WeekEnd Feature: Welcome! Data Matters!

by Ian Thompson, CCSP Staff Editor
January 9, 2004

“Thank you ladies and gentlemen for flying with us today. Your pilot’s name is Ian and he has promised us an entertaining ride. Now, if you could familiarise yourself with the following security procedures…”

I was offered the chance of a 100Mbit fibre-optic off-site connection the other day. Forgive this slight outburst of one-upmanship, but it has a security ‘angle’, promise…

I work as Network Manager and ICT teacher at a UK high school near the city of Leeds. Quick blurb about Leeds; the fastest-growing economic area in the UK outside of London, with around 1 million people in the city itself, and around 2.2 million within a short drive. It’s home to many hi-tech companies, some of them being pioneering in their field, including online/telephone banking and insurance, call centre management and cell phone service providers. Oh, and apparently it’s not far off being the car-crime capital of the UK as well, so even the thieves are good at dealing with complex security… So, a place of mixed blessings.
Amen.

Any way, back to the brag ..er.. topic. The first stage of this upgrade will be completed by April 2004 and will cost us an installation fee of £2,500. For this I’ll be getting a 20Mbit DSL line (10Mbit in either direction) laid between us and the hub in the centre of Leeds, replacing our MetroVPN copper connection (nothing more than a fancy cable network line). There’ll also be some new layer 2 switches to replace the Cisco 2600 router we currently use. I was assured that other similarly sized schools don’t go much above 2-3Mbit, so an upgrade of this capacity would be fine for the foreseeable future.

However, when the guy from the city authority realised that my school’s got one of the biggest networks in Leeds, (not just in high schools, but including all those hi-tech companies), and that we’ve been pushing the technologies available to some of the farthest points possible (okay, slight brag, but he estimated we are in the top 1%), we got chatting on the future possibilities.

You see, unlike copper, fibre is only really limited by what you hang on either end. To provide more speed can be as simple as swapping an old VCR for a new model. I know there are issues with dB loss and the like, but these are nowhere near the limitations of copper, where a combination of very pure metal and short distances between endpoints are vital – for example, Cat5 network cabling inside a building is limited to about 90m, otherwise 100Mbit speeds will be a rarity You can just about get 2Mbit through a 5km cable.

But I digress…
Any way, to cut a long story short (“Too late…!”), the subject of off-site data backup was raised. You see? I cleverly brought this article back into the field of security… Don’t worry; these things have a habit of turning out all right in the end. Don’t ask me how; it’s a mystery.

This led me to think how quickly our regular backup run would take if we used 100Mbit/sec off-site archiving. A full backup of user files currently fills about 80GB, but the incremental stuff tends to be only about 2-3GB per day – we don’t generate that much new stuff, but many files are altered and so on. It can be quite a bit bigger during the few weeks when staff members are entering pupil performance data, but the point is that we run on about 1/40th of our total user space per day.

Quick calculation: 80GB = roughly 80,000MB. 100Mbit/sec = roughly 10MB/sec, which means our full backup would take at least 8,000 seconds, which is just over 2 hours. Our daily backup would be about 1/40th of this, or four minutes… And that’s at raw data speeds. Most backup software allows for data compression by the server, reducing the total archive size and therefore the time taken.

How things change…
Anyone remember doing a “DIR *.* >PRN” of your work directory (aka ‘folder’ for you young ones), then working through the printout to see what files would fit together on a floppy? Then we got disk-spanning Zip archives (leading into the inevitable “It unpacked with the older version!” after an upgrade) – no more worrying about which file went where. I bought my first version of Word, just as version 6 came out in about ’95, and was amazed that it came on eleven 3.5” floppies. IIRC, the full version of Office came out, shortly after that, and took 24 floppies, using Microsoft’s trick 1.68MB disk format. Then someone decided that CD-ROMs were the way to go, and things really got fat…

Well, looking back at the technology I’ve gone through to back up files, the list reads like this:-
 5.25” floppies
 3.5” floppies
 100MB Zip disks (on a parallel port)
 250MB Zip disks (internal ATAPI drive)
 CD-R/RW
 DVD-R/RW

At work, I can also add:-
 4GB Travan tape (and two days to swap six tapes over in the final days of that one – even with Veritas software)
 30/60GB ADR tape (OnStream IDE drive, with Yosemite s/w, taking about 11 hours)
 80/160GB VXR tape (Exabyte SCSI drive, still using MS Backup, at 7 hours)
Add to that list the fact that I’ve also installed a full backup server that uses RoboCopy to mirror the file store areas of the main server – having both redundant PSU’s in an old PDC server fail together tends to make me wary… ever tried to work around having no network for six weeks?!

Moore’s Law; O’Toole’s Rule – now what?
Moore State’s that processing power doubles every 18 months.
O’Toole’s Rule (waddya mean ‘Never heard of it’?) tells us that Murphy was an optimist.
Murphy's Law, of course, states 'anything that can go wrong, will go wrong'..."

Now I give you; Ian’s Slightly Obvious Statement about Backup Capacity: the size of data stored tends to outgrow the backup storage method after about 18 months.

I think you’ll agree, we’re breaking new ground here. However, another staggering revelation is about to be made – the time taken to run a backup is now falling again! It got silly when I was trying to do 26GB on a 4GB Travan drive, so I was glad of the drop to 11 hours on a single tape – my weekend was my own again. Now we see that a daily backup can potentially be made in less time than it takes to make a coffee. And £2,500 isn’t that expensive, compared to a HP SCSI tape drive, for example.

Call it the marginal propensity to “go on a bit”, if you like, but there is a more serious side to all this data diarrhoea. As the reliance on data processing becomes more critical, (even if it is just saving web pages listing cheats for the latest PS2 game), and the volume of data rises, there’s a dangerously high risk of the social life of the network staff reducing.

I congratulate all those technicians at backup technology companies who have worked hard over the years on ever more efficient ways of giving me more time at the pub.

;D

by Ian Thompson

Ian Thompson is a Network Manager of a 500-PC, 5-server, 1700-user school network and is an ICT teacher at a UK high school near the city of Leeds. He has written articles for the Hutchinson Encyclopedia, plus many resources in support of teaching ICT in the UK schools' National Curriculum.

Copyright © Ian Thompson 2004