Serious scalability issues with Oncourse at Indiana U

[R.P. Aditya]

On Sun, Sep 30, 2007 at 10:12:03PM -0400, Charles Hedrick wrote:
> Interesting. Our mysql server shows 691 queries / sec, and that is  

Linda was talking about average http requests per second -- we've managed in
certain load tests to average 1000+ http requests per second, but I would not
characterize any load test to be a simulation of "reality" (reality always
seems to be far _less_ demanding!)  -- db requests per second will be far
lower, about 30% of the max on average (about 40% of http requests we see are
for static resources, and 50% of the remaining are presence related)

> for its whole uptime. This is partly influenced by the fact that we  
> were missing a key patch to tests and quizzes, and thus we were  
> pounding the server when students were doing Tests and Quizzes. But  
> as an average over our uptime, surely our rates for peak periods are  
> a lot higher. We're still in the process of setting up statistics, so  
> I don't have query rates except this one data point.

depending on what the queries are and where the db server is responding to
them it could be entirely prosaic, or you have a usage pattern entirely unlike
UofMs (we don't run tests and quizzes in CTools for example)

Adi

> On Sep 6, 2007, at 5:06 PM, Linda M Place wrote:
>
> >And we've pushed 600+ requests per second during some high end load  
> >tests
> >without killing the Oracle server.  In other words, we know that  
> >the server
> >can handle considerably more load, if load means nothing beyond  
> >requests per
> >second.
>

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[R.P. Aditya]

On Mon, Oct 01, 2007 at 03:16:23AM +0000, R.P. Aditya wrote:

> On Sun, Sep 30, 2007 at 10:12:03PM -0400, Charles Hedrick wrote:
> > Interesting. Our mysql server shows 691 queries / sec, and that is  
> > for its whole uptime. This is partly influenced by the fact that we  
> > were missing a key patch to tests and quizzes, and thus we were  
> > pounding the server when students were doing Tests and Quizzes. But  
> > as an average over our uptime, surely our rates for peak periods are  
> > a lot higher. We're still in the process of setting up statistics, so  
> > I don't have query rates except this one data point.
>
> depending on what the queries are and where the db server is responding to
> them it could be entirely prosaic, or you have a usage pattern entirely unlike
> UofMs (we don't run tests and quizzes in CTools for example)

oh, the other part of it is what is the average response time for those
queries? for example, since this semester began, based on 5-minute samples:

- the time-to-first-byte for http responses has averaged 1.2 seconds across
all hits as measured by our load-balancer (average of 100 http hits per
second)

- the db response time per transaction has averaged 411ms (but also includes
non-CTools hits and backup/updates etc.) as measured by Oracle (average of 18
queries per second)

I'd be very interested in seeing other folks' stats...

Adi

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[jayshao]

On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:

> - the time-to-first-byte for http responses has averaged 1.2  
> seconds across
> all hits as measured by our load-balancer (average of 100 http hits  
> per
> second)
>
> - the db response time per transaction has averaged 411ms (but also  
> includes
> non-CTools hits and backup/updates etc.) as measured by Oracle  
> (average of 18
> queries per second)

Actually, I'm very interested in what stats everyone is keeping and  
how they get them -- also why they're collecting them (capacity  
planning, institutional research, etc.)

Any good writeups already available?

Jason

--

Jason Shao
Application Developer
Rutgers University, Office of Instructional & Research Technology
v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
jay.shao.org


----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Peter Knoop]

Hi Jason,

The Production WG would be a better place to continue discussion on
statistics and their uses for production deployments of Sakai, so I've
cc'ed it here.  (If you can remember, please remove sakai-dev from
further replies.)

-peter

> -----Original Message-----
> From: Jason Shao [mailto:jayshao@...]
> Sent: Thursday, October 04, 2007 6:10 AM
> To: sakai-dev@...
> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
> Oncourse at Indiana U)
>
> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
>
> > - the time-to-first-byte for http responses has averaged 1.2
> > seconds across
> > all hits as measured by our load-balancer (average of 100 http hits
> > per
> > second)
> >
> > - the db response time per transaction has averaged 411ms (but also
> > includes
> > non-CTools hits and backup/updates etc.) as measured by Oracle
> > (average of 18
> > queries per second)
>
> Actually, I'm very interested in what stats everyone is keeping and
> how they get them -- also why they're collecting them (capacity
> planning, institutional research, etc.)
>
> Any good writeups already available?
>
> Jason
>
> --
>
> Jason Shao
> Application Developer
> Rutgers University, Office of Instructional & Research Technology
> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
> jay.shao.org
>
>
> ----------------------
> This automatic notification message was sent by Sakai Collab
> (https://collab.sakaiproject.org/portal) from the DG: Development
> (a.k.a. sakai-dev) site.
> You can modify how you receive notifications at My Workspace >
> Preferences.

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Yuji Shinozaki]

One peripheral question about the Production WG:  Is it joinable?  I  
don't seem to see it.

yuji
----


On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:

> Hi Jason,
>
> The Production WG would be a better place to continue discussion on
> statistics and their uses for production deployments of Sakai, so I've
> cc'ed it here.  (If you can remember, please remove sakai-dev from
> further replies.)
>
> -peter
>
>> -----Original Message-----
>> From: Jason Shao [mailto:jayshao@...]
>> Sent: Thursday, October 04, 2007 6:10 AM
>> To: sakai-dev@...
>> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
>> Oncourse at Indiana U)
>>
>> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
>>
>>> - the time-to-first-byte for http responses has averaged 1.2
>>> seconds across
>>> all hits as measured by our load-balancer (average of 100 http hits
>>> per
>>> second)
>>>
>>> - the db response time per transaction has averaged 411ms (but also
>>> includes
>>> non-CTools hits and backup/updates etc.) as measured by Oracle
>>> (average of 18
>>> queries per second)
>>
>> Actually, I'm very interested in what stats everyone is keeping and
>> how they get them -- also why they're collecting them (capacity
>> planning, institutional research, etc.)
>>
>> Any good writeups already available?
>>
>> Jason
>>
>> --
>>
>> Jason Shao
>> Application Developer
>> Rutgers University, Office of Instructional & Research Technology
>> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
>> jay.shao.org
>>
>>
>> ----------------------
>> This automatic notification message was sent by Sakai Collab
>> (https://collab.sakaiproject.org/portal) from the DG: Development
>> (a.k.a. sakai-dev) site.
>> You can modify how you receive notifications at My Workspace >
>> Preferences.
>
> ----------------------
> This automatic notification message was sent by Sakai Collab  
> (https://collab.sakaiproject.org/portal) from the DG: Development  
> (a.k.a. sakai-dev) site.
> You can modify how you receive notifications at My Workspace >  
> Preferences.

-----
Yuji Shinozaki
Sr. Technical Lead/Project Manager
University of Virginia
Advanced Technologies Group
ys2n@...


----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Anthony Whyte]

The Collab Production work group is public and joinable.  It's titled:

WG: Production

Cheers,

Anthony


On Oct 4, 2007, at 9:23 AM, Yuji Shinozaki wrote:

>
> One peripheral question about the Production WG:  Is it joinable?  
> I don't seem to see it.
>
> yuji
> ----
>
>
> On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:
>
>> Hi Jason,
>>
>> The Production WG would be a better place to continue discussion on
>> statistics and their uses for production deployments of Sakai, so  
>> I've
>> cc'ed it here.  (If you can remember, please remove sakai-dev from
>> further replies.)
>>
>> -peter
>>
>>> -----Original Message-----
>>> From: Jason Shao [mailto:jayshao@...]
>>> Sent: Thursday, October 04, 2007 6:10 AM
>>> To: sakai-dev@...
>>> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
>>> Oncourse at Indiana U)
>>>
>>> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
>>>
>>>> - the time-to-first-byte for http responses has averaged 1.2
>>>> seconds across
>>>> all hits as measured by our load-balancer (average of 100 http hits
>>>> per
>>>> second)
>>>>
>>>> - the db response time per transaction has averaged 411ms (but also
>>>> includes
>>>> non-CTools hits and backup/updates etc.) as measured by Oracle
>>>> (average of 18
>>>> queries per second)
>>>
>>> Actually, I'm very interested in what stats everyone is keeping and
>>> how they get them -- also why they're collecting them (capacity
>>> planning, institutional research, etc.)
>>>
>>> Any good writeups already available?
>>>
>>> Jason
>>>
>>> --
>>>
>>> Jason Shao
>>> Application Developer
>>> Rutgers University, Office of Instructional & Research Technology
>>> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
>>> jay.shao.org
>>>
>>>
>>> ----------------------
>>> This automatic notification message was sent by Sakai Collab
>>> (https://collab.sakaiproject.org/portal) from the DG: Development
>>> (a.k.a. sakai-dev) site.
>>> You can modify how you receive notifications at My Workspace >
>>> Preferences.
>>
>> ----------------------
>> This automatic notification message was sent by Sakai Collab  
>> (https://collab.sakaiproject.org/portal) from the DG: Development  
>> (a.k.a. sakai-dev) site.
>> You can modify how you receive notifications at My Workspace >  
>> Preferences.
>
> -----
> Yuji Shinozaki
> Sr. Technical Lead/Project Manager
> University of Virginia
> Advanced Technologies Group
> ys2n@...
>
>
> ----------------------
> This automatic notification message was sent by Sakai Collab  
> (https://collab.sakaiproject.org/portal) from the DG: Development  
> (a.k.a. sakai-dev) site.
> You can modify how you receive notifications at My Workspace >  
> Preferences.
>
>
>

[see attachment: "smime.p7s", size: 2417 bytes]


Attachments:

smime.p7s
https://collab.sakaiproject.org/access/content/attachment/ff419982-15f1-465c-00a1-269a0898308e/smime.p7s

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Yuji Shinozaki]

Thanks!

Hmmm.  Searching for "Production" in joinable worksites yielded  
nothing and I had scrolled to the "P's" for "Production" not  
realizing it was actually "WG: Production".

Is the site search on sakaiproject.org broken?

yuji
----

On Oct 4, 2007, at 9:36 AM, Anthony Whyte wrote:

> The Collab Production work group is public and joinable.  It's titled:
>
> WG: Production
>
> Cheers,
>
> Anthony
>
>
> On Oct 4, 2007, at 9:23 AM, Yuji Shinozaki wrote:
>
>>
>> One peripheral question about the Production WG:  Is it joinable?  
>> I don't seem to see it.
>>
>> yuji
>> ----
>>
>>
>> On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:
>>
>>> Hi Jason,
>>>
>>> The Production WG would be a better place to continue discussion on
>>> statistics and their uses for production deployments of Sakai, so  
>>> I've
>>> cc'ed it here.  (If you can remember, please remove sakai-dev from
>>> further replies.)
>>>
>>> -peter
>>>
>>>> -----Original Message-----
>>>> From: Jason Shao [mailto:jayshao@...]
>>>> Sent: Thursday, October 04, 2007 6:10 AM
>>>> To: sakai-dev@...
>>>> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
>>>> Oncourse at Indiana U)
>>>>
>>>> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
>>>>
>>>>> - the time-to-first-byte for http responses has averaged 1.2
>>>>> seconds across
>>>>> all hits as measured by our load-balancer (average of 100 http  
>>>>> hits
>>>>> per
>>>>> second)
>>>>>
>>>>> - the db response time per transaction has averaged 411ms (but  
>>>>> also
>>>>> includes
>>>>> non-CTools hits and backup/updates etc.) as measured by Oracle
>>>>> (average of 18
>>>>> queries per second)
>>>>
>>>> Actually, I'm very interested in what stats everyone is keeping and
>>>> how they get them -- also why they're collecting them (capacity
>>>> planning, institutional research, etc.)
>>>>
>>>> Any good writeups already available?
>>>>
>>>> Jason
>>>>
>>>> --
>>>>
>>>> Jason Shao
>>>> Application Developer
>>>> Rutgers University, Office of Instructional & Research Technology
>>>> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
>>>> jay.shao.org
>>>>
>>>>
>>>> ----------------------
>>>> This automatic notification message was sent by Sakai Collab
>>>> (https://collab.sakaiproject.org/portal) from the DG: Development
>>>> (a.k.a. sakai-dev) site.
>>>> You can modify how you receive notifications at My Workspace >
>>>> Preferences.
>>>
>>> ----------------------
>>> This automatic notification message was sent by Sakai Collab  
>>> (https://collab.sakaiproject.org/portal) from the DG: Development  
>>> (a.k.a. sakai-dev) site.
>>> You can modify how you receive notifications at My Workspace >  
>>> Preferences.
>>
>> -----
>> Yuji Shinozaki
>> Sr. Technical Lead/Project Manager
>> University of Virginia
>> Advanced Technologies Group
>> ys2n@...
>>
>>
>> ----------------------
>> This automatic notification message was sent by Sakai Collab  
>> (https://collab.sakaiproject.org/portal) from the DG: Development  
>> (a.k.a. sakai-dev) site.
>> You can modify how you receive notifications at My Workspace >  
>> Preferences.
>>
>>
>>
>
> [see attachment: "smime.p7s", size: 2417 bytes]
>
>
> Attachments:
>
> smime.p7s
> https://collab.sakaiproject.org/access/content/attachment/ 
> ff419982-15f1-465c-00a1-269a0898308e/smime.p7s
>
> ----------------------
> This automatic notification message was sent by Sakai Collab  
> (https://collab.sakaiproject.org/portal) from the DG: Development  
> (a.k.a. sakai-dev) site.
> You can modify how you receive notifications at My Workspace >  
> Preferences.

-----
Yuji Shinozaki
Sr. Technical Lead/Project Manager
University of Virginia
Advanced Technologies Group
ys2n@...


----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Yuji Shinozaki]

On Oct 4, 2007, at 11:01 AM, Yuji Shinozaki wrote:

>
> Thanks!
>
> Hmmm.  Searching for "Production" in joinable worksites yielded  
> nothing and I had scrolled to the "P's" for "Production" not  
> realizing it was actually "WG: Production".
>
> Is the site search on sakaiproject.org broken?
>

To avoid confusion:  I actually meant "Is the joinable sites search  
on collab.sakaiproject.org broken?".

yuji
----

> yuji
> ----
>
> On Oct 4, 2007, at 9:36 AM, Anthony Whyte wrote:
>
>> The Collab Production work group is public and joinable.  It's  
>> titled:
>>
>> WG: Production
>>
>> Cheers,
>>
>> Anthony
>>
>>
>> On Oct 4, 2007, at 9:23 AM, Yuji Shinozaki wrote:
>>
>>>
>>> One peripheral question about the Production WG:  Is it  
>>> joinable?  I don't seem to see it.
>>>
>>> yuji
>>> ----
>>>
>>>
>>> On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:
>>>
>>>> Hi Jason,
>>>>
>>>> The Production WG would be a better place to continue discussion on
>>>> statistics and their uses for production deployments of Sakai,  
>>>> so I've
>>>> cc'ed it here.  (If you can remember, please remove sakai-dev from
>>>> further replies.)
>>>>
>>>> -peter
>>>>
>>>>> -----Original Message-----
>>>>> From: Jason Shao [mailto:jayshao@...]
>>>>> Sent: Thursday, October 04, 2007 6:10 AM
>>>>> To: sakai-dev@...
>>>>> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
>>>>> Oncourse at Indiana U)
>>>>>
>>>>> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
>>>>>
>>>>>> - the time-to-first-byte for http responses has averaged 1.2
>>>>>> seconds across
>>>>>> all hits as measured by our load-balancer (average of 100 http  
>>>>>> hits
>>>>>> per
>>>>>> second)
>>>>>>
>>>>>> - the db response time per transaction has averaged 411ms (but  
>>>>>> also
>>>>>> includes
>>>>>> non-CTools hits and backup/updates etc.) as measured by Oracle
>>>>>> (average of 18
>>>>>> queries per second)
>>>>>
>>>>> Actually, I'm very interested in what stats everyone is keeping  
>>>>> and
>>>>> how they get them -- also why they're collecting them (capacity
>>>>> planning, institutional research, etc.)
>>>>>
>>>>> Any good writeups already available?
>>>>>
>>>>> Jason
>>>>>
>>>>> --
>>>>>
>>>>> Jason Shao
>>>>> Application Developer
>>>>> Rutgers University, Office of Instructional & Research Technology
>>>>> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
>>>>> jay.shao.org
>>>>>
>>>>>
>>>>> ----------------------
>>>>> This automatic notification message was sent by Sakai Collab
>>>>> (https://collab.sakaiproject.org/portal) from the DG: Development
>>>>> (a.k.a. sakai-dev) site.
>>>>> You can modify how you receive notifications at My Workspace >
>>>>> Preferences.
>>>>
>>>> ----------------------
>>>> This automatic notification message was sent by Sakai Collab  
>>>> (https://collab.sakaiproject.org/portal) from the DG:  
>>>> Development (a.k.a. sakai-dev) site.
>>>> You can modify how you receive notifications at My Workspace >  
>>>> Preferences.
>>>
>>> -----
>>> Yuji Shinozaki
>>> Sr. Technical Lead/Project Manager
>>> University of Virginia
>>> Advanced Technologies Group
>>> ys2n@...
>>>
>>>
>>> ----------------------
>>> This automatic notification message was sent by Sakai Collab  
>>> (https://collab.sakaiproject.org/portal) from the DG: Development  
>>> (a.k.a. sakai-dev) site.
>>> You can modify how you receive notifications at My Workspace >  
>>> Preferences.
>>>
>>>
>>>
>>
>> [see attachment: "smime.p7s", size: 2417 bytes]
>>
>>
>> Attachments:
>>
>> smime.p7s
>> https://collab.sakaiproject.org/access/content/attachment/ 
>> ff419982-15f1-465c-00a1-269a0898308e/smime.p7s
>>
>> ----------------------
>> This automatic notification message was sent by Sakai Collab  
>> (https://collab.sakaiproject.org/portal) from the DG: Development  
>> (a.k.a. sakai-dev) site.
>> You can modify how you receive notifications at My Workspace >  
>> Preferences.
>
> -----
> Yuji Shinozaki
> Sr. Technical Lead/Project Manager
> University of Virginia
> Advanced Technologies Group
> ys2n@...
>
>
> ----------------------
> This automatic notification message was sent by Sakai Collab  
> (https://collab.sakaiproject.org/portal) from the DG: Development  
> (a.k.a. sakai-dev) site.
> You can modify how you receive notifications at My Workspace >  
> Preferences.

-----
Yuji Shinozaki
Sr. Technical Lead/Project Manager
University of Virginia
Advanced Technologies Group
ys2n@...


----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Peter Knoop]

Yes, the Production Working Group is joinable.  To join the email list
on Collab (http://collab.sakaiproject.org), look for the "WG:
Production" site in the list of joinable sites.  There is also a
companion Confluence site of the same title. It's a better place for
production-related topics, as there are folks on the production list who
are interested in such things, but who are not on sakai-dev as they do
not have the time to sort through all the traffic for items of interest.

-peter

> -----Original Message-----
> From: Yuji Shinozaki [mailto:ys2n@...]
> Sent: Thursday, October 04, 2007 9:24 AM
> To: Knoop, Peter
> Cc: Jason Shao; sakai-dev@...;
> production@...
> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
> Oncourse at Indiana U)
>
>
> One peripheral question about the Production WG:  Is it joinable?  I
> don't seem to see it.
>
> yuji
> ----
>
>
> On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:
>
> > Hi Jason,
> >
> > The Production WG would be a better place to continue discussion on
> > statistics and their uses for production deployments of Sakai, so
> I've
> > cc'ed it here.  (If you can remember, please remove sakai-dev from
> > further replies.)
> >
> > -peter
> >
> >> -----Original Message-----
> >> From: Jason Shao [mailto:jayshao@...]
> >> Sent: Thursday, October 04, 2007 6:10 AM
> >> To: sakai-dev@...
> >> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues with
> >> Oncourse at Indiana U)
> >>
> >> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
> >>
> >>> - the time-to-first-byte for http responses has averaged 1.2
> >>> seconds across
> >>> all hits as measured by our load-balancer (average of 100 http

hits
> >>> per
> >>> second)
> >>>
> >>> - the db response time per transaction has averaged 411ms (but
also

> >>> includes
> >>> non-CTools hits and backup/updates etc.) as measured by Oracle
> >>> (average of 18
> >>> queries per second)
> >>
> >> Actually, I'm very interested in what stats everyone is keeping and
> >> how they get them -- also why they're collecting them (capacity
> >> planning, institutional research, etc.)
> >>
> >> Any good writeups already available?
> >>
> >> Jason
> >>
> >> --
> >>
> >> Jason Shao
> >> Application Developer
> >> Rutgers University, Office of Instructional & Research Technology
> >> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
> >> jay.shao.org
> >>
> >>
> >> ----------------------
> >> This automatic notification message was sent by Sakai Collab
> >> (https://collab.sakaiproject.org/portal) from the DG: Development
> >> (a.k.a. sakai-dev) site.
> >> You can modify how you receive notifications at My Workspace >
> >> Preferences.
> >
> > ----------------------
> > This automatic notification message was sent by Sakai Collab
> > (https://collab.sakaiproject.org/portal) from the DG: Development
> > (a.k.a. sakai-dev) site.
> > You can modify how you receive notifications at My Workspace >
> > Preferences.
>
> -----
> Yuji Shinozaki
> Sr. Technical Lead/Project Manager
> University of Virginia
> Advanced Technologies Group
> ys2n@...
>

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[Peter Knoop]

Definitely a bug.  See the following jira, if you want to add any more
info:

http://bugs.sakaiproject.org/jira/browse/SAK-11830

-peter

> -----Original Message-----
> From: Yuji Shinozaki [mailto:ys2n@...]
> Sent: Thursday, October 04, 2007 11:02 AM
> To: Whyte, Anthony
> Cc: Knoop, Peter; Jason Shao; sakai-dev@...;
> production@...
> Subject: Re: Production WG (public and joinable)
>
>
> Thanks!
>
> Hmmm.  Searching for "Production" in joinable worksites yielded
> nothing and I had scrolled to the "P's" for "Production" not
> realizing it was actually "WG: Production".
>
> Is the site search on sakaiproject.org broken?
>
> yuji
> ----
>
> On Oct 4, 2007, at 9:36 AM, Anthony Whyte wrote:
>
> > The Collab Production work group is public and joinable.  It's
> titled:
> >
> > WG: Production
> >
> > Cheers,
> >
> > Anthony
> >
> >
> > On Oct 4, 2007, at 9:23 AM, Yuji Shinozaki wrote:
> >
> >>
> >> One peripheral question about the Production WG:  Is it joinable?
> >> I don't seem to see it.
> >>
> >> yuji
> >> ----
> >>
> >>
> >> On Oct 4, 2007, at 9:17 AM, Knoop, Peter wrote:
> >>
> >>> Hi Jason,
> >>>
> >>> The Production WG would be a better place to continue discussion

on

> >>> statistics and their uses for production deployments of Sakai, so
> >>> I've
> >>> cc'ed it here.  (If you can remember, please remove sakai-dev from
> >>> further replies.)
> >>>
> >>> -peter
> >>>
> >>>> -----Original Message-----
> >>>> From: Jason Shao [mailto:jayshao@...]
> >>>> Sent: Thursday, October 04, 2007 6:10 AM
> >>>> To: sakai-dev@...
> >>>> Subject: Re: dbcp vs c3p0 (was Re: Serious scalability issues

with

> >>>> Oncourse at Indiana U)
> >>>>
> >>>> On Sep 30, 2007, at 11:33 PM, R.P. Aditya wrote:
> >>>>
> >>>>> - the time-to-first-byte for http responses has averaged 1.2
> >>>>> seconds across
> >>>>> all hits as measured by our load-balancer (average of 100 http
> >>>>> hits
> >>>>> per
> >>>>> second)
> >>>>>
> >>>>> - the db response time per transaction has averaged 411ms (but
> >>>>> also
> >>>>> includes
> >>>>> non-CTools hits and backup/updates etc.) as measured by Oracle
> >>>>> (average of 18
> >>>>> queries per second)
> >>>>
> >>>> Actually, I'm very interested in what stats everyone is keeping
> and
> >>>> how they get them -- also why they're collecting them (capacity
> >>>> planning, institutional research, etc.)
> >>>>
> >>>> Any good writeups already available?
> >>>>
> >>>> Jason
> >>>>
> >>>> --
> >>>>
> >>>> Jason Shao
> >>>> Application Developer
> >>>> Rutgers University, Office of Instructional & Research Technology
> >>>> v. 732-445-8726 | f. 732-445-5539 | jayshao@... | http://
> >>>> jay.shao.org
> >>>>
> >>>>
> >>>> ----------------------
> >>>> This automatic notification message was sent by Sakai Collab
> >>>> (https://collab.sakaiproject.org/portal) from the DG: Development
> >>>> (a.k.a. sakai-dev) site.
> >>>> You can modify how you receive notifications at My Workspace >
> >>>> Preferences.
> >>>
> >>> ----------------------
> >>> This automatic notification message was sent by Sakai Collab
> >>> (https://collab.sakaiproject.org/portal) from the DG: Development
> >>> (a.k.a. sakai-dev) site.
> >>> You can modify how you receive notifications at My Workspace >
> >>> Preferences.
> >>
> >> -----
> >> Yuji Shinozaki
> >> Sr. Technical Lead/Project Manager
> >> University of Virginia
> >> Advanced Technologies Group
> >> ys2n@...
> >>
> >>
> >> ----------------------
> >> This automatic notification message was sent by Sakai Collab
> >> (https://collab.sakaiproject.org/portal) from the DG: Development
> >> (a.k.a. sakai-dev) site.
> >> You can modify how you receive notifications at My Workspace >
> >> Preferences.
> >>
> >>
> >>
> >
> > [see attachment: "smime.p7s", size: 2417 bytes]
> >
> >
> > Attachments:
> >
> > smime.p7s
> > https://collab.sakaiproject.org/access/content/attachment/
> > ff419982-15f1-465c-00a1-269a0898308e/smime.p7s
> >
> > ----------------------
> > This automatic notification message was sent by Sakai Collab
> > (https://collab.sakaiproject.org/portal) from the DG: Development
> > (a.k.a. sakai-dev) site.
> > You can modify how you receive notifications at My Workspace >
> > Preferences.
>
> -----
> Yuji Shinozaki
> Sr. Technical Lead/Project Manager
> University of Virginia
> Advanced Technologies Group
> ys2n@...
>

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.

[R.P. Aditya]

On Mon, Oct 01, 2007 at 03:33:57AM +0000, R.P. Aditya wrote:

> On Mon, Oct 01, 2007 at 03:16:23AM +0000, R.P. Aditya wrote:
> > On Sun, Sep 30, 2007 at 10:12:03PM -0400, Charles Hedrick wrote:
> > > Interesting. Our mysql server shows 691 queries / sec, and that is  
> > > for its whole uptime. This is partly influenced by the fact that we  
> > > were missing a key patch to tests and quizzes, and thus we were  
> > > pounding the server when students were doing Tests and Quizzes. But  
> > > as an average over our uptime, surely our rates for peak periods are  
> > > a lot higher. We're still in the process of setting up statistics, so  
> > > I don't have query rates except this one data point.
> >
> > depending on what the queries are and where the db server is responding to
> > them it could be entirely prosaic, or you have a usage pattern entirely unlike
> > UofMs (we don't run tests and quizzes in CTools for example)
>
> oh, the other part of it is what is the average response time for those
> queries? for example, since this semester began, based on 5-minute samples:
>
> - the time-to-first-byte for http responses has averaged 1.2 seconds across
> all hits as measured by our load-balancer (average of 100 http hits per
> second)
>
> - the db response time per transaction has averaged 411ms (but also includes
> non-CTools hits and backup/updates etc.) as measured by Oracle (average of 18
> queries per second)

sorry that last stat should be 18 user transactions per second and I think the
Oracle measures of total queries per second is called "executions" per second,
and the average there is 534 per second over the last month -- more in line
with what you're seeing at Rutgers...

Adi

----------------------
This automatic notification message was sent by Sakai Collab (https://collab.sakaiproject.org/portal) from the DG: Development (a.k.a. sakai-dev) site.
You can modify how you receive notifications at My Workspace > Preferences.