Over the next month or two I need to set up a server for my old secondary school to provide user storage for all the pupils. There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible. Currently the school is about half Acorn and half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-). Clearly the storage requirements for the Windows software is of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me. So to the question: I'm happy setting up Linux as a Samba server to do this lot, but does anyone have experience/advice on choice of hardware for the job (I'm happy to build a machine from bits). RAID is a must, but IDE gives us a lot more Mb for our money than SCSI, but maybe at the cost of performance/reliability? Nevertheless, it seems to me that even a modest hard disc can outperform the 100MHz ethernet to which the server will be connected. All advice most welcome. -- Phil Driscoll
Over the next month or two I need to set up a server for my old secondary school to provide user storage for all the pupils.
There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible. Currently the school is about half Acorn and half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-). Clearly the storage requirements for the Windows software is
I'm currently investigating running Windows apps using an LTSP setup. I have been quoted a price of USD 55.99 (per station) from NeTraverse for the Win4Lin server product. Even if you still need to buy multiple Windows licences the fact that you can cut down on the hardware specs more than make up for this.
of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me.
I give them about 20M and only increase it for those who actually need it. Even using Windows apps they will have a hard time filling up 1G. You also need to work out how on earth you are going to back this lot up.
So to the question:
I'm happy setting up Linux as a Samba server to do this lot, but does anyone have experience/advice on choice of hardware for the job (I'm happy to build a machine from bits). RAID is a must, but IDE gives us a lot more Mb for our money than SCSI, but maybe at the cost of performance/reliability? Nevertheless, it seems to me that even a modest hard disc can outperform the 100MHz ethernet to which the server will be connected.
In practice ethernet can often be faster than a HDD, since the quoted figures for HDD's tend to be burst, rather than sustained transfer rates. -- Mark Evans St. Peter's CofE High School Phone: +44 1392 204764 X109 Fax: +44 1392 204763
There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible.
Woah! I have 1200 students, and a total storage space of 5GB....... I can dream..... Regards, Robb Bloomfield, RLS Network Manager -----BEGIN PGP SIGNATURE----- iQA/AwUBOnqp7pLoTHZaFLiuEQKXegCfZBI/ZqaYLSuNPqyYb3O/+AQaoPMAmwZ3 wL2zJ3u67tXeitxrenFk2XgZ =UMho -----END PGP SIGNATURE-----
On Friday 18 May 2001 1:10 pm, you wrote:
There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible. Currently the school is about half Acorn and half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-). Clearly the storage requirements for the Windows software is of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me.
For our 1600 plus sixth form we give 40 or 80Mb for lower school and 80Mb for year 11 & up with individual increases in the 6th form. 1G id more than I've got on the users server!
I'm currently investigating running Windows apps using an LTSP setup. I have been quoted a price of USD 55.99 (per station) from NeTraverse for the Win4Lin server product. Even if you still need to buy multiple Windows licences the fact that you can cut down on the hardware specs more than make up for this.
Have you looked at www.bpbatch.org for thin clients served from Linux? I haven't really looked at Win4Lin but do use VMWare - there are some better apps coming out now that many vendors are going web-delivered.
You also need to work out how on earth you are going to back this lot up.
Here, here. "Even" students deserve to have their work protected.
I'm happy setting up Linux as a Samba server to do this lot, but does anyone have experience/advice on choice of hardware for the job (I'm happy to build a machine from bits). RAID is a must, but IDE gives us a lot more Mb for our money than SCSI, but maybe at the cost
In practice ethernet can often be faster than a HDD, since the quoted figures for HDD's tend to be burst, rather than sustained transfer rates.
Agreed, and SCSI must be better than IDE because of the constant use. IDE is OK for home (I suppose - but think how often folk trash their HDDs!) but I would only use IDE for a low-profile server in a working environment. Specially if you think you can run to 1Tb per student! -- Best wishes, Derek Harding, (BA MIAP) ICT & Network Manager, Warlingham School hardingd@warlingham.surrey.sch.uk
On Friday 18 May 2001 1:10 pm, you wrote:
There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible. Currently the school is about half Acorn and half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-). Clearly the storage requirements for the Windows software is of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me.
For our 1600 plus sixth form we give 40 or 80Mb for lower school and 80Mb for year 11 & up with individual increases in the 6th form. 1G id more than I've got on the users server!
Most often when someone needs their quota increasing they either have piles of junk to remove or they have managed to create huge publisher files. (And want to take them home of floppy!)
I'm currently investigating running Windows apps using an LTSP setup. I have been quoted a price of USD 55.99 (per station) from NeTraverse for the Win4Lin server product. Even if you still need to buy multiple Windows licences the fact that you can cut down on the hardware specs more than make up for this.
Have you looked at www.bpbatch.org for thin clients served from Linux? I haven't really looked at Win4Lin but do use VMWare - there are some better
Didn't get very far with VMWare, since out test server is an AMD... Interestingly this dosn't appear to worry Win4Lin. It is though a case of finding something which is suitable for school use, rather than the giving each user an individual copy of Windows.
apps coming out now that many vendors are going web-delivered.
-- Mark Evans St. Peter's CofE High School Phone: +44 1392 204764 X109 Fax: +44 1392 204763
questions? how many workstations - thus how many concurrent connections? what apps are they going to use - big ones if they need 1GB each is it mainly going to be classroom environment, or walk-in computer labs? If classroom then you need to cater for all users logging off at the same time. if walk-in, then a much flatter workload can be expected. How serious is this data, what backup (online and offline) is required. Could you share the workload over a number of servers, e.g. one per year? Do you need seperate (safer) storeage for staff? There is not enough information here to provide you with any real advice. However, we run a number of servers with up to 100 concurrent users here, and although the main server is an RS6000 with SCSI disks, we don't do raid and we don't have any performances. If you are going to use IDE (as we do) then it may be worth you looking at raid as a means of levelling the load on the disks. Bear in mind tho' that multi-user access tends to squash disk performance as the heads have to fly all over. If you want hot-swap, then you will need to look at SCSI, and probably top end SCSI at that. Personally, I think this would be over-kill. Gary On Friday 18 May 2001 12:40 pm, Phil Driscoll wrote:
Over the next month or two I need to set up a server for my old secondary school to provide user storage for all the pupils.
There are just over 800 kids at the school, so I guess a total of about 1Tb of storage would be sensible. Currently the school is about half Acorn and half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-). Clearly the storage requirements for the Windows software is of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me.
So to the question:
I'm happy setting up Linux as a Samba server to do this lot, but does anyone have experience/advice on choice of hardware for the job (I'm happy to build a machine from bits). RAID is a must, but IDE gives us a lot more Mb for our money than SCSI, but maybe at the cost of performance/reliability? Nevertheless, it seems to me that even a modest hard disc can outperform the 100MHz ethernet to which the server will be connected.
All advice most welcome.
-- Gary Stainburn This email does not contain private or confidential material as it may be snooped on by interested government parties for unknown and undisclosed purposes - Regulation of Investigatory Powers Act, 2000
Thanks for the deluge of responses: Further info/answers to questions posed: The school is an ordinary state sector secondary school with no 6th form, so takes years 7 to 11. The majority of the ICT use seems to be class groups hence big spikes in the load are to be expected at the start and end of lessons. Currently we need to cope with 100 -> 150 concurrent connections but I would be surprised if this doesn't double over the life of the server. The data *is* serious enough to need backing up. I don't think that hot swap is a necessity - as long as the system will keep working with one duff drive which can be swapped out of school hours, that is good enough. Sharing the load between servers on a 'per year group' basis will probably be too expensive, although splitting across two servers may ba acceptable. SCSI vs IDE - I suspect (but don't know for sure) that the kind of SCSI drives we are likely to be able to afford will be of the same mechanical quality as the IDE ones we are likely to be able to afford. Yes there is an issue of saturating a single 100M network - we need to spread the load somehow. I also appreciate that 1GB per student might seem a lot, but I'm sure that before this server ends it's life, it will seem tiny. Cheers -- Phil Driscoll
Phil Driscoll wrote:
I also appreciate that 1GB per student might seem a lot, but I'm sure that before this server ends it's life, it will seem tiny.
That still does seem *very* generous. What do you expect them to do in 1GB? Of course, they shouldn't be installing programs and stuff there. At home, my home directory is approximately 150MB, and that includes years worth of e-mails, old school work etc. The school where I'm at (I'm in the sixth form) allocate 2MB (OK, so that is slightly mean) to lower years, and 10MB to the sixth form. There generally isn't a problem with that. Individual increases are possible. Sadly it's all an appalling RM Connect network, but we can hope for better things ;) Cheers, -- Chris Howells E-Mail: chris.h@gmx.co.uk ICQ: 93699029 Web: http://www.chowells.uklinux.net
On Fri, 18 May 2001, Phil Driscoll wrote:
The majority of the ICT use seems to be class groups hence big spikes in the load are to be expected at the start and end of lessons.
One tip for alleviating this (which may already be used): encourage students to log on as soon as they get to the classroom. I have seen classes told to sit in front of the computers until the teacher tells them all to log on at once, which is a recipe for crawling... Michael
Thanks again for the continuing torrent of good information, and thanks to all who have sent me info and details of their experiences off list. I can see that SCSI, particularly with the tagged queueing which Christopher mentioned is going to give great performance, but I just have to rule it out for reasons of cost. I know a lot of you think that IGb per user is over generous, and indeed it is such a big step up from what they currently have that it will cause them no problems if we phase in storage over the next year. However, I feel that to aim significantly lower would be to short change them. My Risc PC has a total of about 3Gb storage. It is full, but it does contain all the Acorn work I've done since 1987. The three year old NT box I replaced last week, on the other hand, had 11Gb of data on it representing just three years work and nothing within that which I feel I could get rid of. Now I know that my requirements are nothing like someone in a school, nevertheless, the fact is that lots of commonly used modern software products produce outrageously large files. The kids (sadly!) use Powerpoint for multimedia, a language lab which produces large quantities of audio files. They use Photoshop in art, Cubase in music and so on. In short, they have every excuse to use up loads of disc space! So, in view of all that has been said, and our severe cost constraints, I think that we might do somthing along these lines. Build 5 machines with plenty of RAM, 4 large IDE drives on hardware IDE RAID controllers. Each machine will be assigned to a year group, and so will effectively stay with a child for their entire stay at school. This should spread the load nicely, and if we site the servers on different segments of the network we can at least go some way to avoid saturating one segment although short of physically replugging things in sync with the school timetable, I don't know how we would optimise this. There might also be some mileage in getting each server to handle logins and dole out Windows profiles for its year group. On the hardware front, I guess the sensible thing is to go for big and fast but not quite leading edge, so that we reap the benefits of commondity prices without paying a premium for the latest greatest stuff. This might mean somthing like AMD1200 processor, 512Mb 133 RAM and so on. Does anyone have a good cheap source of rackmount cases - the ones I've seen always seem to be several times dearer than large tower cases. Cheers -- Phil Driscoll
On Sat, 19 May 2001, Phil Driscoll wrote:
The kids (sadly!) use Powerpoint for multimedia, a language lab which produces large quantities of audio files. They use Photoshop in art, Cubase in music and so on. In short, they have every excuse to use up loads of disc space!
Do they know about using efficient file formats for e.g. audio/picture data? It might be worth making it a network policy that inefficient formats (.bmp, .wav etc) are banned - you could easily set up an automated routine to scan for such files and e-mail warnings to the owners, or even to automatically convert them to a more efficient format.
Build 5 machines with plenty of RAM, 4 large IDE drives on hardware IDE RAID controllers. Each machine will be assigned to a year group, and so will effectively stay with a child for their entire stay at school. This should spread the load nicely, and if we site the servers on different segments of the network we can at least go some way to avoid saturating one segment although short of physically replugging things in sync with the school timetable, I don't know how we would optimise this.
Bring each segment up to one switch, connect each of these `top-level' switches to a switch shared with the servers. As long as there is very little traffic that isn't of the form server<->workstation, then this will be optimal because the `top of tree' switch will effectively act to dynamically rewire the servers into the appropriate `top-level' switch. Make sure that the `top of tree' switch is a decent one (with cut-through switching and enough intra-switch bandwidth to support full duplex connections on each port simultaneously). Also make sure that you don't introduce any bottlenecks between the `top-level' switches and the servers: make sure that all intervening links run at full duplex at the fastest data rate you can get (1Gbps would not be too fast). And lastly, although you're probably already doing this, don't use any hubs if you can possibly avoid it - stick to switches. At anything higher than 10Mbps, hubs cause major problems. Michael
half Windows, but the Acorns are being phased out in favour of new Windows machines - although they might be new Linux machines if I can make the school see sense :-).
You might be able to set one up as dual-boot then roll that out to all of them.
Clearly the storage requirements for the Windows software is of a different order of magnitude to the Acorn stuff, so 1Gb per child does not seem overly generous to me.
Only an order of magnitude if they use Publisher. Otherwise I, like others, find a Tb rather large, although I did two days ago delete 8GB of mp3 files from the directory of one pupil here, and there are today over twenty with more than 100MB. Drives reduce in price by 40% per annum, so it's extravagant buying now the capacity you will need in two year's time.
I'm happy setting up Linux as a Samba server to do this lot, but does anyone have experience/advice on choice of hardware for the job (I'm happy to build a machine from bits). RAID is a must, but IDE gives us a lot more Mb for our money than SCSI, but maybe at the cost of performance/reliability? Nevertheless, it seems to me that even a modest hard disc can outperform the 100MHz ethernet to which the server will be connected.
Although we now have RAID (software RAID on FreeBSD, dedicated server running NFS, Samba, Appletalk etc) I am not sure it is essential if you have good backups. In our case this is done by overnight dump/gzip scripts to a separate machine, and by September I hope there will be machines at four widely separated locations in the school spending much of every night cross-dumping to each other! You need a fast processor for the compression. Backup drives are 80GB slow IDE (5400 rpm @ 220 pounds each). It's much faster restoring from a hard drive than from tape! I think SCSI is still worth the premium for servers. SCSI is much faster even if it's the same speed because SCSI supports tagged queuing. My guru tells me the latest IBM IDE drives now support that too, BUT. Anyway, here is what he says: =========================================================== With tagged queueing, the host can send several commands (eg. read/write sector) to the drive without waiting for them to complete: the drive then executes them in whatever order is most convenient, reporting the results as each individual command completes. This is particularly effective when writing multiple sectors within a cylinder: if you supply the commands one at a time, you have to wait (on average half a revolution) for the first command's sector to appear under the read head, and then (if writing consecutive sectors) you have to get the next command in pretty quickly to ensure you don't miss the sector going past. Historical systems used to lay out the sectors on each cylinder such as to leave just the right gap to account for the processing delay (eg. in order 1,10,2,11,3,12 etc. round a track, and a similar skew between platters), but with modern drives you just don't have enough information to do this (they don't have a constant number of sectors per track for a start - tracks nearer the edge of the disc have more sectors). A method of cheating (in the write case) is for a drive to enable a write cache, and falsely report to the host that each command is complete as soon as it arrives. This gives you good scores on benchmarks, but means that the host has no idea when data is actually safe on disc. BSD Unix systems have always taken care to write out updates to directories in the right order, and wait for the command to complete before re-using those disc sectors: this is what ensures that you don't lose large numbers of files after an uncontrolled shutdown and fsck. Some IDE drives have been seen to keep sectors in cache, unwritten, for hours at a time if the drive is kept busy - such a drive will seriously corrupt the filesystem if a power failure occurs at an inconvenient moment. Using tagged queueing, you can get the same performance as the 'cheat' without sacraficing reliability. I haven't studied the IDE tagged queueing stuff in detail, but another trick available on SCSI is detach/reconnect: this allows the command, but not its data, to be sent to a drive - the drive will later re-connect to the host to get the data associated with that command, hence allowing more pending commands to be active than can fit into the drive's RAM, with the drive chosing the order in which the data arrives. I don't think that IDE can do this properly - even with tagged queueing - so with IDE you lose performance if you have more than one drive per controller. ================================================================= To summarise, on a single-user workstation the drive rarely has to do more than one operation (load file/save file) at once, so IDE is as fast as SCSI. But on a server there may be dozens of concurrent read/write operations, and this is where SCSI becomes very much faster than IDE. Higher level queuing can ameliorate but not solve this. But manufacturers keep on having new ideas! RAID 5 only requires one spare drive, so you could put a dozen on one SCSI controller and only lose one for security. -- Christopher Dawkins, Felsted School, Dunmow, Essex CM6 3JG 01371-820527 or 07798 636725 cchd@felsted.essex.sch.uk
participants (8)
-
Chris Howells
-
Christopher Dawkins
-
Derek Harding
-
Gary Stainburn
-
Mark Evans
-
Michael Brown
-
Phil Driscoll
-
Royal Latin School