONDS per cycle time, which is really slow.
Is there any workaround for this problem?

You can also read and write to channels in LabVIEW. First you have to configure your 1808 in MAX and save the IAK file. You then load the IAK file in LabVIEW by selecting browse in the FieldPoint IO Point In (constant or control).
In the FieldPoint Device Selection Dialog Windows select the View Configurations Tab.
In this tab navigate to your IAK file and name it in the text box next to where the full file path is displayed.
In the Browse FieldPoint Tab of the same window you should now be able to see all of your FieldPoint modules and channels.
Message Edited by BLAQmx on 03-06-2008 01:12 PM
Mark
LabVIEW R&D

"Full Screen Mode" in Adobe Reader XI for Windows 8 loads way too slow!

I recently got a new laptop with Windows 8, after having used Windows Vista for years. I installed Adobe Reader XI for my new computer. Before, I used Adobe Reader X on Vista. I used Adobe Reader "Full Screen Mode" to read the nice-looking Bible chapters I created with internet audio from Biblegateway.com. In Vista, the "Full Screen Mode" appeared quickly, before the audio fully started, so I could see the text in time for the reading.
However, with Windows 8, "Full Screen Mode" takes around 5-10 seconds to appear at all, and a black screen is present throughout the wait. It does the same thing when I hit "escape" to exit "Full Screen Mode" as well. I even uninstalled Adobe Reader XI from my new laptop, and tried using Adobe Reader X, even though it is not recommened by Adobe to do that on Windows 8, in order to test the speed of "Full Screen Mode" there. Adobe Reader X had the same problem! I do not know if this can be fixed, but I am pretty unhappy about it! I enjoyed the "Full Screen Mode" with my PDF chapters and the online bible audio! Finding that function for the first time with Adobe Reader X was a delight for me! It gave the reading a theater feel!
Has anyone else here had a similiar problem with "Full Screen Mode" loading way too slow on Windows 8? I am very disappointed! If this cannot be fixed by a user, is there a place to complain to Adobe and make a recommendation for software improvement? Thank you!

From: Test Screen Name [email protected]
Sent: Sunday, August 04, 2013 11:32 AM
To: gunner0490
Subject: I'm not able to open PDF files in Adobe Reader XI.
Re: I'm not able to open PDF files in Adobe Reader XI.
created by Test Screen Name <http://forums.adobe.com/people/TestScreenName> in Adobe Reader - View the full discussion <http://forums.adobe.com/message/5565198#5565198

Adobe reader too slow and "not responding," why?

I downloaded adobe reader XI (11.0.10) for my touchscreen laptop with Windows 8.1 but when reading PDFs it's scrolls too slow, gets stuck when highlighting or performing other commands, and then I normally get "not responding."
It's really slowing down my productivity!
Would appreciate your help please. Thanks

Try disabling Protected Mode [Edit | Preferences | Security (Enhanced)].
If that does not resolve the problem, please tell us your operating system.

MSMQ read/write is slow

Sir,
I have installed MSMQ in my Windows server 2008 R2 operation system but the queue reading and writing is very slow .It gives me 150 to 160 messages per second for write or read and the speed decreases if the read and write operation is performed at
a time .
Earlier I have MSMQ on Windows server 2003 .Read/write operations is fine and performance is good.It gives me 1000 to 2000 messages per second .
Regards,
Sandeep

Hi,
Thanks for posting here.
Regarding the current issue, please try to refer to the following article to see if it could improve the performance.
How to improve MSMQ disk performance
http://blogs.msdn.com/b/johnbreakwell/archive/2008/02/13/msmq-disk-performance.aspx
Hope this helps.
Best Regards,
Andy Qi
Andy Qi
TechNet Community Support

Vdbench 5.04 - Solaris - read/write profile problem

Hi, I'm trying to simulate a read/write IO profile to a raw disk. Have a Solaris 10 x86 server connected to a VNX over Fibre Channel.
The profiles are spread over IO sizes ranging from 512b to 128k but I've tried getting this simplified version to work without success.
I am using Vdbench 5.04. This is my parameter file:
sd=sd1,lun=/dev/dsk/c3t60060160C43034009EF8FCAE1522E311d0s2
wd=write,sd=sd1,readpct=0,xfersize=(4k,50,8k,50)
wd=read,sd=sd1,readpct=100,xfersize=(4k,50,8k,50)
rd=test,wd=(write,read),maxdata=4.5g,elapsed=3600,threads=50,iorate=500
I have a dtrace script (iosnoop from the DTraceToolkit) running monitoring the I/O.
From the output of the dtrace script I am seeing no 8k reads or 4k writes.
The statistics from the VNX show the same results as the dtrace output so what is being written to the LUN is correct.
Is xfersize the correct way to simulate a range of I/O sizes or is their a mistake in the parameter file?
Any help appreciated :-)

You are caching your i/o. Please specify /dev/rsdk/

Slow performance to read/write shared variables programatically

We are using datasocket read and write functions to read and write shared variables programatically (in the same machine) but we only achieve a performance of aprox. 200 reads/writes per second. We are using Labview 8.6 with DSC.
Is possible to get better results? That performance is normal?
Any help would be appreciated. Thank you in advance.

Hi MMCDAT,
I think this value can
be normal as you can see in this link:
http://zone.ni.com/devzone/cda/tut/p/id/5037
As you can see, the
limit for datasocket depends on your Ethernet limitations, even if you as using
it just in one PC:
http://digital.ni.com/public.nsf/websearch/6AC9E65734E53F9A8625672400637ECC?OpenDocument
You can improve the
performance changing the update mode or Vis configurations:
http://digital.ni.com/public.nsf/allkb/F8F7DE98856B50588625672400648045?OpenDocument
http://digital.ni.com/public.nsf/allkb/2D9C6D73A160537986256B290076456E?OpenDocument

Socket read write on Solaris 8 too slow.

Similar Messages

Maybe you are looking for