garrya100

Ahh! Petrol stations! I've had to do them too. Sheilded cables in conduit, buried in 1ft of sand, minimum 3ft underground, with vapour pits no more than 30ft apart, voltage suppression arrestors no more than 50VAC...... I hope I never have to do another one!!! BTW the current flow in the neutral conductor can have a large impact on the PD if the buildings have separate earthing systems. One thing you learn in engineering is that earth is not a particularly good earth! Hence the MEN system that has multiple earths tied to the neutral to establish a solid earth. Bear in mind that the earthing system used in power transmission is designed for safety, not low noise. It's to make sure that when there is a fault, the fuse will blow, or the breaker will trip. Reduced sized neutrals in 3 phase systems, and different earthing conductor sizes, while perfectly legal and adequate for a power system can have large impacts on the PD between buildings. A five volt PD between buildings is no issue for a power system, and the induced currents are small, but for a low voltage data system the difference is huge.

No, not entirely, the regs are designed to help stop the induced noise, and minimise the shock hazard, which can still exist if a mains cable is connected some how to a data cable, either by accident, human error or equipment failure, and can still exist even if the transducer in the switch has blown up as a result of being connect to the mains. A fire hazard and shock hazard are two different scenarios. You can still create a shock hazard if a building has a major earth fault and a cable connected to that system is being handled by some poor individual some distance away. A Cat5 cable doesn't contain enough copper to sustain a fire hazard, but certainly can sustain a current of 30ma, which is all that's required to kill someone! In that situation it's only the impedance of the poor individual on the end of the cable that limits the current flow till the breaker trips or the copper blows. Before somebody brings up lightning strikes again, if you remember the theories of electro-magnetic induction, it's the sudden motion of the electrical field produced by the lightning strike that creates the voltage rise in the cable, and a lightning strike miles away can generate enough voltage in a copper cable to destroy a switch transducer. Again optical fibre is immune to lightning strikes. I can tell you it still goes against the grain designing systems that run data cables in the same conduit as 11Kv lines even though it's perfectly legal using fibre. Copper has to be a minimum of 6 feet away and shielded and then have its own earthing system. In a correctly installed system, with solid earths, the shock hazard is minimal, and the cable itself can't support a fire hazard, but there will still be PD between the buildings that can't be easily eliminated, though it can be minimised. The induced noise can still beat you, even though you've done everything correctly! I can add more theory regarding voltage drops/current flow/cable CSA if you wish, but it may be a little heavy for this forum.

The problem with switches being on different phases being connected by Cat5 copper has nothing to do with the fire risk. The transducer on each end of the copper with blow long before enough current can flow to create a fire risk. Even if you connected a length of Cat5 cable directly across two phases between to buildings the copper would blow like a fuse before a significant fire risk occurred. There simply isn't enough copper in the cable to sustain the current levels needed to create a fire risk even with a significant earth fault. The issue with the potential difference between buildings (which always exists) is the small currents generated by the PD creates additional noise on the cable. With enough PD between to buildings the induced noise will actually swamp the data signal hence no data stream. Even small currents/noise caused by the PD between buildings can cause bit errors, hence re-transmition, and a slow down in data through put, despite a lot of network traffic. Optical fibre being either glass or plastic is an insulator, thus no current can flow regardless of the PD between the switches. The other benefit is that fibre can go to 10G whereas copper's effective limit over any distance is in reality 400Mbs. Garry Anderson (AS Dip Elect Eng if it helps to show I know what I'm talking about )

Dual Cisco 3550-12G's running HSRP at the core, feeding out via redundant 3508's to 26 x 3524/3548's at the access layer through 4 x optical fibres in a dual redundant PVST config for 'poor mans' load balancing. The Campus is spread over 25 acres so I have a mixture single mode and multimode fibre. The longest run is just over 4000ft. Server/SAN switches are Dell 5224's, which are surprisingly good for the money. All areas are VLANed to minimize broadcast traffic across the network. ACL's to restrict student access to appropriate VLANs. 40 x 3Com 2620 dual radio AP's controlled by a WX5002 controller. Staff, student, guest and mobile device wireless VLAN's for QOS policy enforcement. Our own managed ADSL2+ direct access conection through a Cisco 878 and firewalled/filtered/managed by a Sonar appliance. (Its nice having control over your own DNS records!) We host our own mail server, LMS's, terminal server and staff portals which are all available externally. The remote management access is a godsend! A mixture of 5 physical and 24 virtual servers on 3 x ESX hosts running S2003R2/S2008R/Ubuntu 9.04 Pre-school to 12 with 250 PC's, 40 Mac's, 48 Netbooks, 150 student laptops, 80 student iPhones (and growing!), and shortly 45 iPads. The 3524/3548's are old school, but the have the features I require and I can buy them for $60 each. When they die I just replace them. I'll start replacing them shortly with 3550/3560 POE's if we decide to go VOIP. As all the switches are end of life I carry spare units, and also a small stock of PSU's and fans.

From the other side of the fence- I'm the manager, and I'm not a failed IT guy, and I'm a highly qualified IT guy and trainer. When I'm interveiwing candidates I'm looking for experience and some qualifications. I want the guy/girl who has the drive to get some qualifications of his/her own and some experience. They're the ones that are keen to learn, are also the easiest to train. I would prefer the guy with a CCENT, or Network+, or the equivalent from Juniper or HP or something from M$ than the guy who believes he's good enough with any further learning. In this game if you're not prepared to update your skills you might as well get out and go and direct traffic. My advice, find something you're interested in learning about, it doesn't really matter if it's Cisco, Microsoft, HP, VMware or CompTIA, show your prospective employer you're keen to learn. Get some old hardware and pull it apart/rebuild it and learn some practical skills. Find someone to give you work experience and get some references from them. Show any prospective employer you have the basic knowledge to start a job, and you know where to look for more info. That's the sort of candidate employers are looking for. On your journey collect more skills and move up the ladder. Over the years I've collected qualifications from Dell, Apple, Microsoft, VMWare and Cisco. I have also studied and received qualifications that were not a job requirement in Education and Training, Business Management, Project Management, as well as my original qualifications in Electrical Engineering/Industrial Computing and Telecommunication. I'm not an exam junkie, I simply took each opportunity to learn a new skill when a job situation required it. Today I'm well paid, and run a department with a team of 3.5 and a budget of over $1m, and the responsibility for the design, installation and management of all the IT infrastruture in the school, and also all the teacher IT training. Anybody with a bit of go in them could do it too.

The biggest difference between business and enterprise is that you don't get Bitlocker with business (or Pro) but do Ent. (there are some other minor differences) If you don't want Bitlocker than it's a wash. If you do want Bitlocker and you're not a VL client then you have to buy ulimate.

Win 7 Enterprise is only available to you if you're a volume licence customer (Schools Agreement or Select etc) so you would need to buy Pro licences to connect them to a domain if you're not a VL customer. If you're a VL customer you buy them with Win7 starter as the schools agreement is an upgrade licencing scheme. I have a fleet of Dell Latitude 2100's running Win 7 Ent on the domain and they work fine, but its not the cheapest option.

Technically putting an OEM licence on an existing box is a no no.... You should really be buying an upgrade licence. Also a OEM licence is not transferable so it has to stay with the box if you move it on. If you own an upgrade you can legally down grade the 'old' box and move your OS to another box. (ie buy a new box with vista basic or windows 7 home and upgrade it to pro.)

Talk to the Guys at Dell or Computelec. Dell have negotiated volume licencing for schools that are a better deal than Computelec if your an AIS member. I don't know if the're better than Computelec if your commercial. Linda Pryce is the schools licensing contact at Dell Aust.

Renaming a DC is relatively painless, you need to demote it first, rename it and then re-promote it, however if you only have one DC than you may be better of building another DC with the correct name, promote it to a DC, wait for everything to replicate and then demoting and removing the old one I've just done all that moving over from 2003R2 to 2008R2 servers, I changed the names of the old 2003 servers so I could use the same names for the 2008 servers. The 2003 servers become olddc-1 etc till all was replicated and they were decommissioned. Removing a DC from a domain with only one DC is frought with danger, as is trying to rename a DC with out demoting it. Last time I tried that trick I totally wrecked the DNS, and spent way to much time fixing it. As for renaming your forest, I've never seen it done, and don't think its possible with building a new one. It may be do-able with some serious AD tools but I wouldn't like to try! If somebody knows how I'd like to know too!

A 2008R2 sever will happily be a member server on a 2003 domain without having to raise the AD schema to 2008. You will need to extend the group poilicy to use the extra features of 2008 servers via GPO's if you are staying with a 2003 AD schema. You only have to raise the AD schema to 2008 if you want the 2008R2 server to be a DC. If just done exactly that to migrate all our settings to new 2008R2 DC's. The proceedure is fairly painless. Build 2008R2 servers, add features (DHCP,DNS etc) connect to domain, run adprep on new 2008R2 server, run DC promo, wait 24 hours to ensure replication, check logs to ensure its all OK, demote old 2003 servers, wait another 24 hours to ensure all is correct, check logs once more, finally kill old servers. Mission acomplished!

I tried the Mini10 and the Latitude 2100. The Mini10 wins the "smallest" the 2100 wins the most useful. I've set up a few Mini10s as "photo storage and email" for traveling teachers due to there size, but I'd personally opt for the 2100 with the big battery option.

Before you commit to Cat5e, check out some of the newer Cat6 cable. It's not as bulky as the older Cat6, and not as expensive. It also gives you some degree of future proofing your network. While 1Gig to the desktop is not a big advantage yet, that day may come sooner than we realise. P.S DMX512A runs at 250Khz, capacitance is a bigger factor for DMX cables due the rise time and lack of error checking on the receiving device. The standard does not require the transmission of all 512 channel hence the impossibility of any parity/error checking. All you have to do is transmit the start code and then the channel data you wish to send up to 512 channels. (too many years running a production company, now I do it for fun!!) BTW anybody want a Jands Hog 1000, I've got one I no longer require!!

I agree with Steve. I used to keep updating all my images to keep the patches up to date. Now I just have a set of standard images and my WSUS does the rest. Much less work. And of course WSUS is a free download.

I would go for some Cisco Quals, preferably do your CCNA, or at least a CCENT. If you have been a network manager, and are as good as you say you are, then a couple of decent books from Cisco press will get you through the exams. CCENT is really much the same as Network+ but with the Cisco flavour. CCNA adds basic routing, ACL's and a few other skils. Cisco aren't the only network quals, but the are the 'industry standard'. Speaking from the employer side of the fence, I allways look more favourably on the candidate who has shown the willingness to improve their quals. For equal expirience, the guy who is working to become more knowledgeable wins everytime. Age has very little to do with it, practical skills are more important

And a wise piece of advise from an old hand - "if a piece of equipment has an auto function, you auto-not trust it!"

It re-enforces what we were told in Cisco school, "always hard code the speed and duplex unless you need auto!!" And yes it's caught me out too!

I may have confused you with my definenitions. Both S2008 servers are indeed from the same code base, otherwise there would be too many problems with being able to use the same drivers and programs etc. But S2008R2 is not a 'bug fix' as is Win 7. The code has been extended and recompiled. Much the same as Win2K and XP are the same code base, but the're not the same OS. S2003R and S2003R2 are actually the same OS with a major service pack being the diference. S2008R2 is not S2008 with a major service pack, but a recompile of the OS. If you like, S2003R2 is a .001 revision while S2008R2 is a .1 revision.

Win 7 32 bit is a "bug fix" based on the Vista code and as such is not a 'subset' of the S2008R2 code. I imagine it would be the same for Win 7 64bit. S2008R2 is an extention of the S2008 64bit code to suit the new 'features' of Win 7. M$ has always said they only wanted to support 1 server code base, so S2008 32 is the last 32 bit server code. (Vista SP1 shares some code with S2008, and the kernals share similar instructions). If M$ decide to build another 'new' consumer OS, I suspect it will be 64 only. If it's another extension of the existing code base there could be both flavours. There are still a lot of embeded devices out there running XP 32, so a market for a basic 32 OS still exists.

Hi Tim It's time to do some serious thinking about what you wish to acheive. I have an Openfiler box running on my "test suite", and it works OK. But, you need to know about iSCSI to get the best out of it. Performance is not as good as a dedicated iSCSI box, but probably adequate for up to 12 VMs. The main reason for not considering Openfiler as a production solution is that it leaves you open to a single point of failure. If you have 4 servers on boxes and one box dies you have 3 boxes, with the possibility of moving some processes to another box. If your Openfiler box fails, you're dead in the water, with few options. The second big point of virtualisation for me after the hardware savings was the fact I could build in redundancy. With my system you can pull the power or network on any part of the "server core" and the system will auto recover, send me an email telling me all about it, all with very little interuption to the end users. Some other things to consider with the MD3000i, performance is average with SATA drives. Dell are currently working on it, but I personally wouldn't consider SATA for VM's anyway. You can mix SATA and SAS drives in the MD3000i, which means you can have 'VM Luns" and "Storage Luns" if you like and save a little money. Also, remember that when you place drives in a RAID array you take a performance hit (up to 50%!) compared to a single drive.(Yes, RAID0 is the exception, but RAID0 as the name suggests is not really RAID its really just AID!) A mirrored drive has only a small difference, the biggest hit is RAID6. The main reason for this is the seek time of the drives, which is why I went for 15K drives, that have the fastest seek time. 600GB/10k drives end up being around the same price, but I went for the 15k drives anyway to maximise performance. SAS drives also have a longer 'rated life' than SATA, but all drives failures are becoming rarer these days. Think about your SAN switches. I ended up using Dell 5224's. Being a "Cisco Geek", I would have prefered to use Cisco switches like the rest of my network, but the cost was prohibative. The 5224's are not "featureiffic" and the CLI is somewhat clunky, but once you've done the initial setup, you can configure them via a web gui. I would give your Dell Rep a call and see if you can upgrade you MD3000, but by the time you upgrade 2 iSCSI controllers, and possibly some HDDs, it may be cheaper to trade it on a MD3000i. Finally (hope you got something out of all this) there are some great 'white papers' about all this on the VMware site and also the Dell forums. While they mostly refer to VMware, the concepts are identical to all virtualisation products. Go take a read, you may find all you want to know there, otherwise, PM me and I'll try to help.

Unfortunatley a lot of it is restricted as it contains passwords and proprietry info. I'll have a look when I'm in the office to see if I can have enough left to send if I cut out all the sensitive bits!!

How do you get that much info into 20 pages! My network operations manual that includes all the diagrams, copies of all the configs, a fairly detailed "how to" section designed for a junior tech, and a trouble shooting guide runs to just over 100 pages! I currently updateing the server/admin system manuals, and they too will be weighty tomes!!

I have 3 x 2950's with dual quad CPU, and 24GB Ram connected to a MD3000i, running a total of 24 VM's on VSphere. The MD3000i has 15x 400GB 15k drives configured as a RAID5 Array with 6 Different LUN's. VM are connected to LUN's depending on their expected I/O requirement. EG the file sever and exchange server each have their own LUN's, the backup DC and SCCM server share a LUN (with some other low I/O processes) The SAN has multiple redundant data paths and the system runs DRS and HA. Performance is good, the only time I have a performance hit is when you override the stagered start up and start too many servers at once. Hyper-V when tuned correctly should be at least as good, and with only 4 servers per box should be fine. I would run 2008 Core on the local drives instead of loading the OS from the SAN, simply because at least that way it gives you another option for recovery if you have a complete SAN meltdown (I have a collection of complete VM's backed up that I can load on local drives as a last resort disaster recovery option). To move over you should only have to set up your SAN correctly and then migrate the storage to the iSCSI box. BTW the MD3000 is a NAS box, not iSCSI, the MD3000i is the iSCSI version. If you have a MD3000, you may be better off to leave the VM's on the local drives and use the MD3000 as storage.

You never seem to get all the slips back, and there's allways one student who says he/she never got it. So our AUP is on the logon screen, in the College diaries and also presented to all students at orinentation days and the first day of each term. It starts with, "By logging onto and using the College network you agree to........' no slips required and so far no complaints.

Netstumbler is your friend! You can even link it to a GPS to produce an area map. I've used it many times to produce network surveys for clients. Once you understand the program and wireless design, it makes life easy. NetStumbler.com