You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by David kerber <dc...@verizon.net> on 2011/03/11 21:02:12 UTC
Performance 5.5 vs 6 vs 7
How might I expect the performance of 7 to compare to a late rev of 5.5
or 6.0.x, running under JRE 6? Specifically, my application has lots of
small, simple requests, and the app does nothing but de-obfuscate (it's
not really sophisticated enough to call it encryption) the data and save
it to disk. But it does a LOT of them, 60 to 100 per second during
normal operation, more during very busy times. The overall average when
I checked last week was about 60 requests per second for the previous 10
days, around the clock, >4.5M per day.
I spent quite a bit of effort in optimizing my app last year, and got
some big gains in it, but now I'm again starting to not be able to keep
up with the increased flow as our customer has doubled their number of
installations, so I'm looking elsewhere for potential improvements. I've
already checked my connection bandwidth, and that still has some
headroom, though not a lot. Can I expect TC 6 or 7 to handle larger
numbers of very simple requests than 5.5 can?
Dave
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by David kerber <dc...@verizon.net>.
On 3/14/2011 8:21 AM, Peter Crowther wrote:
> On 14 March 2011 12:08, David kerber<dc...@verizon.net> wrote:
>
>> Dave, could you give us any more information about your network? What is
>> the piece that's at 80% utilisation when you see the trouble? Is it a
>> point-to-point connection, or an Ethernet LAN, or what? If it's Ethernet,
>> what hardware are you using for connection?
>>
>> It's our internet connection to the outside world, which is a T-1, with a
>> Cisco ASA for the firewall. The connections we're processing are from a
>> bunch of separate customer sites, but which all connect to us through the
>> customer's gateway.
>
>
> A T1 used to be 1.5 Mbit/s bidirectional, back when I did my basic network
> training (which, I admit, is a good few years ago). Is it still? If so,
Yes, 1.5Mbps bidirectional
> that's quite low. I have 34 Mbit/s down, 1.5 Mbit/s up to my house for the
> price of a decent restaurant meal for one per month. What might your
Me, too (mine is 10 down/2 up), but FIOS isn't available where we are,
or I'd go that route in about half a heartbeat. Cable is, and is
faster, and we actually use that for our backup, but the reliability
isn't good enough for it to be our primary connection (frequent short
outages).
> upgrade options be?
We're in the process of upping our bandwidth with our current ISP, but
the options ae rather limited without having to change our IP address,
and therefore all our customers' firewall settings, Lan-2-Lan vpns, etc.
So changing that, while doable, is a last resort.
> Also, see if the ASA can give you any usage statistics. There's a small
> chance that you're saturating your firewall if you have a very large number
> of short-lived connections. Cisco's figures don't always stack up here - in
Yeah, I've been looking at that as well, but as you say, the numbers
don't always stack up.
> the past, I've had to debug load problems with Cisco routers that were
> supposed to be able to handle two ISDN PRIs, and couldn't when there was
> heavy call setup and termination load. I accept this isn't quite a parallel
> case!
>
> Tomcat might still be the problem, but I'd certainly take a long hard look
I never really suspected tomcat as the cause after I did my app tuning
work a year ago or so, but since that's the easiest to test and
eliminate, I started there.
> at your network infrastructure. Does the server really have to be on your
> site, or can you have a server in a bunker somewhere that passes hourly /
> daily reports to your site via a heavily compressed file format? Our
> telemetry system does that, for example, just running in a VM (for
> reliability, believe it or not).
Hourly or daily wouldn't be enough, but every minute or every 5 minutes
might be a possibility, and one we hadn't thought of; thanks for the
suggestion. We're also looking at moving the entire process to a data
center hosting service as well.
D
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Peter Crowther <pe...@melandra.com>.
On 14 March 2011 12:08, David kerber <dc...@verizon.net> wrote:
> Dave, could you give us any more information about your network? What is
> the piece that's at 80% utilisation when you see the trouble? Is it a
> point-to-point connection, or an Ethernet LAN, or what? If it's Ethernet,
> what hardware are you using for connection?
>
> It's our internet connection to the outside world, which is a T-1, with a
> Cisco ASA for the firewall. The connections we're processing are from a
> bunch of separate customer sites, but which all connect to us through the
> customer's gateway.
A T1 used to be 1.5 Mbit/s bidirectional, back when I did my basic network
training (which, I admit, is a good few years ago). Is it still? If so,
that's quite low. I have 34 Mbit/s down, 1.5 Mbit/s up to my house for the
price of a decent restaurant meal for one per month. What might your
upgrade options be?
Also, see if the ASA can give you any usage statistics. There's a small
chance that you're saturating your firewall if you have a very large number
of short-lived connections. Cisco's figures don't always stack up here - in
the past, I've had to debug load problems with Cisco routers that were
supposed to be able to handle two ISDN PRIs, and couldn't when there was
heavy call setup and termination load. I accept this isn't quite a parallel
case!
Tomcat might still be the problem, but I'd certainly take a long hard look
at your network infrastructure. Does the server really have to be on your
site, or can you have a server in a bunker somewhere that passes hourly /
daily reports to your site via a heavily compressed file format? Our
telemetry system does that, for example, just running in a VM (for
reliability, believe it or not).
Cheers,
- Peter
Re: Performance 5.5 vs 6 vs 7
Posted by David kerber <dc...@verizon.net>.
On 3/14/2011 3:12 AM, Peter Crowther wrote:
> On 13 March 2011 21:01, Tony Anecito<ad...@yahoo.com> wrote:
>
>> As someone mentioned the network can imit you. If your bandwidth
>> utilization is
>> at 60% or over you are in trouble since collisions start to become a
>> serious
>> issue.
>>
>> Collisions may or may not be an issue, depending on the exact mode of
> operation of the congested network link.
>
> Collisions are a feature of the original Ethernet. They're still an issue
> on many Ethernets, but there are provisos even here. If your switch does
> store-and-forward, or you simply have a switched network and all the traffic
> is between two stations, then collisions are relatively unlikely and the
> Ethernet usage can go considerably higher than 60% before collisions and
> retries start to become significant. Sure, if you're on wireless or using
> an old 10baseT or Cheapernet hub, 60% is about the knee in the curve. But
> I'd be a little surprised if Dave were using such ancient technology for a
> critical piece of infrastructure such as this server :-).
We're not, though you would probably be surprised how long it lasted
before we modernized... ;-D
> Point-to-point links, such as most broadband connections, do not suffer from
> collisions as the routers at each end store and forward traffic across the
> link. That said, by the time you get to 80% usage then each packet is
> typically going to have to queue behind several others. Even if this only
> adds a few milliseconds of latency each time, it builds up when you multiply
> it by the TCP 3-way handshake, 4-way FIN/ACK sequence, plus any data you're
> transferring.
Makes sense to me.
> Dave, could you give us any more information about your network? What is
> the piece that's at 80% utilisation when you see the trouble? Is it a
> point-to-point connection, or an Ethernet LAN, or what? If it's Ethernet,
> what hardware are you using for connection?
It's our internet connection to the outside world, which is a T-1, with
a Cisco ASA for the firewall. The connections we're processing are from
a bunch of separate customer sites, but which all connect to us through
the customer's gateway.
D
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Tony Anecito <ad...@yahoo.com>.
Hi Peter,
Thanks for your comments. I am using 1Gb ethernet.
My loopback testing included the motherboard network circuitry, TCP/IP stack,
OS, CPU and client and Tomcat App code. The best I got for a single request
round trip that reused the http connection was 600 microseconds. As I added more
threads running on the client sending requests it got slower.
When not using loopback I got around 1.2msec so the extra components included my
1Gb switch and 1Gb router.
When I ran the client from my laptop on the network that had 1Gb ethernet
support I got over 2msec.
I googled on the internet and I was seeing 2msec as the fastest people were
saying for a network card even one with a built-in CPU.
I would love to eliminate the network portion from the path and go directly from
client to Tomcat to see what Tomcat truely can do.
Any recommendations are appreciated. I did move on to other projects since I
could not think how more timing info.
Regards,
-Tony
----- Original Message ----
From: Peter Crowther <pe...@melandra.com>
To: Tomcat Users List <us...@tomcat.apache.org>
Sent: Mon, March 14, 2011 1:12:12 AM
Subject: Re: Performance 5.5 vs 6 vs 7
On 13 March 2011 21:01, Tony Anecito <ad...@yahoo.com> wrote:
> As someone mentioned the network can imit you. If your bandwidth
> utilization is
> at 60% or over you are in trouble since collisions start to become a
> serious
> issue.
>
> Collisions may or may not be an issue, depending on the exact mode of
operation of the congested network link.
Collisions are a feature of the original Ethernet. They're still an issue
on many Ethernets, but there are provisos even here. If your switch does
store-and-forward, or you simply have a switched network and all the traffic
is between two stations, then collisions are relatively unlikely and the
Ethernet usage can go considerably higher than 60% before collisions and
retries start to become significant. Sure, if you're on wireless or using
an old 10baseT or Cheapernet hub, 60% is about the knee in the curve. But
I'd be a little surprised if Dave were using such ancient technology for a
critical piece of infrastructure such as this server :-).
Point-to-point links, such as most broadband connections, do not suffer from
collisions as the routers at each end store and forward traffic across the
link. That said, by the time you get to 80% usage then each packet is
typically going to have to queue behind several others. Even if this only
adds a few milliseconds of latency each time, it builds up when you multiply
it by the TCP 3-way handshake, 4-way FIN/ACK sequence, plus any data you're
transferring.
Dave, could you give us any more information about your network? What is
the piece that's at 80% utilisation when you see the trouble? Is it a
point-to-point connection, or an Ethernet LAN, or what? If it's Ethernet,
what hardware are you using for connection?
Tony - I also don't believe 2ms is the best you can get on Ethernet. If it
was, most big machine rooms would have silted up with network traffic long
ago ;-). It depends on what network technology you're using. For example,
gigabit Ethernet latencies are considerably lower than good ol' 10 meg.
What link speeds do you have to your router? 100 Mbit/s?
- Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Peter Crowther <pe...@melandra.com>.
On 13 March 2011 21:01, Tony Anecito <ad...@yahoo.com> wrote:
> As someone mentioned the network can imit you. If your bandwidth
> utilization is
> at 60% or over you are in trouble since collisions start to become a
> serious
> issue.
>
> Collisions may or may not be an issue, depending on the exact mode of
operation of the congested network link.
Collisions are a feature of the original Ethernet. They're still an issue
on many Ethernets, but there are provisos even here. If your switch does
store-and-forward, or you simply have a switched network and all the traffic
is between two stations, then collisions are relatively unlikely and the
Ethernet usage can go considerably higher than 60% before collisions and
retries start to become significant. Sure, if you're on wireless or using
an old 10baseT or Cheapernet hub, 60% is about the knee in the curve. But
I'd be a little surprised if Dave were using such ancient technology for a
critical piece of infrastructure such as this server :-).
Point-to-point links, such as most broadband connections, do not suffer from
collisions as the routers at each end store and forward traffic across the
link. That said, by the time you get to 80% usage then each packet is
typically going to have to queue behind several others. Even if this only
adds a few milliseconds of latency each time, it builds up when you multiply
it by the TCP 3-way handshake, 4-way FIN/ACK sequence, plus any data you're
transferring.
Dave, could you give us any more information about your network? What is
the piece that's at 80% utilisation when you see the trouble? Is it a
point-to-point connection, or an Ethernet LAN, or what? If it's Ethernet,
what hardware are you using for connection?
Tony - I also don't believe 2ms is the best you can get on Ethernet. If it
was, most big machine rooms would have silted up with network traffic long
ago ;-). It depends on what network technology you're using. For example,
gigabit Ethernet latencies are considerably lower than good ol' 10 meg.
What link speeds do you have to your router? 100 Mbit/s?
- Peter
Re: Performance 5.5 vs 6 vs 7
Posted by Tony Anecito <ad...@yahoo.com>.
I am not in the habit of defending Microsoft but the question is would you
rather have hard coded functions or ones you dynamically put in?
Just saying something without seeing the actual code of what they did it is hard
to say who as the most optimized code.
I will setup the test with the adaptor and see what the capacity of Tomcat 7 is
using the "Windows Loopback interface". I am hoping for better than the 600
microseconds I saw without it.
I think DOS will beat everything today because size does matter (I do not miss
the Osborne luggable). I feel the same way about 32-bit versus 64-bit java and
OS on the same hardware. I was around during the early NT days and I really did
not like it. But now big monolithic OS's are the trend. Writing code to fit on a
single 1.4MB floppy or fit in 64K of memory is a lost art.
Best Regards,
-Tony
----- Original Message ----
From: Christopher Schultz <ch...@christopherschultz.net>
To: Tomcat Users List <us...@tomcat.apache.org>
Sent: Mon, March 14, 2011 3:09:35 PM
Subject: Re: Performance 5.5 vs 6 vs 7
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Tony,
On 3/14/2011 3:08 PM, Tony Anecito wrote:
>>>From the definitions and discussions on google regarding loopback interface
> (localhost) it will go through the nic card for Windows unless you install a
> loopback adaptor. For Linux you do not need a nic card.
>
> I was using windows and thus it used my nic card.
Oh, I didn't realize that the Windows TCP/IP stack was that poorly
optimized.
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk1+hA8ACgkQ9CaO5/Lv0PD6rACdFyW/vYPubdDCYqU1VKqYCo85
mqkAoIRlwv/+UsBhUX/SxEGPwFejPdO0
=GxPR
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Tony,
On 3/14/2011 3:08 PM, Tony Anecito wrote:
>>>From the definitions and discussions on google regarding loopback interface
> (localhost) it will go through the nic card for Windows unless you install a
> loopback adaptor. For Linux you do not need a nic card.
>
> I was using windows and thus it used my nic card.
Oh, I didn't realize that the Windows TCP/IP stack was that poorly
optimized.
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk1+hA8ACgkQ9CaO5/Lv0PD6rACdFyW/vYPubdDCYqU1VKqYCo85
mqkAoIRlwv/+UsBhUX/SxEGPwFejPdO0
=GxPR
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Tony Anecito <ad...@yahoo.com>.
>From the definitions and discussions on google regarding loopback interface
(localhost) it will go through the nic card for Windows unless you install a
loopback adaptor. For Linux you do not need a nic card.
I was using windows and thus it used my nic card.
-Tony
----- Original Message ----
From: Christopher Schultz <ch...@christopherschultz.net>
To: Tomcat Users List <us...@tomcat.apache.org>
Sent: Mon, March 14, 2011 11:25:35 AM
Subject: Re: Performance 5.5 vs 6 vs 7
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Tony,
On 3/13/2011 5:01 PM, Tony Anecito wrote:
> Also, I have measured around 600microseconds for resonse time when using
> localhost and running the client on the same server as Tomcat but that was
>still
>
> going through the network card om by motherboard.
If you really were using localhost, I suspect your NIC did not
participate in the conversation.
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk1+T48ACgkQ9CaO5/Lv0PAHvgCgkMaYJQgto4WgituoxS7Dobog
S9oAoJPVvmoEZ9iRHLA+O3ToOkkEeQwM
=50hY
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Tony,
On 3/13/2011 5:01 PM, Tony Anecito wrote:
> Also, I have measured around 600microseconds for resonse time when using
> localhost and running the client on the same server as Tomcat but that was still
> going through the network card om by motherboard.
If you really were using localhost, I suspect your NIC did not
participate in the conversation.
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk1+T48ACgkQ9CaO5/Lv0PAHvgCgkMaYJQgto4WgituoxS7Dobog
S9oAoJPVvmoEZ9iRHLA+O3ToOkkEeQwM
=50hY
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by David Kerber <dc...@verizon.net>.
On 3/13/2011 5:01 PM, Tony Anecito wrote:
> Hi Don,
>
> As someone mentioned the network can imit you. If your bandwidth utilization is
> at 60% or over you are in trouble since collisions start to become a serious
> issue.
Ah, that's something I had not thought of. I'm running ~80% network
utilization at the times when I see the trouble, so that might well be
the root of the issue.
Thanks!
D
> Also, I have measured around 600microseconds for resonse time when using
> localhost and running the client on the same server as Tomcat but that was still
> going through the network card om by motherboard. I looked ito gettting a faster
> network card but I kept hearing 2ms was the best you could get but I got down to
> 1.2ms running from my laptop to my motherboard so my router (I have a high speed
> gaming router) and network circuitry on my systems is the current limiting
> factor and not Tomcat.
>
> Regards,
> -Tony
>
>
>
> ----- Original Message ----
> From: David Kerber<dc...@verizon.net>
> To: Tomcat Users List<us...@tomcat.apache.org>
> Sent: Sat, March 12, 2011 6:41:30 AM
> Subject: Re: Performance 5.5 vs 6 vs 7
>
> I've thought about that, but I would think I would be maxing out the cpu
> if tomcat's processing were the limit. I'm running only about 8% cpu
> usage across all 4 cpu's.
>
>
> On 3/11/2011 4:23 PM, Tony Anecito wrote:
>> Hi David,
>>
>> I recently ran performance tests against version 7 and during that whole
>> process
>> discovered at best for a single instance of Tomcat 7 running on a 6 core AMD
>> 2.8Ghz processor the best I could get running the client on the server and
>> using
>> loopback network (Localhost) was around 32K per second hitting a servlet that
>> had no code. I was using APR also. By adding the network domain name where it
>> forced the requests to go through my gaming router I could only get 2.2K
>> requests per second. If I ran the client from a laptop on the lan I got around
>> the same number.
>>
>> I have not looked into adding another instance of Tomcat but I suspect as you
>> add latency and some form of code in a servlet it might get down to 100
>> requests
>> per second. Remember if a single request takes 10msec at Tomcat and depending
>> upon the amount of data and latency (say 100msec) you are already down to 100
>> req per second at best. You might only get that at best.
>>
>> Regards,
>> -Tony
>>
>>
>>
>> ----- Original Message ----
>> From: David kerber<dc...@verizon.net>
>> To: Tomcat Users List<us...@tomcat.apache.org>
>> Sent: Fri, March 11, 2011 1:02:12 PM
>> Subject: Performance 5.5 vs 6 vs 7
>>
>> How might I expect the performance of 7 to compare to a late rev of 5.5 or
>> 6.0.x, running under JRE 6? Specifically, my application has lots of small,
>> simple requests, and the app does nothing but de-obfuscate (it's not really
>> sophisticated enough to call it encryption) the data and save it to disk. But
>> it does a LOT of them, 60 to 100 per second during normal operation, more
>> during
>> very busy times. The overall average when I checked last week was about 60
>> requests per second for the previous 10 days, around the clock,>4.5M per day.
>>
>> I spent quite a bit of effort in optimizing my app last year, and got some big
>> gains in it, but now I'm again starting to not be able to keep up with the
>> increased flow as our customer has doubled their number of installations, so
>> I'm
>> looking elsewhere for potential improvements. I've already checked my
>> connection
>> bandwidth, and that still has some headroom, though not a lot. Can I expect
> TC
>> 6 or 7 to handle larger numbers of very simple requests than 5.5 can?
>>
>> Dave
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Tony Anecito <ad...@yahoo.com>.
Hi Don,
As someone mentioned the network can imit you. If your bandwidth utilization is
at 60% or over you are in trouble since collisions start to become a serious
issue.
Also, I have measured around 600microseconds for resonse time when using
localhost and running the client on the same server as Tomcat but that was still
going through the network card om by motherboard. I looked ito gettting a faster
network card but I kept hearing 2ms was the best you could get but I got down to
1.2ms running from my laptop to my motherboard so my router (I have a high speed
gaming router) and network circuitry on my systems is the current limiting
factor and not Tomcat.
Regards,
-Tony
----- Original Message ----
From: David Kerber <dc...@verizon.net>
To: Tomcat Users List <us...@tomcat.apache.org>
Sent: Sat, March 12, 2011 6:41:30 AM
Subject: Re: Performance 5.5 vs 6 vs 7
I've thought about that, but I would think I would be maxing out the cpu
if tomcat's processing were the limit. I'm running only about 8% cpu
usage across all 4 cpu's.
On 3/11/2011 4:23 PM, Tony Anecito wrote:
> Hi David,
>
> I recently ran performance tests against version 7 and during that whole
>process
> discovered at best for a single instance of Tomcat 7 running on a 6 core AMD
> 2.8Ghz processor the best I could get running the client on the server and
>using
> loopback network (Localhost) was around 32K per second hitting a servlet that
> had no code. I was using APR also. By adding the network domain name where it
> forced the requests to go through my gaming router I could only get 2.2K
> requests per second. If I ran the client from a laptop on the lan I got around
> the same number.
>
> I have not looked into adding another instance of Tomcat but I suspect as you
> add latency and some form of code in a servlet it might get down to 100
>requests
> per second. Remember if a single request takes 10msec at Tomcat and depending
> upon the amount of data and latency (say 100msec) you are already down to 100
> req per second at best. You might only get that at best.
>
> Regards,
> -Tony
>
>
>
> ----- Original Message ----
> From: David kerber<dc...@verizon.net>
> To: Tomcat Users List<us...@tomcat.apache.org>
> Sent: Fri, March 11, 2011 1:02:12 PM
> Subject: Performance 5.5 vs 6 vs 7
>
> How might I expect the performance of 7 to compare to a late rev of 5.5 or
> 6.0.x, running under JRE 6? Specifically, my application has lots of small,
> simple requests, and the app does nothing but de-obfuscate (it's not really
> sophisticated enough to call it encryption) the data and save it to disk. But
> it does a LOT of them, 60 to 100 per second during normal operation, more
>during
> very busy times. The overall average when I checked last week was about 60
> requests per second for the previous 10 days, around the clock,>4.5M per day.
>
> I spent quite a bit of effort in optimizing my app last year, and got some big
> gains in it, but now I'm again starting to not be able to keep up with the
> increased flow as our customer has doubled their number of installations, so
>I'm
> looking elsewhere for potential improvements. I've already checked my
>connection
> bandwidth, and that still has some headroom, though not a lot. Can I expect
TC
> 6 or 7 to handle larger numbers of very simple requests than 5.5 can?
>
> Dave
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by David Kerber <dc...@verizon.net>.
I've thought about that, but I would think I would be maxing out the cpu
if tomcat's processing were the limit. I'm running only about 8% cpu
usage across all 4 cpu's.
On 3/11/2011 4:23 PM, Tony Anecito wrote:
> Hi David,
>
> I recently ran performance tests against version 7 and during that whole process
> discovered at best for a single instance of Tomcat 7 running on a 6 core AMD
> 2.8Ghz processor the best I could get running the client on the server and using
> loopback network (Localhost) was around 32K per second hitting a servlet that
> had no code. I was using APR also. By adding the network domain name where it
> forced the requests to go through my gaming router I could only get 2.2K
> requests per second. If I ran the client from a laptop on the lan I got around
> the same number.
>
> I have not looked into adding another instance of Tomcat but I suspect as you
> add latency and some form of code in a servlet it might get down to 100 requests
> per second. Remember if a single request takes 10msec at Tomcat and depending
> upon the amount of data and latency (say 100msec) you are already down to 100
> req per second at best. You might only get that at best.
>
> Regards,
> -Tony
>
>
>
> ----- Original Message ----
> From: David kerber<dc...@verizon.net>
> To: Tomcat Users List<us...@tomcat.apache.org>
> Sent: Fri, March 11, 2011 1:02:12 PM
> Subject: Performance 5.5 vs 6 vs 7
>
> How might I expect the performance of 7 to compare to a late rev of 5.5 or
> 6.0.x, running under JRE 6? Specifically, my application has lots of small,
> simple requests, and the app does nothing but de-obfuscate (it's not really
> sophisticated enough to call it encryption) the data and save it to disk. But
> it does a LOT of them, 60 to 100 per second during normal operation, more during
> very busy times. The overall average when I checked last week was about 60
> requests per second for the previous 10 days, around the clock,>4.5M per day.
>
> I spent quite a bit of effort in optimizing my app last year, and got some big
> gains in it, but now I'm again starting to not be able to keep up with the
> increased flow as our customer has doubled their number of installations, so I'm
> looking elsewhere for potential improvements. I've already checked my connection
> bandwidth, and that still has some headroom, though not a lot. Can I expect TC
> 6 or 7 to handle larger numbers of very simple requests than 5.5 can?
>
> Dave
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Tony Anecito <ad...@yahoo.com>.
Hi David,
I recently ran performance tests against version 7 and during that whole process
discovered at best for a single instance of Tomcat 7 running on a 6 core AMD
2.8Ghz processor the best I could get running the client on the server and using
loopback network (Localhost) was around 32K per second hitting a servlet that
had no code. I was using APR also. By adding the network domain name where it
forced the requests to go through my gaming router I could only get 2.2K
requests per second. If I ran the client from a laptop on the lan I got around
the same number.
I have not looked into adding another instance of Tomcat but I suspect as you
add latency and some form of code in a servlet it might get down to 100 requests
per second. Remember if a single request takes 10msec at Tomcat and depending
upon the amount of data and latency (say 100msec) you are already down to 100
req per second at best. You might only get that at best.
Regards,
-Tony
----- Original Message ----
From: David kerber <dc...@verizon.net>
To: Tomcat Users List <us...@tomcat.apache.org>
Sent: Fri, March 11, 2011 1:02:12 PM
Subject: Performance 5.5 vs 6 vs 7
How might I expect the performance of 7 to compare to a late rev of 5.5 or
6.0.x, running under JRE 6? Specifically, my application has lots of small,
simple requests, and the app does nothing but de-obfuscate (it's not really
sophisticated enough to call it encryption) the data and save it to disk. But
it does a LOT of them, 60 to 100 per second during normal operation, more during
very busy times. The overall average when I checked last week was about 60
requests per second for the previous 10 days, around the clock, >4.5M per day.
I spent quite a bit of effort in optimizing my app last year, and got some big
gains in it, but now I'm again starting to not be able to keep up with the
increased flow as our customer has doubled their number of installations, so I'm
looking elsewhere for potential improvements. I've already checked my connection
bandwidth, and that still has some headroom, though not a lot. Can I expect TC
6 or 7 to handle larger numbers of very simple requests than 5.5 can?
Dave
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Performance 5.5 vs 6 vs 7
Posted by Peter Crowther <pe...@melandra.com>.
On 11 March 2011 20:02, David kerber <dc...@verizon.net> wrote:
> I've already checked my connection bandwidth, and that still has some
> headroom, though not a lot.
>
> What's "not a lot"? If latency across the connection is starting to
increase due to congestion, that could increase the time to process
requests, therefore increase the number of requests still active, therefore
consume more resources on the server.
Also, what is the connection and what's upstream of it? Might upstream
congestion be an issue?
- Peter
Re: Performance 5.5 vs 6 vs 7
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
David,
On 3/11/2011 3:02 PM, David kerber wrote:
> How might I expect the performance of 7 to compare to a late rev of 5.5
> or 6.0.x, running under JRE 6? Specifically, my application has lots of
> small, simple requests, and the app does nothing but de-obfuscate (it's
> not really sophisticated enough to call it encryption) the data and save
> it to disk. But it does a LOT of them, 60 to 100 per second during
> normal operation, more during very busy times. The overall average when
> I checked last week was about 60 requests per second for the previous 10
> days, around the clock, >4.5M per day.
>
> I spent quite a bit of effort in optimizing my app last year, and got
> some big gains in it, but now I'm again starting to not be able to keep
> up with the increased flow as our customer has doubled their number of
> installations, so I'm looking elsewhere for potential improvements. I've
> already checked my connection bandwidth, and that still has some
> headroom, though not a lot. Can I expect TC 6 or 7 to handle larger
> numbers of very simple requests than 5.5 can?
Do you know if your problem is latency or throughput?
If latency is a problem, switching to the NIO connector may help handle
more connections within a given environment.
Tomcat 6 claims to have "memory usage optimizations" and "advanced IO
capabilities" which would indicate in general that performance is improved.
Tomcat 7 has "lots of internal code clean-up" which often, though not
always, results in a performance improvement.
As always, you should test your actual application to determine any
performance boost you would expect to see in a production environment.
I think you may be looking at an infrastructure upgrade, though.
Good luck,
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk16jkkACgkQ9CaO5/Lv0PBGIgCfSl8CkdzBhrR1hoXTEIHG9ehK
q18An3niUMS11uk1c5F+Hq9Imknwi0n2
=ahcF
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org