headers
Ross Gardler | 2 Aug 2003 12:24

Re: Added costings to pilot plan and resources

Taran Rampersad wrote:
> Ross Gardler wrote:
> 
>> relaxer@... wrote:
>>
>>> If there is any other specfication other than 1000VA please let me 
>>> know befor next week.
>>
>>
>>
>> There is no specification other than the recomendations of this list. 
>> I know nothing about these things and so do not know what we need.
>>
>> It has been suggested (and supported) that we would be better off 
>> using a single UPS for each machine in the network. The machines will 
>> need to stay alive long enough for users to save their work and shut 
>> down cleanly. The suggested spec was 350Va for each machine rather 
>> than 1000Va for a group of machines. 
> 
> 
> 
> I agree. Single point failure for a single UPS is asking a bit much, but 
> then again - is about $450/UPS for a Cyberpower 585 AVR (as an example) 
> too much per machine?

That would really depend on the money we manage to raise and the 
availability of money in the schools themselves. In a great many cases I 
expect the schools will not be able to put any cash in (although there 
are people looking into government funding we may be able to tap).
(Continue reading)

Permalink | Reply |
headers
Taran Rampersad | 3 Aug 2003 22:49

Re: Added costings to pilot plan and resources

Ross Gardler wrote:

> Taran Rampersad wrote:
>
>> Ross Gardler wrote:
>>
>> I agree. Single point failure for a single UPS is asking a bit much, 
>> but then again - is about $450/UPS for a Cyberpower 585 AVR (as an 
>> example) too much per machine?
>
>
> That would really depend on the money we manage to raise and the 
> availability of money in the schools themselves. In a great many cases 
> I expect the schools will not be able to put any cash in (although 
> there are people looking into government funding we may be able to tap). 

Schools like Presentation College (as an example) also depend heavily on 
the past students and so forth. They do not get full government funding, 
and this needs to be realized as a variable.

>
>
>> (I, too, have price lists. I deal with wholesalers when I source 
>> things.)
>
>
> When the list is stable I think the best thing to do is to ask people 
> to give us the best price they can get. The lowest price is the one we 
> use. Personally I don't care if someone is making a few dollars off 
> the deal as long as the project gets the lowest price possible. Of
(Continue reading)

Permalink | Reply |
headers
Ross Gardler | 2 Aug 2003 14:30

Re: Added costings to pilot plan and resources

Taran Rampersad wrote:
> Ross Gardler wrote:
> 
>> Taran Rampersad wrote:
>>
>>> Ross Gardler wrote:
>>>
>>> I agree. Single point failure for a single UPS is asking a bit much, 
>>> but then again - is about $450/UPS for a Cyberpower 585 AVR (as an 
>>> example) too much per machine?
>>
>>
>>
>> That would really depend on the money we manage to raise and the 
>> availability of money in the schools themselves. In a great many cases 
>> I expect the schools will not be able to put any cash in (although 
>> there are people looking into government funding we may be able to tap). 
> 
> 
> Schools like Presentation College (as an example) also depend heavily on 
> the past students and so forth. They do not get full government funding, 
> and this needs to be realized as a variable.

This is not a variable for the pilot proposal under discussion. This is 
a proposal to get computers into underdeveloped primary schools - money 
is a real issue here and government funding *may* be available.

Of course, if you want to write a proposal for presentation college I am 
sure this community (certainly myself) will be able to offer support, 
but I'm afraid you will have to write that proposal, not I.
(Continue reading)

Permalink | Reply |
headers
Ace Suares | 3 Aug 2003 03:32
Gravatar

Re: Added costings to pilot plan and resources


> Regarding maintenance of the donated PC's, you have a good point. I
> suggest the best thing to do is to maintian a stock of parts from
> butchered machines. It is inevitable that not all donated machines will

There is two ways going about this, in my mind.

1. Make sure the project gets huge - and order new parts per the container.

For instance, order 1000 casings, 1000 motherboards, 1000 flatscreens, 1000 
256 MB DDR modules. The motherboard should have 'everything' on board: sound, 
video, LAN, USB, IEEE1394, everything. 

Then make a project out of 'maintining and repairing CIS-computers'.

That is, let the students of some tech school learn by repairing, replacing, 
diagnozing these computers. Free repair and maintenance shop !

And 'order by the 1000' gives you a real nice price on things.

2. Make sure the project gets huge - and make sure you get 1000's of donated 
computers. 

Then make a project out of 'maintining and repairing CIS-computers'.
That is, let the students of some tech school learn by repairing, replacing, 
diagnozing these computers. Free repair and maintenance shop !

Oh, and then there's the third way: make sure your software contains enough 
drivers so that they can work on any old thin client. Or make it easy to 
generate new disks on the basis of brand, type etc etc.
(Continue reading)

Permalink | Reply |
headers
Ross Gardler | 3 Aug 2003 05:17

Re: Added costings to pilot plan and resources

Ace Suares wrote:
> 2. Make sure the project gets huge - and make sure you get 1000's of donated 
> computers. 
> 
> Then make a project out of 'maintining and repairing CIS-computers'.
> That is, let the students of some tech school learn by repairing, replacing, 
> diagnozing these computers. Free repair and maintenance shop !

<snip/>

> PS I am not saying it's *easy* to get 1000 donated computers - but if you have 
> your software in place, and a Cis-repair shop in place, I am sure you can do 
> mracles.

I am with you on this approach. Although it need not be huge to start 
with. If we get 150 machines donated. Install 130 of them and strip the 
rest for spares (remember in a thin client there is very little to go 
wrong, almost no moving parts). That will serve those 120 installed 
machines for quite some time.

I also agree with your observations that having loads of drivers on the 
installation will help considerably.

And I like your idea of getting the students themselves to do the 
maintenance. Computers are so simple inside that this could even happen 
(under close supervision) at the primary level.

> 
> Reality Check:
> PS2 looking for http://www.furbie.nl/. a foundation that refurbishes donated
(Continue reading)

Permalink | Reply |
headers
Taran Rampersad | 4 Aug 2003 00:07

Re: Added costings to pilot plan and resources

Ross Gardler wrote:

> Taran Rampersad wrote:
>
>> Ross Gardler wrote:
>>
>>> Taran Rampersad wrote:
>>
>>
>>
>> Schools like Presentation College (as an example) also depend heavily 
>> on the past students and so forth. They do not get full government 
>> funding, and this needs to be realized as a variable.
>
>
> This is not a variable for the pilot proposal under discussion. This 
> is a proposal to get computers into underdeveloped primary schools - 
> money is a real issue here and government funding *may* be available.
>
> Of course, if you want to write a proposal for presentation college I 
> am sure this community (certainly myself) will be able to offer 
> support, but I'm afraid you will have to write that proposal, not I.
>
> I can assure you that you will be given access to the lab and if we 
> should get a surplus of donated machines they will also be made 
> available (when talking to potential sponsors I am not just "selling" 
> the pilot project but the interests of this community as a whole). 

I used a school as an example. Be advised that the example may be one 
that you find in other schools. This is why I took the time to label the
(Continue reading)

Permalink | Reply |
headers
Ace Suares | 2 Aug 2003 21:25
Gravatar

Re: Added costings to pilot plan and resources


> This is not likely to happen. The machines (except the server) are thin
> clients. All we need to do is keep the monitor and machine alive long
> enough to allow the students to save their work, it really doesn't
> matter if the sound system doesn't work etc

Actually, the session gets saved on the server - no need for the thin clients 
to remain open. Al least, that's what I think how thin clients work. No data 
is in thin clients memory - it's just an extention of a virtual screen, 
keyboard, mouse from the server. Do a little bit of research into that and it 
saves you huge on UPS's. Just one for the server, overdimensioned as much as 
possible, it's much better if the server keeps ronin for half an hour then 5 
minutes. Hey, for all you know the blackout is only 5 minutes, and everyone 
takes off where they left after a coffee break.

Surge protectors and a GOOD grounding on the thin clients are needed, though !

_)Ace
Permalink | Reply |
headers
Ross Gardler | 2 Aug 2003 13:30

Re: Added costings to pilot plan and resources

Ace Suares wrote:
> 
>>This is not likely to happen. The machines (except the server) are thin
>>clients. All we need to do is keep the monitor and machine alive long
>>enough to allow the students to save their work, it really doesn't
>>matter if the sound system doesn't work etc
> 
> 
> Actually, the session gets saved on the server - no need for the thin clients 
> to remain open. Al least, that's what I think how thin clients work. No data 
> is in thin clients memory - it's just an extention of a virtual screen, 
> keyboard, mouse from the server. Do a little bit of research into that and it 
> saves you huge on UPS's. Just one for the server, overdimensioned as much as 
> possible, it's much better if the server keeps ronin for half an hour then 5 
> minutes. Hey, for all you know the blackout is only 5 minutes, and everyone 
> takes off where they left after a coffee break.
> 

Excellent. I was hoping that the this would be the case. This will 
massively reduce the cost of our installations (around 40%!). I will be 
conducting some experiments when I return to T&T (actually I don't leave 
till Monday but that's moot). I'll provide a definitive answer on this then.

Ross
Permalink | Reply |
headers
Donny Mark Ramdathsingh | 4 Aug 2003 13:15
Picon

RE: Added costings to pilot plan and resources

Geez, 

These weekends conversations are HOTT !! :)

Seriously now are we saying their will be no protection on the client
side (UPS) ?? Ace is spot on with regards to sessions being executed on
the 
Server (HD and Mem) the only concern is if the clients don't have UPS 
In the event of a power failure and dies, although the Data resides on
the "server" there are concerns of abnormal termination and defunct
forks 
Which will eventually affect system performance, and should be taken
into consideration..

Donny... 

-----Original Message-----
From: owner-cis@... [mailto:owner-cis <at> saafe.org] On Behalf
Of Ross
Gardler
Sent: Saturday, August 02, 2003 7:30 AM
To: cis@...
Subject: Re: [cis] Added costings to pilot plan and resources

Ace Suares wrote:
> 
>>This is not likely to happen. The machines (except the server) are
thin
>>clients. All we need to do is keep the monitor and machine alive long
>>enough to allow the students to save their work, it really doesn't
(Continue reading)

Permalink | Reply |
headers
Ace Suares | 4 Aug 2003 17:14
Gravatar

Re: Added costings to pilot plan and resources


> In the event of a power failure and dies, although the Data resides on
> the "server" there are concerns of abnormal termination and defunct
> forks
> Which will eventually affect system performance, and should be taken
> into consideration..
>
> Donny...

Hi Donny,
IN developing a *good* system for CIS, we need to look into the possibility of 
'taking over' dead sessions on the server (which is protected by ups) or find 
a way of automatically 'saving' sessions before any shutdown of the server. 
Here is where genius or a good search on sourceforge can help ;-)

Think of vmware (You know vmware ?) when you close it it saves the ENTIRE 
VIRTUAL MACHINE as one file on the  harddisk. Next time, the mouse pointer is 
even at the same position as when you killed it ;-)

_Ace

PS Good to see so many of your links and tips!!!
Permalink | Reply |
headers
Richard Hamel-Smith | 5 Aug 2003 01:13

Re: Added costings to pilot plan and resources

Ace Suares wrote:
> 
>>In the event of a power failure and dies, although the Data resides on
>>the "server" there are concerns of abnormal termination and defunct
>>forks
>>Which will eventually affect system performance, and should be taken
>>into consideration..
>>
>>Donny...
> 
> 
> Hi Donny,
> IN developing a *good* system for CIS, we need to look into the possibility of 
> 'taking over' dead sessions on the server (which is protected by ups) or find 
> a way of automatically 'saving' sessions before any shutdown of the server. 
> Here is where genius or a good search on sourceforge can help ;-)
> 
> Think of vmware (You know vmware ?) when you close it it saves the ENTIRE 
> VIRTUAL MACHINE as one file on the  harddisk. Next time, the mouse pointer is 
> even at the same position as when you killed it ;-)
> 
> _Ace
> 
> PS Good to see so many of your links and tips!!!
> 
> 
> 
One thing which I have seen do this is VNC. You can quit a VNC session 
on one terminal and connect to it from another. It works even if the 
first terminal is an X-window session running on Linux and the second
(Continue reading)

Permalink | Reply |
headers
Ace Suares | 5 Aug 2003 04:13
Gravatar

Re: Added costings to pilot plan and resources


> There's a ThinVNC version as well now.

I forgot all about VNC but it has great capabilities, indeed.

No need to run X then :-)

Keep this VNC thing in mind, do some tests in the lab once ross has set it up.

I *think* that the capability of 'redirecting' a session is there in X, too, 
need to research that.

_Ace
Permalink | Reply |
headers
Donny Mark Ramdathsingh | 4 Aug 2003 17:33
Picon

RE: Added costings to pilot plan and resources

Ayup, I use VMWARE , it's a memory hog though :( and it's commercial :(

We really need to be on the "ball" for the system;
In UNIX, AIX to be specific, when an Xterm is 
Shutdown abnormally a defunct fork is established
The memory does not automatically allocate back to the system 
A simple ps -aux | more can see the before and after results.
I guess a rudimentary solution would be to show the 
Admin how to search and execute a KILL
But it's better to design availability work in the design phase.
>From my use of TS and the like, Memory is always the bottleneck
And defunct processes play a BIG role is System performance..

The last thing we want is a CIS solution that is NOT available :(

Cheers my friend...
Donny...

-----Original Message-----
From: owner-cis@... [mailto:owner-cis <at> saafe.org] On Behalf
Of Ace
Suares
Sent: Monday, August 04, 2003 11:15 AM
To: cis@...
Subject: Re: [cis] Added costings to pilot plan and resources

> In the event of a power failure and dies, although the Data resides on
> the "server" there are concerns of abnormal termination and defunct
> forks
> Which will eventually affect system performance, and should be taken
(Continue reading)

Permalink | Reply |
headers
Ace Suares | 4 Aug 2003 22:08
Gravatar

Re: Added costings to pilot plan and resources


Hi,

I am thinking of a simple function like 'redirect session x to screen:0:x'
so that a sysadmin can hijack any session and shut it down properly.

Don't know if it is possible (but actuallty I do know it should be possible).

_Ace
Permalink | Reply |
headers
Donny Mark Ramdathsingh | 4 Aug 2003 22:18
Picon

RE: Added costings to pilot plan and resources

I've never tried to Hijack the session , I don't even know if  the 
Rules of AAA will allow it, you can FOR SURE initiate an X 
Using  <at> a:0.0 usage but to take a session and shut it down orderly
That will be SWEET if it can be done, but with MAC (Manadatory Access
Control) I have never seen it.....very much worth the lookup !!!!

Remember the Solution *should* be native :)

Donny...

-----Original Message-----
From: owner-cis@... [mailto:owner-cis <at> saafe.org] On Behalf
Of Ace
Suares
Sent: Monday, August 04, 2003 4:09 PM
To: cis@...
Subject: Re: [cis] Added costings to pilot plan and resources

Hi,

I am thinking of a simple function like 'redirect session x to
screen:0:x'
so that a sysadmin can hijack any session and shut it down properly.

Don't know if it is possible (but actuallty I do know it should be
possible).

_Ace
(Continue reading)

Permalink | Reply |
headers
Taran Rampersad | 3 Aug 2003 22:58

Re: Added costings to pilot plan and resources

Ross Gardler wrote:

> Ace Suares wrote:
>
>>
>>> This is not likely to happen. The machines (except the server) are thin
>>> clients. All we need to do is keep the monitor and machine alive long
>>> enough to allow the students to save their work, it really doesn't
>>> matter if the sound system doesn't work etc
>>
>>
>>
>> Actually, the session gets saved on the server - no need for the thin 
>> clients to remain open. Al least, that's what I think how thin 
>> clients work. No data is in thin clients memory - it's just an 
>> extention of a virtual screen, keyboard, mouse from the server. Do a 
>> little bit of research into that and it saves you huge on UPS's. Just 
>> one for the server, overdimensioned as much as possible, it's much 
>> better if the server keeps ronin for half an hour then 5 minutes. 
>> Hey, for all you know the blackout is only 5 minutes, and everyone 
>> takes off where they left after a coffee break.
>>
>
> Excellent. I was hoping that the this would be the case. This will 
> massively reduce the cost of our installations (around 40%!). I will 
> be conducting some experiments when I return to T&T (actually I don't 
> leave till Monday but that's moot). I'll provide a definitive answer 
> on this then.
>
Well, there we go - if I had read this I could have saved myself a lot
(Continue reading)

Permalink | Reply |

Gmane