Saturday, 18 April 2009

Don't get me wrong

Don't get me wrong - I am not bashing OneTel Mk II because I think a fibre network is a silly idea. I am bashing it because the economics don't add up.

Over the last 7 years, I've put in a number of fibre networks. The first one I put in linked a number of sites in a regional city - the company I was working for had hundreds of staff and several server rooms scattered over a dozen sites in the city in question. I discovered that many of the sites had dark fibre running along their boundaries, and I arranged to replace a whole bunch of expensive 2MB Telstra frame relay links with a 100MB fibre network. We consolidated over a dozen servers in 6 locations down to three servers in one location, and improved the IT setup for all the staff working in that city. I even managed to tap into another fibre going back to Sydney, and finangled a 30MB link from the regional centre back to the Sydney CBD.

That fibre network paid for itself within a year - that company replaced servers every 3 to 4 years, and they cost a minimum of $10,000 each. Simply reducing the number that needed replacing from 12 down to 3 realised a cost saving of $90,000 - a saving that would be repeated from thereon every 3 to 4 years. Due to the then hideous cost of frame relay, renting a dark fibre and installing our own switches and fibre drivers ended up costing no more than the frame relay links, even though we got 50 times more bandwidth.

A year after we installed that 100MB network, the backbone was upgraded to 1GB - the cost of the fibre drivers had declined to the point where it was economic to buy them instead. We were still stuck with a 30MB link to Sydney, as that link required a pair of $50,000 repeaters half way down, and it didn't make sense to spend that much on a link we didn't need.

After that, I helped rollout another fibre ring around Sydney, linking dozens of offices on a 1GB backbone. Again, the economics made sense because we eliminated over 50 servers and 10 server rooms in the process. The 1GB backbone allowed us to consolidate lots of scattered servers (and support staff) back into a central data centre, and to replace lots of piddly servers with a small number of bigger units. I also tried to organise renting dark fibre from Trans Grid to another regional city where we had a presence, but nothing came of that.

Not a sod was turned until I had written a water tight business case that included a mass of cost/benefit calculations. I'm pretty tight with spending company money, so the business case had to be convincing enough to convince me first of all. I wasn't going to try and sell a crock of shit to my boss if it didn't add up.

At the time of that rollout, I was frustrated that we didn't use that network to do more - we could have saved even more money, but internal politics prevented that from happening. I was even more frustrated that we were unable to use that network to offer savings to other companies in our group - intra company politics got in the way there as well. However, that fibre network still did a lot of good, and I am proud that I had a part in getting it installed.

I put in another fibre network for another company last year, linking various sites around Sydney at 1GB. If their financial situation allows it, next year they are going to increase that network to 10GB. Putting that network in allowed that company to make some major changes to how it runs its IT shop, and it saved quite a lot of money in some areas.

I also put a huge amount of fibre into a data centre last year - that site had over 4000 fibre end points, and I personally installed nearly 1500 fibre patch leads as part of the job. You could say I am a bit over fibre. Just the fibre alone for that job came in at $750,000 and took two months to install, and it was connected up to $5 million worth of switches. I finished another smaller site in the new year - I spent days on my guts under the floor sorting out fibre cabling. In the end though, the result was good.

The thing is though that all this work was expensive. Hideously expensive. Back in the late 1990s, I had to link two buildings in the CBD via fibre - they had a microwave link between them, but it was unreliable and not fat enough. It took months of negotiation with Energy Australia (to get access to the ducts under the street), the landlords of both buildings (to get access to the basement and the risers) and the Council (to allow the fibre to be laid). Even though the buildings were only a few blocks apart, and were in line of sight, it still cost $70,000 to lay that fibre end to end, and the monthly rental on the duct space was expensive.

You don't have to convince me that fibre is a good thing. Putting in that first 100MB network was a huge leap forward for the company concerned. However, it cost a lot, and I see no point in spending big dollars unless you are going to reap even bigger returns. I'm simply not convinced that most households have deep enough pockets to pay for it, and that they will achieve any worthwhile gains at all.

No comments: